KEY QUESTIONS IN SECOND LANGUAGE

ACQUISITION

This highly accessible introductory textbook carefully explores the main issues
that have driven the field of second language acquisition research. Intended for
students with little or no background in linguistics or psycholinguistics, it explains
important linguistic concepts, and how and why they are relevant to second
language acquisition. Topics are presented via a “key questions” structure that
enables the reader to understand how these questions have motivated research in
the field, and the problems to which researchers are seeking solutions. It provides
a complete package for any introductory course on second language acquisition.

Bill VanPatten, formerly Professor of Spanish & Second Language Studies at

Michigan State University, is now an independent scholar. He has had a long and
distinguished academic career and has published widely in the fields of second
language acquisition and second language teaching.

Megan Smith is an Assistant Professor at Mississippi State University. Her main

research interests are in second language acquisition, particularly L2 syntax and
psycholinguistics.

Alessandro G. Benati is Head of the English Department and Professor of

English and Applied Linguistics at the American University of Sharjah. He is
visiting and honorary professor at the University of York St. John and the
University of Portsmouth.

2
KEY QUESTI ONS I N SECOND
LANGUAGE ACQUI SI TI ON
An Introduction
Bill VanPatten
Chowchilla, CA, USA

Megan Smith
Mississippi State University

Alessandro G. Benati
American University of Sharjah, UAE

3
 University Printing House, Cambridge CB2 8BS, United Kingdom

One Liberty Plaza, 20th Floor, New York, NY 10006, USA

477 Williamstown Road, Port Melbourne, VIC 3207, Australia

314–321, 3rd Floor, Plot 3, Splendor Forum, Jasola District Centre, New Delhi –
110025, India

79 Anson Road, #06–04/06, Singapore 079906

Cambridge University Press is part of the University of Cambridge.

It furthers the University’s mission by disseminating knowledge in the pursuit of

education, learning, and research at the highest international levels of excellence.

www.cambridge.org

Information on this title: www.cambridge.org/9781108486668

DOI: 10.1017/9781108761529

© Bill VanPatten, Megan Smith and Alessandro Benati 2020

This publication is in copyright. Subject to statutory exception and to the provisions of

relevant collective licensing agreements, no reproduction of any part may take place
without the written permission of Cambridge University Press.

First published 2020

Printed in the United Kingdom by TJ International Ltd, Padstow, Cornwall

A catalogue record for this publication is available from the British Library.

ISBN 978-1-108-48666-8 Hardback

ISBN 978-1-108-70817-3 Paperback

Cambridge University Press has no responsibility for the persistence or accuracy of URLs
for external or third-party internet websites referred to in this publication and does not
guarantee that any content on such websites is, or will remain, accurate or appropriate.

4
To Murphy

– BVP

To Hideki and Ayako

– MS

To Bernadette, Grace, and Francesco and in memory of AnnaMaria and

Orazio Benati

– AB

5
 Contents
Foreword
Acknowledgments
List of Abbreviations
Prologue

1 What Are the Origins of Second Language Acquisition as a Research

Field?
The Revolution in Linguistics and L1 Acquisition
The Revolution in Second Language Acquisition
The Driving Question
Some Observations on Data Types
Conclusion
Recap
Following Up

2 What Does Development Look Like?

Morpheme Orders
Developmental Sequences
U-Shaped Development
Markedness Relationships
Summary So Far
Knowing More Than What You Are Exposed To
Constraints on Acquisition
Back to Variation for a Moment
Recap
Following Up

3 What Are the Roles of Input and Output?

The Nature of Input

6
 The Necessity of Input
Output
Recap
Following Up

4 What Is the Initial State?

Universal Grammar
Universal Grammar as the Initial State
The L1 Is the Initial State
Partial Transfer of the L1 in the Initial State
Quick Recap
Other Perspectives
Recap
Following Up

5 Can L2 Learners Become Nativelike?

What Is Nativelikeness?
Second Language Acquisition
Recap
Following Up

6 Does Instruction Make a Difference?

Instruction and Ordered Development
Instruction and Rate of Development
Instruction and Ultimate Attainment
The Effects of Instruction and Short-Term Studies
Recap
Following Up

7 What Role Does Explicit Learning Play in L2 Acquisition?

A Brief Discussion of “Learning”
Can Explicit Knowledge Turn into Implicit Knowledge?
Can Explicit Knowledge/Learning Influence the Development of

7
 Implicit Knowledge?
Putting Explicit Learning into Perspective
Recap
Following Up

8 What Are Individual Differences and How Do They Affect L2

Acquisition?
Individual Differences in L1 Acquisition
Individual Differences in L2 Acquisition
What Do Individual Differences Explain?
Recap
Following Up

Epilogue

Glossary
Index

8
 Foreword
Interest in second language acquisition (SLA) has been on a sharp rise for a few
decades. Examining the acquisition and learning of non-native languages is
important for multifarious reasons and, thus, of interest to students and scholars of
linguistics, psychology, cognitive science, sociology, and many more disciplines.
There is no shortage of introductory books on second language acquisition on the
market, which is not surprising when one considers the many paradigms that
contribute to the larger field. Key Questions in Second Language Acquisition: An
Introductionis a unique and refreshing such book that stands out in this well-
populated landscape. As an introduction to the field of second language studies in
the broad sense, the book is as impressive as it is useful for a varied audience. The
authors do a great job walking a novice student audience through decades of work
in SLA, while providing a level of history of our discipline that offers insights and
succinct review for even the more seasoned researcher. What stands out most
from other texts available is the format. The tone and structure, aligning to key
questions as the title suggests, make for a quite enjoyable and easy experience. I
especially enjoy the “Consider this …” excerpts, which facilitate this book being
used at various levels of teaching on undergraduate and even graduate courses.
The layout of the book is also logical in its flow as it incrementally builds
knowledge that is needed for questions/topics covered downstream – another
advantage for quick adoption of the book as a teaching tool.
Often with books of this type accuracy is sacrificed for the sake of perceived
teaching needs: Details are changed knowingly or certain topics are foregone
since they might be “too complex” for an introductory book. The authors have
managed to find the perfect tone to engage, excite even, the reader while

9
delivering a seriously large amount of information. Many difficult/dynamic
questions are addressed without infantilizing the audience. And still, little to no
prior knowledge is needed to get the full benefits of this book. In this respect, this
book would be a very nice introduction even for the very seasoned linguist or
psychologist who works on theory, language acquisition, or processing yet does
not do so in the domain of sequential bi-/multilingualism. In a nutshell, this book
is current, complete, well informed and well articulated. It is a must-have for
everyone who studies or is interested in second language acquisition.
Jason Rothman
University of Reading
UiT, The Arctic University of Norway

10
 Acknowledgments
Many thanks are in order for a book such as this one. First and foremost, we
would like to thank the folks at Cambridge for their work in getting this book out,
especially Helen Barton, who shepherded the proposal and the final manuscript
through the review stages, and Charlie Howell, who managed the book during its
production phase. Second, we would like to thank our family and friends for their
support over the years. Such support cannot be understated because when the
times get rough, they are the people you need to lean on. We also thank all the
scholars who have influenced us over the years with their research and writings. A
book such as this is always built on the work of many people. Finally, we thank
our students who have inspired us to write an introductory book such as this one.
We hope to have done the best job possible.

11
 Abbreviations

ACC
accusative case

CEI
communicatively embedded input

CP
complementizer phrase

CPH
the Critical Period Hypothesis

ESL
English as a second language

FDH
the Fundamental Difference Hypothesis

FEM
feminine

FT/FA
Full Transfer/Full Access

L1
first language

12
L2
second language

MASC
masculine

MLAT
Modern Language Aptitude Test

NNS
non-native speaker

NOM
nominative case

NP
noun phrase

NS
native speaker

OPC
overt pronoun constraint

OVS
object-verb-subject (word order)

POS
poverty of the stimulus

PP
prepositional phrase

RC

13
relative clause

SLA
second language acquisition (as a research field)

SOV
subject-object-verb (word order)

SSH
the Shallow Structure Hypothesis

SVO
subject-verb-object (word order)

TP
tense phrase

UG
Universal Grammar

V2
verb second

VP
verb phrase

WM
working memory

14
 Prologue

Second language acquisition (SLA) is a field of empirical and theoretical

research that has been around since the early 1970s. For scholars interested in
cognitive science (i.e., the interdisciplinary study of the mind and its processes),
research on how a second language (L2) is acquired adds valuable data that can
shed light on various hypotheses and offer novel insights into how the mind
works. For teachers, basic facts about L2 acquisition can be informative for
decision-making processes about methods, what to teach, what not to teach, and
so on. (Note to the reader: In this book, we use SLA to represent a field of study
and L2 acquisition to mean the actual acquisition of another language. Thus, when
the reader sees SLA, that should be interpreted as a field of study interested in L2
acquisition.)
Regardless of the motivation for one’s interest in research on L2 acquisition,
it should be noted that the field of SLA is driven by underlying questions. Like
any science or social science, SLA seeks answers to questions. The present
volume is unique compared to other similar introductions to the field because it is
organized around questions and not topics, theories, or research methodology. As
the reader will see, Chapter 1 situates the origins of contemporary L2 research
within the fundamental and original question of whether L2 acquisition is similar
to or different from first language (L1) acquisition. As we see it, all questions
currently driving the field stem from this fundamental question; that is, whether
stated or not, lurking behind current research in L2 acquisition is the issue of
similarities and differences between L1 and L2 acquisition. In most cases, this

15
issue is implicit, as the researcher tends to have an underlying assumption about
similarities and differences between the two contexts. So, for example, the
researcher interested in motivation in L2 acquisition – a factor of little interest to
researchers in L1 acquisition – may have already “determined” that something is
different about L2 acquisition and this is why motivation is important to examine
(see Chapter 8). As another example, researchers have wondered whether
instruction on the formal elements of language (what people typically call
“grammar”) has an impact on how formal elements are actually acquired. In
asking this question, the researcher is explicitly or implicitly making claims about
how different L2 acquisition is from L1 acquisition (where there is no
instruction).
It is impossible for a purposely short and introductory book to be exhaustive,
so for the present endeavor we have selected a limited number of questions that
we believe drive the bulk of L2 research (and theorizing) these days. We list them
here with some brief introductory remarks – and the reader will note that most
also form the chapter headings.

Are L1 and L2 acquisition fundamentally the same or fundamentally

different? We show how this question launched the field of SLA research
in the early 1970s in Chapter 1. We do not address this question in this
chapter but return to it in the Epilogue once the reader has reviewed
research on the major questions that form the focus of this book.

What does development look like? In Chapter 2, we describe various

aspects of the ordered and staged development of formal features of
language (e.g., sentence structure, inflections, and endings on verbs and
nouns) while also touching on variation in such ordered development and
any observed L1 influence on this development.

What are the roles of input and output? One of the fundamental findings of
the early research on L2 acquisition was the critical role that
communicatively embedded input plays as the data for language
acquisition. Chapter 3 reviews this idea while also addressing later

16
hypotheses that communicatively embedded output (i.e., learner
production) somehow affects acquisition.

What is the initial state? Of concern to researchers since the early 1970s is
whether or not L2 learners begin the task of acquisition by massively
transferring their L1 into the “hypothesis space” for L2 acquisition. In
Chapter 4, we explore research related to this question.

Can L2 learners become nativelike? In Chapter 5 we address this question,

looking at the outcome of L2 acquisition from a variety of perspectives.
What will emerge from this discussion is that nativelikeness is not an all-
or-nothing proposition, but that it likely depends on what aspects of
language and language use we examine.

Does instruction make a difference? Almost since the foundation of

contemporary L2 research there has been debate about the extent to which
instruction on the formal properties of language (i.e., what most people
call “grammar”) affects acquisition. We take up this question in Chapter 6.

What are the roles of explicit and implicit learning? Increasingly, L2

researchers are converging on the importance and fundamental nature of
implicit learning in L2 acquisition (i.e., learning without intent and
without awareness of what you are learning). However, there is debate
about the contribution of explicit learning and explicit processes in L2
acquisition. In a sense, the question raised here is related to the previous
question about instruction. We take these issues up in Chapter 7.

What are individual differences and how do they affect acquisition?

Individual differences refer to non-linguistic things such as motivation,
attitude, aptitude for learning, and working memory, among others. In
Chapter 8, we focus on the most researched of the individual difference
variables, namely, motivation, aptitude, and working memory. We
purposely situate this chapter and the question it addresses last in the book

17
 as much of how we interpret the research involves an understanding of
how we answer other questions.

Because this is an introductory volume, we have purposely avoided a

“scholarly style and tone.” Instead, we have used a style that we hope is reader-
friendly for the novice student of SLA. The three authors are experienced
educators, having taught courses on SLA to teachers-in-training, novice
researchers, graduate students, and other populations. We have concluded that
introductory books that assume little background on the part of the reader are
difficult to come by. So our target audience, again, is that novice student who is
reading about the field of SLA for the first time. By extension, the reader could
also be someone outside the field such as an administrator, parent, or anyone
interested in the basics of L2 acquisition. We assume no background in any area
related to SLA (e.g., linguistics, cognitive psychology) and do our best to explain
as we go and simplify complex concepts where possible. In so doing, and in
trying to keep this book as short as possible, we have necessarily been brief on
some of the topics. Some topics that readers might find in a lengthier and more
advanced work are not included in this volume. In addition, the reader will not
find an exhaustive review of research with citation after citation that is found in
tomes that are intended for more advanced readers and scholars (e.g., in the many
“handbooks” on second language acquisition that are on the market these days).
This book is different in intent and is not meant to compete with books used in
advanced courses or advanced degree programs.
Along the same lines, we narrow our scope of discussion to what are
traditionally called “syntax” and “morphology.” Syntax refers to sentence
structure while morphology refers to word structure (e.g., roots, inflections on
words). We do this purposefully. One reason is to keep the book short. A second
reason is that the crux of what we wish to talk about – differences between L1 and
L2 acquisition – has largely been tied to these domains of inquiry. This does not
mean that such things as the sound system (phonology), vocabulary (the mental
lexicon), and other things aren’t of interest to L2 acquisition more generally, and

18
occasionally we make reference to them. But when it comes to comparing L1 and
L2 acquisition, the locus of discussion has tended to be syntax and morphology.
Because we have limited the scope of inquiry, we don’t intend that readers will be
experts in the field by the time they turn the last page. Instead, we hope to instill
curiosity in readers so that continued engagement with more detailed volumes as
well as primary sources is a result. For those readers who will not pursue more
reading or work in the field of SLA, we hope we provide enough of the “basics”
for whatever purpose that reader has in mind.
We also need to admit our biases up front. All scholars in SLA come to the
task of synthesizing the field from a variety of perspectives. The three of us are
trained linguists and thus we come to the task with a linguistic slant or bias. For
example, we assume that the product of acquisition is an abstract, complex, and
implicit system called “language” that defies easy explanation and looks nothing
like textbook rules or descriptions language learners might find in a Google
search. This perspective pushes us to always look at the research in terms of what
it means to speak to the creation of this underlying “mental representation” of
language. The particular approach that underlies this book is called “generative,”
which we will discuss in Chapter 1. In linguistics, the generative approach is the
dominant framework informing theoretical and empirical advances. However, not
all L2 researchers adopt a generative approach. Many have no theoretical
approach to language at all. But we are firm believers that if we want to talk about
language acquisition, we have to have a working theory of language. Readers who
go on to pursue academic work in the field of SLA will encounter other
approaches and can compare them to the one taken in this volume. Still, where
appropriate, we point the reader to alternative ideas. And because none of us
works in the field of social and cultural aspects of acquisition, the book leaves
these perspectives to other experts. We do, however, touch on social aspects of
acquisition in a special feature called “What about Social Factors?” in which we
offer brief glimpses into how we might bridge the gap between linguistic
approaches and social approaches. This may not be enough for the experts in

19
social factors, but for an introductory book with a linguistic focus, we hope our
intent is well taken.
We also point out that in addition to readings and follow-up questions at the
end of each chapter, internal to each chapter are three features we hope will help
the novice reader. One feature involves bolded words and phrases. Bolded words
and phrases represent key constructs and ideas. Although these constructs and
ideas are defined as we go, they are also included in a glossary at the end of the
book. A second feature consists of call-outs, which are intended to draw the
reader to a main idea in a particular section of the chapter. These are useful for
reviewing after reading the chapter to recall key ideas. A third feature consists of
boxes called “Consider this …” The intent of these boxes is to invite readers to
stop and think about something they have just read, to reflect before going on. We
have found that such “stop and think” suggestions help readers consolidate ideas
and also break up the reading process. If this book is used in a class, such boxes
can also be used for classroom discussion.
We hope we have done our job for the novice reader. We hope we have done
our job for the instructor who adopts this text for classroom use. Most
importantly, we hope to instill in that novice reader a sense of wonder at the
complexity of L2 acquisition both as a human endeavor and as a research
enterprise. Although we have come a long way since the foundation of
contemporary L2 acquisition research in the late 1960s, there are still things to be
sorted out, research methodologies to be explored, and details that need
examination. But at the heart of it all are the key questions that drive what we as
researchers do, whether we acknowledge it or not. We hope these key questions
remain with the reader long after finishing this book.

20
 1
What Are the Origins of Second
Language Acquisition as a Research
Field?
◈

In science, knowledge builds upon previous knowledge. New areas of inquiry and
new disciplines have their origins in work that preceded them. Modern-day
physics is built upon the physics of the nineteenth century, for example, which in
turn was built upon Newton’s ideas, which in turn had roots traceable to ancient
Greece.
Although we may not be able to trace contemporary second language (L2)
research to Plato or Aristotle, we can describe its origins in the twentieth century.
And like other sciences, L2 acquisition also had its predecessors: linguistics and
first language (L1) acquisition to be exact.

21
 The Revolution in Linguistics and L1 Acquisition
The 1950s and the 1960s were an exciting time for both linguistics and language
acquisition. Until that time, linguistics had been dominated by what was called
structural linguistics. Structural linguistics looked at patterns of language; that
is, it looked at observable aspects of sentences as they were actually used by
people in order to make statements about the rules of language. For example, the
structural linguist would look at the words cat, cats, dog, dogs, plate, and plates
and determine that a regular singular noun is turned into a plural noun by adding
an ‘s’ sound. That ‘s’ sound added to nouns would be called a “plural morpheme.”
(A morpheme is the smallest unit of language that carries meaning. Note that the
‘s’ sound on a noun carries the meaning of “more than one”: dog = one dog, dogs
= more than one dog. We will examine morphemes in detail in Chapter 2.)
Similarly, the structural linguist might look at sentences such as John is tall, Is
John tall?, Mary is studying, and Is Mary studying? and conclude that a simple
yes/no question with the verb be is formed by inverting the subject and the verb
be. That linguist would also look at the following sentences, Megan likes
linguistics and Does Megan like linguistics? and determine that a simple yes/no
question with a full verb is formed by inserting an appropriate form of do before
the subject and eliminating any inflection (ending) on the verb. Thus, the
structural linguist would have two different “rules” or patterns for the formation
of yes/no questions in English: one for be questions and one for do questions. At
the same time, the structural linguist might look at what are called wh-questions:
those questions containing interrogative words such as why, who, and where.
They might look at sentences such as John went to the store and Where did John
go? and determine yet a third rule for how questions are made in English.
At the same time, human (and animal) learning was dominated by the
psychological theory of behaviorism. Under behaviorism, learning just about
anything was seen as a result of an organism’s response to a stimulus and the

22
subsequent reaction to that response. Appropriate responses got some kind of
reward. Inappropriate responses would get no reward or maybe even some kind of
negative reaction. For example, a chicken runs to the gate of the coop (reaction)
every time it hears the farmer approaching (stimulus). Why? Because in the past
the chicken has repeatedly received food when the farmer comes to the coop. The
chicken associates the footsteps of the farmer with impending food (reward).
Another example is the daughter who answers the phone and hears her mother
say, “I’m so glad you’re home” (stimulus). The daughter tightens up inside
(reaction) because she associates this phrase with all the times in the past her
mother has called and used the same phrase before delivering bad news or
complaining to her daughter about something. This kind of learning was termed
operant conditioning.

Under operant conditioning, language was seen as a behavior and not a

mental construct. As such, it was learned like any other behavior: through
the learning system of stimulus, response, and reward

B. F. Skinner applied behavioral psychology to language acquisition in his

seminal book Verbal Behavior published in 1957. His claim was that children
learn the structural patterns of the language around them the same way they learn
anything else: through the learning system of stimulus, response, and reward.
Skinner saw language primarily as a behavior and not as a mental construct. As
such, he argued that language was learned like any other behavior. The prevailing
notion of the time was that children imitated (response) what they heard
(stimulus) and because of continued conversation by parents (one type of reward)
or because parents said “how cute” and smiled (another type of reward) or
because the child received something like a drink or food (yet another type of
reward), children acquired language.
But something was brewing in the 1950s. For some, the basic ideas of
structural linguistics and behaviorism as an explanation for language acquisition

23
just didn’t sit right. Two publications appeared at roughly the same time that gave
voice to the concerns of those scholars. In 1958, Jean Berko Gleason published
her famous study in which she tested young children’s knowledge of word
endings (inflections) for nouns, verbs, and adjectives in English. Without going
into details here (see Exemplary Study), Berko found that children did not just
imitate the speech around them. Instead, something internal to the child was
extracting rule-like knowledge from the examples they were storing in their
minds/brains. The conclusion from Berko’s study was that children were much
more “active” learners of language than could be claimed under something like
behaviorism. Children weren’t just imitating habits and then getting rewarded.
They were actively (re-)creating language in their minds as they were exposed to
language in their environment.
The following year, in 1959, Noam Chomsky published his famous critical
review of Skinner’s Verbal Behavior. In this seminal essay, Chomsky argued two
basic points. The first was that unconscious linguistic knowledge was much more
abstract and complex than a set of internalized habits, laying the groundwork for
what would later be called the generative revolution in linguistics (more on that
later). As we will see, Chomsky did not see language as something relatively
simple to learn via stimulus and reinforcement, but as a very complex and
intricate system that defied description by the psychologists of the time. For him
language was not behavior; it was a mental construct. In other words, language
was a system that existed inside the mind; something we could study through
introspection. So we can introspect and determine that the word himself in John
knows himself well refers to John and not to somebody else. But we can also think
about the sentence John told Bobby to buy the picture of himself and determine
that himself can only refer to Bobby and not to John. Then there’s a third
sentence, John showed Bobby a picture of himself, and we might say “Geez.
Himself can refer to either Bobby or John.” And finally, we might consider the
sentence Himself knows John pretty well and say, “Well, that sentence is just not
possible.” We are not gathering sentences from people speaking. We are

24
introspecting and thinking about language. And as we introspect, we begin to
think about how language is constructed inside the mind.
Chomsky’s second point was that learning anything could not be reduced to
operant conditioning, no matter what neat experiments with rats, dogs, and
chickens would suggest. He forcefully critiqued the very basic constructs of
operant conditioning (e.g., stimulus, response, feedback) and showed how
inadequate they were when one considered how quickly children acquire both
language and concepts. He argued that children were not blank slates to be written
on by their environments, but instead came to the learning task with something
internal that guided their active processing of information to convert it into
something usable at a given point in time. In a sense, Chomsky argued that
children came prewired to learn certain things in certain ways. One of those things
was language.

Exemplary Study

Berko, J. (1958). The child’s learning of English morphology. Word, 14,

150–177.

25
 Participants
12 native-speaking English adults

19 children learning English as an L1 between 4 and 5 years old (12

girls and 7 boys)

61 first-graders (5–7.5 years old, 35 girls and 26 boys)

26
 Major Target Items
noun plurals

past tense on verbs

third-person singular

singular and plural possessive

comparative and superlative of adjectives

Note: The items involved largely nonsense words for a total of 28 items.

27
 Materials and Procedure
picture cards with simple line drawings

testing children individually, the experimenter held up a picture card

and prompted the child to complete a thought. Example: [a picture with
one bird-like animal] “This is a wug.” [another picture with two of the
bird-like animals] “Now there is another one. There are two of them.
There are two _____.” [the expectation was for the child to complete
the sentence]

28
 Results
The 12 adults were unanimous in the way they supplied endings to
nouns, verbs, and adjectives. Example: wug → wugs [wʌgz]; gutch →
gutches [gʌʧɪz].

There were no differences between boys and girls when matched for
age.

First-graders were better than pre-schoolers on some items, but of

interest is where the two groups clearly deviated from adult-like
patterns. These included plurals that needed to add /-ɪz/ as in gutch →
gutches, past tense that required /-ɪd/ as in mot → motted, and third-
person singular that required /-ɪz/ as in nazz → nazzes. What emerged
was that children had adult-like rule-based knowledge for adding
simple consonants and could do so on the nonsense words, but they
tended not to be able to add an ending to nonsense words that required a
syllable such as /-ɪz/ or /-ɪd/. So, they could easily pluralize wug →
wugs [wʌgz] but could not easily pluralize gutch → gutches [gʌʧɪz].

29
 Conclusion
Children possess unconscious rules for noun, verb, and other
inflections. They are not limited to imitating what they hear in the
environment.

After the Berko Gleason and Chomsky publications, it seemed as though the
dam had broken, and pushback against both structural linguistics and behaviorism
began. Chomsky followed up his critique of Skinner’s work with a number of
publications that launched a new kind of linguistics called generative linguistics.
The most cited of his early works was Aspects of the Theory of Syntax, which
appeared in 1965. Generative linguistics sought to understand a number of things
about language, including – but not limited to – exactly how languages were
different and why they were different, what hidden properties there might be in
languages that we don’t see when we hear or read sentences, and how language is
organized in the mind/brain of the competent adult speaker. The basic idea behind
generative linguistics was that a person could generate an infinite number of
sentences yet must have a finite set of “rules” to work with. How was this
possible? And how was it that children acquired these “rules” without help?

Generative linguistics sought to understand a number of things about

language, including but not limited to exactly how languages were
different and why they were different, what hidden properties there might
be in languages that we don’t see when we hear or read sentences, and
how language is organized in the mind/brain of the competent adult
speaker.

30
 Although generative linguistics has evolved over the years, it remains
committed to understanding language as mental representation (competence) and
not as a set of patterns observable in verbal behavior (some kind of performance).
If we return to the examples of the yes/no questions from earlier, generative
linguists would look to see not two different rules for yes/no questions depending
on whether the verb was be or do, but instead to deeper properties that might
actually reflect a “rule” that unified both types of question. The generative
linguist might also include the wh-questions we saw earlier in this analysis to see
if there is something common to all three types of question. Let’s look at a classic
example that Chomsky used with questions with be: The man who is tall is sick
and its corresponding yes/no question Is the man who is tall sick? A simple rule or
pattern for yes/no questions with be would be to move the first be to the front of
the sentence, based on many examples one might hear, such as John is sick/Is
John sick?, Mary is studying/Is Mary studying?, and The baby’s diaper is wet/Is
the baby’s diaper wet? But moving the first be in The man who is tall is sick
results in an incorrect question: *Is the man who tall is sick? (An asterisk is used
in linguistics to indicate a sentence that is not permitted in a language.) Chomsky
reasoned that there must be something going on deeper and more abstract than
“move the first be to make a yes/no question.” What is more, he pointed out that
no child ever makes an error such as *Is the man who tall is sick? during language
acquisition. And indeed the early research on child language acquisition showed
that children’s non-adult language was constrained; potential errors based on
“imitation” and surface features of sentences just never happened. Chomsky
surmised that something internal was guiding both the nature of language and the
nature of language acquisition.
On the heels of Berko Gleason’s work, data on L1 acquisition began popping
up in the United States and Europe. Roger Brown’s famous work on child
language acquisition began to appear in the 1960s and culminated in his 1973
seminal book A First Language. In his work (and his colleagues’ as well), Brown
showed that language acquisition was stage-like and that particular parts of

31
language were acquired in a particular order in English, regardless of children’s
social environment or status (as long as the children were not mentally impaired).
As a preview to what we will see later, Brown showed, for example, that all
English-speaking children first acquired -ing on the ends of verbs (e.g., The car is
slowing down) before past-tense endings (e.g., The car slowed down) and both of
these before present-tense third-person -s (e.g., The car slows down). No child
acquires past tense before -ing, for example.
Also emerging in the 1960s and 1970s was work by Ursula Bellugi on
developmental stages. What Bellugi showed was that in the acquisition of a
particular structure, such as negation, one could find stages. That is, children do
not magically begin producing negative sentences such as John does not live here
but instead go through stages in which bits and pieces of the syntax of negation
are added and refined until the child arrives at what looks like an adult-like
representation. Again as a preview to what we will see in a later chapter, Bellugi
showed that children start off with simple no + word or phrase to indicate
negation (e.g., no milk, no more soap) followed by something that looked like a
sentence because it had a subject and verb but no was still the negator (e.g., I no
want milk). The staged development would continue until finally the child could
produce helping verbs (do, have) and modals (e.g., can, would, might) with the
correct negator not (e.g., He doesn’t like milk, I can’t fix this). Ordered
development, whether the acquisition of a group of morphemes or the acquisition
of a particular construction, showed up repeatedly in child language acquisition
and in various different languages around the world.
At the same time, researchers began to look at the interactions between
children and their caregivers in order to ascertain whether feedback (e.g.,
reinforcement, correction) and other aspects of behaviorism were at play. None
could be found. In fact, repeated attempts to deliberately teach children language
and to correct them did not work. Note: We’re not talking about school-aged
children who are corrected for the use of ain’t, for example, by teachers and
parents. We are talking about 2- and 3-year-olds corrected for feets and Nobody

32
don’t like me, as in the following example from David McNeill, published in the
mid 1960s. A mother is attempting to overtly correct her child to see if the
correction works.

CHILD: Nobody don’t like me.

MOTHER: No, say “nobody likes me.”

CHILD: Nobody don’t like me.

(Eight repetitions of this exchange)

MOTHER: No, now listen carefully; say “nobody likes me.”

CHILD: Oh! Nobody don’t likes me.

McNeill’s example was a deliberate attempt by one mother. Such attempts

are rare. What the research did show was that parents and caregivers corrected
children only for content or truth-value. Note in the following interchange how
the parent does not correct for the form of the verb but for what the child is
talking about.

CHILD: Look. Doggie lie down! (pointing to a cow in the field)

PA R E N T: No, honey. That’s not a doggie. That’s a cow.

CHILD: (still pointing) Cow! Cow lie down!

PA R E N T: Yes. Maybe he’s tired.

By the time the mid 1970s had arrived, child language acquisition research –
along with the generative revolution in linguistics – had pretty much sounded the
death knell for behaviorism as an adequate explanation for child language
acquisition. Children did not imitate their parents’ language and get rewarded for
it. Yes, they got rewarded for participating in conversations and for being truthful,
but not for how they said anything. Children seemed to gradually work out the
“rules” of language on their own without external influences.

What about Social Factors?

33
The picture painted here is that there is some sort of universalism in child
language acquisition for speakers of the same language. A conclusion
might be that social factors or context doesn’t matter. Social factors and
context matter a great deal for a variety of things in child language
acquisition but not for the formal properties of language (e.g., syntax,
rules of the sound system). Social context affects how children participate
in conversations, the particular words they learn as opposed to others, and
even how they might eventually see themselves as speakers of a particular
language. For example, children in one social context may be raised so
that swearing is okay, while others may be raised in an environment where
swearing is considered ill mannered and to be avoided.
Of course, social context influences the particular dialect of a
language that a child learns (e.g., English in South Boston, English in
rural Georgia, English in Belfast, Black Vernacular English). Dialects are
clearly tied to identity and children tend to identify with the community in
which they are raised.
And, to be sure, social context and factors affect communicative
modes that use language, such as writing and composition. Here the effect
is not on the language itself (i.e., the formal properties) but on style,
rhetoric, word choice, spelling, and even whether a person writes or
doesn’t.
But, again, social context does not seem to affect the acquisition of
the underlying formal properties of language (i.e., how these things
develop over time, as in the Berko Gleason and Brown studies). So, the
child learning English in South Boston, the child learning English in rural
Georgia, and the child learning English in Belfast, all acquire -ing before
they acquire third-person -s, for example.

34
 The Revolution in Second Language Acquisition
As in child language acquisition prior to the 1960s, L2 acquisition theory was
dominated by behaviorism. L2 learners were viewed as acquirers of new habits,
and these habits were learned using the same constructs as in L1 acquisition:
stimulus, response, feedback, and their correlates. The situation was seen to be
more complicated for the L2 learner because while learning new habits, the
learner had to suppress old habits – the habits of the L1. Language-teaching
circles in the military latched onto these ideas with the advent of the Army
Method in the 1940s, which was translated into academic environments as the
Audiolingual approach in the late 1950s and early 1960s. This approach to
language acquisition in the classroom involved memorization of dialogues and
various kinds of “drills” to practice vocabulary and sentence patterns. Suppression
of L1 habits and the avoidance of errors were central to the Audiolingual
approach. Errors were seen as either failure to learn or intrusive in the learning
process. What is more, the L1 was considered to be the source of all errors. These
ideas are best represented in the works of Robert Lado and his contemporaries.
Spurred by the rapidly changing landscape in child L1 acquisition and
linguistics in the 1960s, several scholars interested in L2 acquisition began to
question the application of behaviorism to non-native language acquisition. In this
regard, two publications stand out as pioneering – and should be read by any
student of L2 acquisition. The first was S. Pit Corder’s famous essay, “The
Significance of Learners’ Errors” published in 1967. Corder argued that
systematic errors in L2 acquisition were not symptomatic of faulty learning, but,
as in child language acquisition, were windows into how the mind processed and
stored language. Better yet, they were ways to understand how L2 learners (re-
)created language in their heads. By careful study of “errors,” we could come to
know how L2 acquisition proceeds, determine some of the underlying processes,

35
and look for commonalities across learners and languages. As Corder himself
said,

A learner’s errors provide evidence of the system of the language that he is

using (i.e., has learned) at a particular point in the course (and it must be
repeated that he is using some system, although it is not yet the right system)
… [Errors] provide to the researcher evidence of how language is learned or
acquired, what strategies or procedures the learner is employing in the
discovery of the language.

(Corder, 1967, p. 167)

36
 Consider this …
Corder discussed the significance of learners’ “errors” and claimed that
they would probably provide evidence of a system. And if L2 acquisition
turned out to be like L1 acquisition in that the learner did not learn by
imitation but by gradually building up an internal system over time, are
“errors” actually errors? Think of this: An adult frog starts out as an egg,
eggs hatch to release tadpoles with gills, tadpoles grow back legs, then the
tadpoles’ gills are replaced by lungs as they also grow front legs, and then
the adult frog emerges. Is a tadpole an error because it isn’t a frog yet? Is a
tadpole with legs and gills an error because it doesn’t have lungs?

In 1967, the linguist S. Pit Corder argued that systematic errors in L2

acquisition were not symptomatic of faulty learning but, as in child
language acquisition, were windows into how the mind processed and
stored language.

The other publication that helped launch the field of contemporary research
on L2 acquisition was Larry Selinker’s 1972 essay, “Interlanguage.” In an
ambitious attempt to connect many ideas, Selinker argued a number of points.
Here, we will consider only three. The first was that learners create what he called
an interlanguage. This language was neither the L1 nor the L2 but was reflective
of how the learner moved from one toward the goal of another. In this way,
Selinker likened L2 acquisition to L1 acquisition but added in the one ingredient
that the L1 learner did not have: an L1. That is, he added language transfer or
influence as an essential ingredient in L2 acquisition. A second idea was that
language was a complex and abstract mental representation in the vein of
Chomsky’s generative perspective. As such, L2 learners were as active in (re-

37
)creating this representation as L1 learners were. A third major idea that Selinker
offered was that L2 acquisition, for whatever reasons, was not uniformly
successful. Whereas all children growing up in the same environment converge on
the same linguistic system (mental representation), L2 learners do not. They are
routinely non-nativelike in a variety of domains. Selinker termed this aspect of L2
acquisition fossilization.
Inspired by Roger Brown’s empirical work in L1 acquisition, research on
ordered development in the L2 context began to appear in the 1970s. Although we
will visit this line of research in some detail in Chapter 2, a number of pioneers
launched this line of empirical L2 research. (Prior to this time period, there was
speculation about acquisition vis-à-vis behaviorism but no actual empirical
research.) These pioneers included Heidi Dulay and Marina Burt, who in a series
of studies in the early to mid 1970s demonstrated that child L2 acquisition of
morphemes was ordered over time independently of the L1. For adult L2 learners,
Stephen Krashen and his colleagues conducted similar research on the acquisition
of morphemes in the mid to late 1970s. Again, ordered development was found
and it looked similar to what children did.
About the same time and into the early 1980s, work inspired by Ursula
Bellugi began to appear as L2 scholars looked at the acquisition of particular
structures, namely negation, as in John doesn’t live here, and wh-questions such
as Where does John live? and What time are you coming? This work was
published by scholars such as Henning Wode, John Schumann, Roar Ravem, and
a number of others. Quickly, the research on ordered development spread from
English as an L2 to other languages, notably Spanish, German, and French, with
both classroom and non-classroom language learners.
In short, research on L2 acquisition was virtually non-existent prior to the
1970s. But because of the pioneering work in L1 acquisition and the field of
linguistics – and also because of the seminal essays by Corder and Selinker –
research on L2 acquisition seemed to explode, and by the 1980s, L2 research was
a fruitful domain of inquiry. It was a young field but was quickly taking off. And

38
it was challenging previously held notions about how second languages are
acquired.

39
 The Driving Question
Although the flurry of research that cropped up in the 1970s and continued
afterward was inspired by the work in L1 acquisition, very quickly the field began
to coalesce around a singular question: Is L2 acquisition like L1 acquisition? To
the average person on the street this may seem like a dumb question. Of course L2
acquisition is not like L1 acquisition! Adults are adults and children are children.
Their brains are different, their motivations are different, their environments are
different … and the list continues. Nonetheless, both Corder and Selinker
suggested that at least in some respects – perhaps at their core – L1 and L2
acquisition shared similarities and probably involved identical underlying
processes in spite of the differences we see on the outside. The key term here is
“underlying,” with the suggestion that no matter what the motivations,
environments, or other non-language-related factors, something deep in the mind
of humans was at the core of both L1 and L2 acquisition. Here is a quote from
Corder on this matter:

It still remains to be shown that the process of learning an L2 is of a

fundamentally different nature from the process of primary acquisition. If we
postulate the same mechanism, then we may also postulate that the
procedures or strategies adopted by the learner of the L2 are fundamentally
the same.

(Corder, 1967, p. 164)

The driving question that got the field of L2 acquisition going was
whether it was similar to or different from L1 acquisition in terms of
underlying processes.

40
 As the empirical research emerged in the 1970s and 1980s, at first it seemed
as though L1 and L2 acquisition were more similar than they were different. As
we will see later on ordered development (Chapter 2), there was considerable
overlap between ordered development in English L1 and English L2, for example.
But enough differences seemed to surface that the pendulum swung in the
opposite direction and some claimed that L1 and L2 acquisition were different.
One of the most influential hypotheses to emerge came from Robert Bley-Vroman
in the late 1980s and was called the Fundamental Difference Hypothesis. We will
review Bley-Vroman’s hypothesis in detail in Chapter 5. For now, we will suggest
that answering the question of L1 and L2 acquisition similarities and differences
hinges on how we define terms like “process,” “procedure,” “strategies,” and so
on. What is more, it also hinges on what kind of data we look at – a point we will
touch on below – and what we mean by both “similar” and “different.”
We will not answer the question of whether L2 acquisition is the same as L1
acquisition here. It is our aim to lay out the issues in each chapter of this book and
then revisit this question in the Epilogue. For now, we will simply say that, for
whatever reason, many scholars simply decided that L2 acquisition was different
from L1 acquisition and the early driving question of the empirical research
seemed to have faded by the 1990s. Indeed, many people accepted Bley-
Vroman’s hypothesis that the two acquisition situations were fundamentally
different. However, it is our contention that the question has never really gone
away and remains central to the research and theorizing in L2 acquisition. It lurks
beneath the major questions that scholars have been addressing over the years –
the questions that are the core of this book. For this reason, you will see us return
to this driving question from time to time and, again, address it in a summary
form in the Epilogue to this book. It is also our contention that Bley-Vroman’s
hypothesis was flawed in a number of ways and not sustainable given the vast
amount of research conducted since the 1980s. Does this mean that L1 and L2
acquisition are fundamentally the same? Once more, we will refrain from
addressing that question until the end of this book.

41
 Some Observations on Data Types
A recurring difference in comparing L1 and L2 acquisition is the type of data
gathered. Because pre-school children tend not to be literate, the kinds of data
gathered are oral or comprehension based. Oral data may be spontaneous; that is,
the researcher may simply put a recording device in a room and record the child
during play, eating, and other everyday activities. Oral data may also be elicited
using puppets or pictures. If you review the Berko experiment in the Exemplary
Study in this chapter, you will see the use of elicited oral data with children.
Another kind of oral data used with children are those collected when the
researcher says something like “Bill, ask the lady in the picture where she wants
to go” or “Ask the doll what time it is.”
Researchers might also collect comprehension data in L1 acquisition. For
example, researchers have tested children on active sentences (The cow kicked the
horse) and passive sentences (The horse was kicked by the cow). The child would
be given stuffed-animal toys and asked to act out a sentence spoken by the
researcher. If the child understood the sentences, he or she would demonstrate this
by manipulating the stuffed animals accordingly. The real target of such research
was the comprehension of passives. If children interpreted passives like actives,
they would mistakenly show the horse kicking the cow for The horse was kicked
by the cow (which pre-school children regularly do – it seems that passive
sentences are a “late-acquired structure” in English L1).
Clearly, researchers are limited in the kind of data they can obtain from pre-
school and pre-literate children. However, the same is not true for, say,
adolescents and adults acquiring an L2. Nonetheless, in the early days of L2
research, researchers tended to use the same kind of methods used in child
language acquisition research. For example, the research conducted by Heidi
Dulay and Marina Burt mentioned earlier used a picture narration task, as did
some of the research by Stephen Krashen and his colleagues, also mentioned

42
above. In these studies, participants saw an array of pictures that told a story. The
researcher would ask participants a question or several questions about each
picture to elicit language production focused on meaning (i.e., talking about the
story). Other researchers used semi-spontaneous oral production data in which
they interviewed L2 learners about topics related to their lives, hopes, dreams,
experiences, and so on. Still others used comprehension data, and some combined
both oral and comprehension data.
Since the early days of L2 research, however, researchers have generally
moved away from oral production data – especially spontaneous and semi-
spontaneous oral data. Just about every type of data has been used. Here is a
partial list.

spontaneous writing (e.g., email samples, chat room samples)

elicited writing (e.g., sentence combining, picture description, video

summary, compositions assigned in classes)

grammaticality judgments (indicating whether a sentence is possible or

impossible in a language)

truth-value tasks (indicating whether a sentence “fits” with what preceded

it)

sentence interpretation tasks, aural and written (e.g., matching what is

heard/read to a picture, answering questions about who did what to whom)

self-paced reading (comparing reading times on different kinds of

sentences, e.g., grammatical versus ungrammatical, logical versus
illogical)

eye-tracking (e.g., tracking unconscious pupil movement during sentence

reading using similar stimuli to self-paced reading)

paper-and-pencil tests (e.g., cloze tests in which every fifth, seventh, or

tenth word has been deleted from a reading and the participant fills them

43
 in, classroom tests, results from standardized language tests such as the
TOEFL)

grades in language courses.

Again, this list is partial. You might guess that the kind of data collected
depends on what the researcher is looking at, and you will be partially correct.
However, over the short life span of L2 research, we have seen almost an
explosion of techniques to gather data – and we have also seen that questions
researched in the past could be researched again with a new data-gathering
technique to see if the same results obtain. We have also seen that sometimes
techniques are misapplied or that conclusions are reached prematurely because the
data collected are limited in what they can actually address.
These and other issues regarding the type of data collected and the
conclusions based on those data will surface periodically in this book. As we will
see, the issue always comes back to whether or not a particular technique
measures underlying mental representation. For now, if you are unfamiliar with
empirical L2 research, it is important for you to simply keep in mind that two
researchers with two distinct data-collection techniques could come to different
conclusions regarding the same question they are investigating. This makes the
field a bit complicated and there have been a number of volumes already
published on data-collection techniques in L2 research.

44
Consider this …
Given the differences in types of data used in L1 and L2 acquisition, do
you think they have an impact on how we view first “versus” second
language acquisition and whether they are similar or different? In what
way?

45
 Conclusion
As of the writing of this book, L2 acquisition as a research field is only 50 years
old. If we take Corder’s 1967 essay as the starting point for all contemporary
theory and research on L2 acquisition, then compared to other empirical fields,
including the sciences and many social sciences, L2 research is a relative
newcomer. After all, physics, chemistry, and math have been around for centuries
if not millennia. Nonetheless, a lot has happened in five decades in terms of
research findings, theory development, debates, and general growth of the field.
When Corder published his seminal essay, a person could count on one hand the
number of journals devoted to L2 theory and research. Now, the journals number
well beyond a dozen (e.g., Studies in Second Language Acquisition, Language
Learning, Second Language Research, Linguistic Approaches to Bilingualism,
Bilingualism: Language and Cognition, Applied Psycholinguistics, The
International Review of Applied Linguistics, Instructed Second Language
Acquisition, The Modern Language Journal). In addition, major publishing
houses in Europe, North America, and elsewhere have series and sub-series
devoted to L2 research (e.g., Routledge/Taylor Francis, John Benjamins,
Multilingual Matters, Blackwell, Cambridge, Oxford, Bloomsbury, Georgetown
University Press). There are dozens of conferences in which L2 research forms
the core of the conference or is regularly featured in the conference (e.g., The
Second Language Research Forum, The American Association of Applied
Linguistics, The British Association of Applied Linguistics, EuroSLA, Generative
Approaches to Second Language Acquisition, The Boston University Conference
on Language Development).
All of this leads us to the simple claim that L2 acquisition is a vibrant and
busy field of inquiry. Although its roots are in linguistics and child language
acquisition, the field of L2 research has evolved to include cognitive and social
perspectives. Sometimes these perspectives are at odds. Sometimes they are not.

46
And sometimes it seems they are at odds but a closer look suggests that perhaps
they are not.

47
 Recap
In the present chapter, we have touched upon these major ideas and issues.

In the 1950s and 1960s, we saw the beginnings of a revolution in

linguistics and in language acquisition that would challenge prevailing
beliefs.

Language is not a set of rules or patterns but something much more

abstract and generative in nature.

L1 acquisition does not involve the internalization of “habits” mimicked

from the environment. Instead, a linguistic system evolves in the child’s
mind/brain over time. That evolution involves staged and ordered
development.

The revolution in L2 research began in the 1970s and demonstrated that as

with children, a second language is not learned through repetition and
imitation. Instead, the learner’s linguistic system evolves over time.

The initial driving question – and one that lurks beneath the surface of
current research – is to what extent L1 and L2 acquisition are the same
thing or involve the same processes.

How we measure acquisition (i.e., the type of data we examine) influences

how we talk about acquisition and the conclusions we draw from it.

48
 References and Readings

Bellugi, U. (1967). The acquisition of negation. Unpublished doctoral

dissertation, Cambridge, MA: Harvard University Press.

Berko, J. (1958). The child’s learning of English morphology. Word, 14,

150–177.

Brown, R. (1973). A first language: The early stages. Cambridge, MA: Harvard
University Press.

Chomsky, N. (1957). Syntactic structures. The Hague: Mouton.

Chomsky, N. (1959). A review of B. F. Skinner’s Verbal Behavior. Language, 35,

26–58.

Corder, S. Pit. (1967). The significance of learners’ errors. International Review

of Applied Linguistics, 5, 161–170.

Dulay, H. & Burt, M. (1974). Natural sequences in child second language

acquisition. Language Learning, 24, 37–53.

Krashen, S. D., Houck, N., Giunchi, P., Birnbaum, R., Butler, J., & Srei, J.
(1977). Difficulty order for grammatical morphemes for adult second language
performers using free speech. TESOL Quarterly, 11, 338–341.

Pinker, S. (2007). The language instinct: How the mind creates language. New
York: Harper Perennial.

Rowland, C. (2014). Understanding child language acquisition. London:

Routledge.

Selinker, L. (1972). Interlanguage. International Review of Applied Linguistics,

10, 209–230.

49
Skinner, B. F. (1957). Verbal Behavior. New York: Appleton-Century-Crofts.

50
 Following Up
1. Read S. Pit Corder’s 1967 essay on the significance of learners’ errors (see
References and Readings). List and describe five major points he makes in
this essay.

2. It is common belief that children learn an L1 effortlessly and easily. We

now know the following to be true:

Acquisition begins in the womb as the later-stage fetus starts to process

sounds from the environment.

Children first babble, then enter a one-word stage.

At some point, children begin to put two words together, eventually

getting to something that looks like a sentence by the age of 3 years.

An adult-like grammar is in place by the age of 4.5 years but there are
gaps in some areas.

Some of these gaps aren’t filled in until the child is much older, well
into primary school and sometimes middle-school age.

It would seem that if children do nothing but spend their waking hours
learning language, they spend thousands of hours doing so: over 13,000
hours by the age of 4.5. How does this scenario fit with the “effortlessly and
easily” description of acquisition? What do you think people mean by
“effortlessly and easily”?

3. When children acquire the past tense in English, they tend to follow this
sequence:

In the first stage, verbs are not marked for past tense. Children use
“bare” verbs such as eat and talk when referencing past tense.

51
 In the next stage, children exhibit control over high-frequency irregulars
(e.g., ate, went) but not regular past tense.

In the subsequent stage, regular past-tense markings emerge (e.g.,

talked, looked) but there is also an unlearning of the irregulars so that
they are treated as regulars (e.g., eated or ated, wented or goed).

In a final stage, the irregulars reassert themselves (e.g., ate, went) and
past-tense marking looks adult-like. It takes the child until about the age
of 5 to get to this stage, and there may still be a few gaps for infrequent
verbs.

Can these stages be accounted for by imitation and reward? Why or why not?

4. Structural linguists were concerned with observable patterns of language.

As such, they were concerned with possible sentences in a language and tried
to describe those. Generative linguists have been concerned with both what
is possible and what is not possible in a language. For example, the
contraction wanna is possible in this sentence, Who do you wanna call about
this? but it is not possible in this sentence, *Who do you wanna call Jim
about this? (instead you have to say Who do you wanta/want to call Jim
about this?). Likewise, the contraction I’ve is possible in I’ve done it but not
possible in *Should I’ve done it? To look at what is possible and impossible,
generative linguists rely on intuitions about language or sometimes they ask
speakers “Which sounds better to you? X or Y?” Structural linguists relied
on what people said. Which do you think is a better way to determine
“what’s in people’s heads” (i.e., their underlying competence)?

5. Conduct an online search for recent Second Language Research Forum

(SLRF) meetings and/or for recent EuroSLA meetings. Examine the titles for
papers on L2 acquisition. What strikes you about the titles?

52
 2
What Does Development Look Like?
◈

In nature, there is ordered development. Any person who has taken a science class
has probably seen how chicken eggs are fertilized and what the development of a
chick embryo looks like over the 21 days it takes for the chick to emerge from the
shell. If not, you might want to Google “development of chick embryo.” What is
clear is that the development of the embryo is ordered or staged, such that
something must happen before other things can happen. This is all pre-
programmed into the genetics of the egg, of course. The same is true for the
development of an oak tree from an acorn and the development of a snowflake
from molecules of water.
So too in L2 acquisition there is ordered development. That is, there are
ways in which an L2 system grows in the mind of the learner that are universal
across all learners of the same language. Such ordered development does not
mean that there is no variation among individuals, but such variation does not
alter the stages themselves. And of course, in Chapter 8 we will touch on
individual variation that affects rate of development and how far learners get. To
examine ordered development, we will review four major research areas in L2
research:

morpheme orders

developmental sequences

U-shaped development

53
 markedness relationships.

At the appropriate points, we will also discuss the role of L1 influence in the
developmental process of L2 acquisition.

54
 Morpheme Orders
You may recall from Chapter 1 that morpheme order research began in response
to research in child L1 acquisition. Roger Brown studied how children acquire
particular morphemes over time in English. To review, a morpheme is the
smallest unit of language that carries meaning. So, a word like dog is a morpheme
and carries the meaning of a canine-like animal. The -s marker that is attached to
the end of a noun is a morpheme and means “more than one.” So, dogs consists of
two morphemes: the morpheme dog and the morpheme -s. The plural -s marker is
what is called an inflectional morpheme, meaning that it gets attached to a noun
to add to its meaning (i.e., dog has one meaning and dogs has the same meaning
with the added meaning of more than one). Likewise, the verb watch is a
morpheme and means using the eyes to focus on something. We can add the
inflectional morpheme -ed to the verb and add the meaning of “happened in the
past”: watch + -ed = watched.
As we touched on in Chapter 1, Roger Brown investigated how children
acquired various morphemes over time. These morphemes included inflectional
morphemes for nouns (plural marker, possessive marker, as in the boy’s hat), for
verbs (past-tense marker, third-person marker for present-tense -s as in eats,
progressive marker -ing as in eating), in addition to articles (the, a/an), copular
(“linking”) verbs (be, and contracted be, as in the boy’s hungry), be as auxiliary
(as in the boy is coming and the boy’s coming), irregular past-tense verbs (such as
ate, went), among a few others. What Brown found was that linguistically
unimpaired children, regardless of social status or home life, acquired the
morphemes in a universal order over time. The order looks like this (with a few
things left out for simplicity here):

progressive -ing (e.g., eating)

55
 plural -s

irregular past (e.g., went, ate)

possessive -s

full copula be (e.g., Snail is slow).

articles (e.g., a/an, the)

regular past (e.g., watched, cooked)

third-person -s (e.g., eats, cooks)

full auxiliary be (e.g., The snail is crawling)

contracted copula be (e.g., (The snail’s slow)

contracted auxiliary be (e.g., The snail’s crawling)

By acquired, Brown meant that in the samples he took, the children were
accurately using a morpheme 90 percent of the time in spontaneous speech over
several contiguous time periods. Of course, there are individual differences in
how long children take to acquire each morpheme and how long they take overall,
but the point here is that regardless of differences in rate, all children acquired the

56
morphemes in the same order over time. Brown’s work was subsequently
corroborated by others.
In the 1970s, L2 researchers began to look at children and adults acquiring
English as an L2 in and out of classrooms. Some of these researchers were Heidi
Dulay and Marina Burt and Stephen Krashen and his colleagues, among many,
many others. In fact, by the early 1990s, dozens of studies looking at morpheme
orders and their universality had been reported. Taken together, these studies
found that, with some slight variations, there was a universal order of morpheme
development for all L2 learners of English. This prompted Stephen Krashen to
claim there was a “natural” order of English morpheme acquisition for all learners
of English regardless of context, L1, age, and so on. His order looked like this,
with groups of morphemes acquired before other groups of morphemes:

-ing, noun plural, copula be

auxiliaries, articles

irregular past

regular past, third-person singular present tense (-s as in talks), possessive on

nouns (-s as in John’s)

Other researchers, notably Roger Andersen and Bill VanPatten, took another
look at the morpheme studies and questioned whether it was appropriate to group
noun-related morphemes with articles with verb-related morphemes (and so on)
all in the same study. After all, why should we care about the relative order of,
say, plural -s and past tense? Andersen and VanPatten independently looked at,
for example, just verb-related morphemes and found a universal and inalterable
order in all studies. What is more, this order matched that found for first-language
learners. That order looked like this:

57
 progressive -ing (eating)

regular past (e.g., watched, cooked)

irregular past (e.g., went, ate)

third-person -s (e.g., eats, cooks)

In short, when morphemes were tracked by the category to which they

belong, it looks like indeed there is a universal or “natural” order for L2
acquisition and that this order closely resembles the order for child L1 acquisition.

When morphemes are grouped by category (e.g., verb-related

morphemes), it appears that there is a universal, invariant developmental
order for all learners regardless of context, L1, and other external factors.

However, a universal morpheme order only obtained with spontaneous or

semi-spontaneous speech. That is, the data establishing morpheme orders for L2
learners was the same kind of data used for establishing orders for child L1
learners: speech samples in which the participants are actively engaged in
communicating information. The morpheme orders could be altered, especially
for adults, if non-spontaneous, non-communicative data were used, such as fill-in-
the-blank tests, untimed writing samples where participants’ attention was
directed to “how well they wrote,” and other activities where the adult’s attention
could be directed to what Stephen Krashen has called monitored output.
Monitored output can occur in those situations in which a learner has formally
studied grammar and under certain conditions (e.g., there is time) can apply
consciously learned rules and formal features to edit speech or writing. Most
researchers now accept that monitored output is not reflective of what is actually

58
in a learner’s implicit mental representation of language, a point that will be
further developed in this book. For this reason, the kind of data researchers use to
talk about language acquisition becomes paramount to understanding L2
acquisition.

59
 Consider this …
Why would researchers want to restrict data to that collected via
spontaneous or semi-spontaneous speech? Why not use monitored output?
One reason, of course, is to have a more direct comparison to child L1
acquisition and child/adult L2 acquisition. If the data sets aren’t similar,
then differences found between the two kinds of acquisition could be
traceable to research methods and not to actual differences. Another
reason is that researchers are more interested in what’s in the unconscious
or implicit system in the learner and not what that learner knows
consciously. This is an important issue and one that we will visit again
later in this book when we talk about the effects of instruction (Chapter 6).

The vast majority of research on morpheme orders has been conducted with
English. The reason for this is that once it was established that English morpheme
orders were unaffected by instruction, context, gender, the L1, and other variables
(see other chapters in this book), researchers concluded that there must be orders
in other languages as well. That is, we could posit the idea in the abstract of
morpheme orders because of the compelling evidence from English; we did not
actually need to see these other orders. Nonetheless, in spite of the general
acceptance that there is some kind of universality to morpheme orders, some
researchers have suggested that the orders were affected by the L1. In 2009, Zoe
Pei-su Luk and Yashuro Shirai published an article in which they reviewed
morpheme order development by learners with Japanese as an L1. What they
found was that for the noun-related morphemes, Japanese did seem to affect what
the English L2 learners were doing compared to learners who, for example, had
Spanish as a native language. They did not report the same for verb-related
morphemes, which to date seem to form a universal order regardless of L1. A
study published in 2016 by Akira Murakami and Theodora Alexopoulou also

60
suggested L1 influence on morpheme order development. However, their
conclusions must be taken with a grain of salt because of the numerous
methodological differences in their study. For example, they looked at only six
morphemes, did not group them by function (as did Andersen and VanPatten, for
example), used written data collected from exams (which of course is subject to
Krashen’s monitored output), and also did not follow the standard procedure of
looking at morpheme development over time. In short, it is difficult to take their
research as any serious challenge to universality of morpheme development.
At this point in time, it seems that sets of morphemes are acquired in a
particular order regardless of the factors already mentioned. It may be, however,
that noun-related morphemes (plural marker -s, possessive -s) are somehow
affected by first-language influence (at least in the case of Japanese as L1). What
is more, learners do not zip through morpheme orders and acquire them quickly;
the process is slow. In some cases for some learners, late-acquired morphemes
hardly appear in their communicative speech.
Not a lot of work has been done on morpheme orders in other languages.
However, research in various languages has looked at non-native patterns of
development for small sets of morphemes in Spanish, Swedish, German, and
other languages. For example, in Spanish as an L2, it has been documented that
learners acquire 3rd-person singular before they acquire other verb forms and that
singular verb forms are acquired before plural verb forms. For the reader
unfamiliar with Spanish, the language has unique person–number markings on
verbs in almost all tenses and moods. In the present tense, for example, the
morphemes are these:

1st singular: -o (or the variant –oy). Examples: como ‘I eat’; voy ‘I go’

2nd singular: -s. Examples: comes ‘you eat’; vas ‘you go’

3rd singular: nothing/“bare” verb. Examples: come ‘he/she eats’; va ‘he/she

goes’

1st plural: -mos. Examples: comemos ‘we eat’; vamos ‘we go’

61
 2nd plural: -is or –n depending on dialect. Examples: comeis/comen ‘you all
eat’; vais/van ‘you all go’

3rd plural: -n. Examples: comen ‘they eat’; van ‘they go’

Thus, learners of Spanish exhibit a morpheme order for person–number endings

resembling this:

nothing/bare verb (e.g., come)

-o, -s (e.g., como, comes)

-n, -mos, -is (e.g., comen, comemos, comeis).

62
 Developmental Sequences
In the previous section, we reviewed the ordered development of different parts of
language (e.g., noun-related morphemes, verb-related morphemes, articles,
copular verbs). Developmental sequences refers to the acquisition of a particular
item or structure. How does a learner get control over past-tense endings? How
does a learner get control over making yes/no questions (e.g., Does Bill write
about acquisition? Do you like to read about acquisition?)? How does a learner
get control over negation (e.g., Bill doesn’t write about physics. You won’t read
about acquisition.)? As we saw in Chapter 1, research on developmental
sequences was inspired by the research of Ursula Bellugi and others in
acquisition. Edward Klima and Ursula Bellugi published a paper in 1966 in which
they established three stages of development that all child L1 learners traverse
during the acquisition of negation in English.

Stage 1. The negator ‘no’ (sometimes ‘not’) is placed in front of or behind a

sentence or phrase. Examples: Not have coffee, No singing song, No I see
truck, Not Fraser read it.

Stage 2. The negator moves internal to the sentence. Unanalyzed chunks of

language such as can’t and don’t are used and may mean the same thing as
no not. Examples: There no squirrels, You can’t dance (meaning ‘You aren’t
dancing’).

Stage 3. Auxiliaries and modals are acquired and used correctly with not, in
both contracted and uncontracted forms. But children may still omit auxiliary
or copular be. Examples: I don’t have a book, Paul can’t have one, I not
crying.

In the late 1960s and early to mid 1970s, researchers such as Roar Ravem,
Henning Wode, and John Schumann (among others) began to look at negation in
L2 development. What emerged after a number of years was that the acquisition

63
of negation was stage-like just as it was in L1 acquisition. What is more, the
stages (1) were strikingly similar to those for L1 acquisition and (2) were similar
across learners of different L1s.

Stage 1. The negator ‘no’ is placed in front of a sentence or phrase.

Examples: No book, No is happy, No you pay it.

Stage 2. The negator moves internal to the sentence. Unanalyzed don’t

appears as a variant of no (i.e., the learner has not yet acquired do as an
auxiliary) Examples: He no can shoot good, They not working, I don’t have a
car, I don’t can play good, He don’t like job.

Stage 3. Modals appear and are used correctly with not, but largely in
contracted form. Examples: I can’t play, It wasn’t so big.

Stage 4. Do appears as an auxiliary and is used correctly with not in a variety

of persons and tenses. Examples: She didn’t believe me, He didn’t say it, He
doesn’t know anything.

Three points about developmental sequences are important to mention here.

The first is that learners do not seem able to skip stages. That is, a learner does not
go from Stage 1 to Stage 3 of negation without spending some time in Stage 2. To
be sure, learners may go through a stage very quickly and if we are not there to
document it, we might not see it. A second point about developmental sequences
is that they cannot be reordered. Stage 3 cannot become Stage 2 and vice versa.
Finally, although the stages are universal, learners show individual rates of
passing through each one, just like children in L1 acquisition are shown to have
individual rates of progressing through their development. Thus, one learner may
make it through the stages of acquisition in two years while another might take
four years. Others may never make it to the final stage after a decade.
Developmental sequences have been well documented for other structures in
English, such as wh-questions (what, when, where, why, how) and relative clauses
(e.g., The man who came …, The man who I saw …). Developmental sequences
have been established in other languages for various structures as well, such as

64
word order in German and copular verb acquisition in Spanish. In short, staged
development for a particular structure is a well-researched phenomenon. As in the
case of morpheme orders, learners take time to go through these stages – years, in
fact. Some learners do not make it to the last stages, as noted earlier for negation.

The L1 does not seem to affect the universality of staged development.

But the L1 may affect how long a learner stays in a particular stage of
development of a given structure.

Once again, something like staged development seems impervious to L1

influence. Whether you are a Chinese speaker, a Spanish speaker, or a Turkish
speaker, you will show evidence of these stages. You won’t skip stages and you
won’t reorder the stages. However, your L1 may influence how quickly you pass
through them or how long you spend in a stage. For example, Spanish is a
language with pre-verbal negation, as in Juan no estudia mucho ‘John doesn’t
study much.’ French is a language with post-verbal negation, as in Jean n’étudie
pas beaucoup ‘John doesn’t study much.’ (Pas is the negator in French; the ne/n’
is redundant and is often eliminated in spoken French, e.g., Jean étudie pas
beaucoup.) Both French speakers and Spanish speakers will show evidence of
Stages 1 and 2 of negation, but the Spanish speakers may linger longer in these
stages presumably because of the similarity to Spanish. In other words, L1
influence may be “triggered” in developmental sequences when a stage in
acquisition resembles something in the L1. This idea was advanced by Roger
Andersen back in the 1980s and, while not widely researched, has received
support in the empirical research on developmental sequences.
To be sure, there is some variation among learners but not in terms of the
stages themselves. For example, two different learners do not create two different
developmental sequences for negation. The four stages for negation outlined
above are universal. But the two learners may show some individual differences
in a given stage. For example, one learner might use no more as a negator than not

65
in Stages 1 and 2 whereas another learner might use them equally or may not use
not at all. What is more, stages aren’t neatly divided such that Stage 1 suddenly
disappears and Stage 2 suddenly appears. There is overlap among stages so that if
you collect data at just the right time, you might get a learner showing evidence of
Stages 2 and 3 at the same time, for example. This makes the learner’s internal
system look variable at a given point in time. And as in the case of morpheme
orders, research on developmental sequences is based largely on spontaneous or
semi-spontaneous speech used to communicate ideas.

66
Consider this …
Developmental sequences are said to be universal. This suggests that
everyone learning English negation begins with the no + word/phrase
stage. What does this suggest about the starting point for L2 acquisition?
Do learners begin with universal tendencies or predispositions? Is there
L1 transfer from the outset? Can you think of some other possibility? In
short, what do you think about this: The L1 may be the “starting point” for
L2 acquisition, but as soon as the data start coming in to the learner,
transfer may be “suppressed” until a time when a stage of acquisition
looks something like a structure in the L1. We will review these and
related ideas in Chapter 4.

67
 U-Shaped Development
Another phenomenon documented in the acquisition of particular structures over
time is what is called U-shaped development. The name derives from the shape
of a graph line if you plotted out what the development looks like over time. An
ideal situation is depicted in Figure 2.1.

Figure 2.1 Classic U-shaped development

What Figure 2.1 represents is that at Time 1, the first time the researcher
observed the participant, the participant seemed to do rather well on the structure,
with an accuracy of about 95 percent. Then at Time 2, the participant seems to
have got worse in performing with the structure (35 percent accuracy), finally
bottoming out at Time 3 and Time 4 (5 percent accuracy). Then at Time 5 we see
a marked increase in accuracy (45 percent), with the participant performing with
almost perfect accuracy in the end (Time 6). When we step back and look at the
graph line, it resembles a “U.” Hence the name of the phenomenon.
On what kinds of structures do we see a developmental trend like that in
Figure 2.1? The classic case is with the acquisition of irregular past tense – and
this seems to be true for L1 and L2 learners. At first, the learner performs rather
well with a handful of highly frequent irregulars, such as was, came, went, and
ate. Less frequent irregulars like drove, swam, and bought haven’t even been

68
picked up by the learner yet. Subsequently, the learner begins to develop regular
past tense, which is formed by adding some form of the /t/ sound to the ends of
verbs (e.g., watched ends in a [t] sound, warmed ends in a [d] sound, and wanted
ends in a [ɪd] sound). As regulars begin to creep into the learner’s system, the
irregulars get “regularized.” So the learner stops saying went and might say goed
or even wented. Accuracy in production of these irregulars may drop to zero.
After a period of time, the irregulars begin to creep back in, especially the highly
frequent ones, until the learner is once again saying went, ate, and came with 100
percent accuracy. When the accuracy of these irregular verb forms in the past is
plotted over time on a graph, it looks like the one in the graph.
Another typical U-shaped pattern emerges with plural marking on nouns.
Learners may start off correctly with various irregular plurals such as feet and
sheep. Later, as they acquire regular plurals such as dogs, watches, and books,
they tend to regularize the irregulars, producing variations such as foots and feets,
sheeps, and others. Eventually, these drop out and the irregular forms the learner
started with emerge once again.
U-shaped development has been more studied in L1 acquisition than in L2
acquisition, yet there are a few studies that have documented U-shaped patterns.
Again, once the pattern showed up in early L2 research and was found not to be
influenced by L1 (except see below with markedness), scholars took such
behavior to be a given. Related to U-shaped behavior is a term called
backsliding. Backsliding refers to learners seemingly going back to an earlier
stage of learning a structure and thus performing in a “worse” or less accurate
way. So if a learner is clearly at Stage 3 of negation and then Stage 2-type
structures begin to dominate, we might say that the learner is backsliding.
Backsliding does not appear to be a predominant feature of L2 acquisition but it
has been documented for some learners.

69
 Markedness Relationships
Languages have what linguists called unmarked and marked structures or
lesser-marked and more marked structures. Markedness refers, in a general
sense, to how typical and frequent one version of a structure is compared to other
versions. Markedness often entails what is called an implicational hierarchy. In an
implicational hierarchy, if a language has Z it also has A, but if it has A it does
not necessarily mean it has Z. One of the classic examples of markedness and an
implicational hierarchy is found in relative clauses.
Relative clauses are those dependent clauses that modify a noun. So, for
example, in The man who came to dinner …, who came to dinner is a relative
clause that modifies the man. There are different types of relative clauses
depending on the grammatical role they fulfill. Here are all of the relative clauses
in English.

Subject. The relative clause modifies the subject of a sentence. In the man
who came to dinner was Fred, the relative clause who came to dinner
modifies the man, which is the subject of the main clause The man was Fred.

Object. The relative clause is the object of a verb. In I know the man who you
invited to dinner, the man is the direct object of the verb invited.

Indirect object. The relative clause is the indirect object of a verb. In I know
the man who you served dinner to, the man is the indirect object of the
phrase served dinner to.

Oblique. The relative clause is the object of a preposition. In I know the man
who you talked about, the man is the object of the preposition about.

Genitive. The relative clause is the possessor of something. In I know the

man whose dinner you burned, the man is the possessor of the dinner.

Object of comparison. The relative clause is the object of a comparison

involving than. In I know the man who you are a better cook than, the man is

70
 the object of the comparison better cook than.

In English, and other languages that have similar relative clauses – not all do
– subject relative clauses are by far the more frequent. They are considered
unmarked or the least marked of the relative clauses. Object of comparison
clauses, on the other hand, are infrequent and are said to be marked or more
marked and less typical. What is more, these relative clauses exist in an
implicational hierarchy. Each relative clause lower on the list implies the clauses
above it but not below it. So if a language has oblique relative clauses, it also has
subject, object, and indirect object relative clauses. But we can’t make any
predictions about clauses below oblique; the language may or may not have them.
Looking at the two extremes, if a language has object of comparison clauses, then
it has all of the clauses above it. If a language has subject relative clauses, we
cannot infer anything about other clauses. We’d have to see if the language has
any of those.
Markedness plays a complex role in L2 acquisition. First, as in L1
acquisition, unmarked or lesser-marked structures tend to be acquired before
marked or more marked structures. In the case of relative clauses, we would
expect learners to comprehend and produce subject relative clauses (way) before
they could comprehend and produce object of comparison clauses. We might also
expect learners to make fewer non-nativelike versions of subject relative clauses
before they would do the same with something like object of comparison clauses.
What does this mean?
Some languages have what are called resumptive pronouns within relative
clauses. Resumptive pronouns duplicate the grammatical function of the relative
clause itself. For example, for an object relative clause, the resumptive pronoun
him is inserted into the clause: I know the man who you invited him to dinner. In
this case, the object pronoun him duplicates the function of the man, which is the
original object of the verb invited (i.e., I know the man. You invited him to
dinner.). For object of comparisons, the sentence with a resumptive pronoun
would be I know the man who you are a better cook than him (i.e., I know the

71
man. You are a better cook than him.). Languages like Arabic require the use of
such pronouns in relative clauses. Languages like Spanish and English do not; in
fact, they aren’t allowed in these languages. Yet both Spanish and Arabic learners
are known to produce resumptive pronouns when first working with relative
clauses or accept them on what are called “grammaticality judgment tests” (tests
in which you give a learner sentences, some grammatical and some
ungrammatical, and ask the learner to indicate whether a sentence sounds “good”
or “bad”). However, resumptive pronouns should “disappear” earlier from learner
language with the unmarked or lesser-marked relative clauses than before the
marked or more marked relative clauses. And if learners are not making non-
native sentences with resumptive pronouns in oblique relative clauses, they are
probably not making them in sentences above these clauses on the implicational
hierarchy we reviewed earlier.
Markedness also plays a role in L1 influence. There was considerable
research conducted in the 1970s and 1980s in which it was shown that marked
structures are less likely to be “transferred” into L2 learning than unmarked
structures. Transfer is a somewhat ill-defined term in L2 acquisition but can refer
to any situation in which there appears to be L1 influence on L2 acquisition or
performance. For example, if you hear someone speaking English and detect a
“foreign accent,” you are likely noticing some kind of L1 transfer of that person’s
L1 sound system into the L2 sound system. How transfer actually happens is not
entirely clear, which in part is why the concept is a bit ill-defined. Nonetheless,
the idea of transfer (or L1 influence) is widely accepted as a phenomenon in L2
research circles.

Markedness plays a role in constraining language transfer. Learners seem

less likely to transfer marked structures from their L1 into the L2 learning
process than they are to transfer unmarked ones.

72
 Returning to the issue of markedness and transfer, here’s one example of
how markedness constrains transfer (meaning how it keeps transfer from doing
whatever it wants to do). The example comes from phonology (the study of sound
systems). Languages may have what is called a voiced/voiceless contrast for
consonants. Voicing occurs when the vocal cords vibrate. The cords do not
vibrate when a consonant is not voiced. You can try this yourself by placing your
fingers gently on your throat and making the /s/ sound several times and then
making the /z/ sound several times. You should feel a vibration when
pronouncing /z/ but not /s/. Other voiced/voiceless pairs of consonants include /d/
and /t/, /b/ and /p/, and /g/ and /k/. In terms of markedness, there is an
implicational hierarchy for voiced/voiceless contrasts of consonants. The
unmarked position is word initial, as in dot vs. tot. The more marked position is in
word-final or syllable-final position, as in Tod vs. tot. Some languages have
voicing contrasts in all positions, such as English. Others do not. What has been
observed in L2 research is that learners are less inclined to transfer the
voiced/voiceless contrast in final position into the L2, whereas they may readily
transfer the contrast in initial position. In addition, learners whose L1 has the
voiced/voiceless contrast in all word and syllable positions seem to have no
difficulty acquiring these contrasts in their L2. Learners whose L1 does not, tend
to have a harder time acquiring marked voicing contrasts. Thus, markedness
constrains or affects the role of L1 influence.

73
 Summary So Far
In this chapter we have reviewed several key findings about how L2s grow in the
mind of the typical, unimpaired learner. We will summarize them here.

Language development is ordered. It is ordered in that various forms and

structures are acquired over time in orders that seem universal to all
learners of a given language. We saw this in three ways.

◦ Morpheme orders. Morphemes, the smallest units of meaning, are

acquired in a particular order over time. So, for example, in English, -
ing on verbs is always acquired before 3rd-person -s, and past-tense
marking is acquired somewhere in between.

◦ Developmental sequences. In the acquisition of a given structure,

learners will show evidence of staged development. We demonstrated
this with the acquisition of negation in English, describing the four
stages established by researchers.

◦ U-shaped behavior. In the acquisition of a given structure, learners

sometimes show evidence of starting out well with something (normally
with a few target samples of the structure or form), then their accuracy
“plummets.” Over time, they regain accuracy on the form or structure.

Language development is constrained. We saw how the construct of

markedness affects how language is acquired. In general, we noted the
following:

◦ Learners acquire unmarked structures before marked or more marked

structures.

◦ Learners are more likely to exhibit non-nativelike performance with

more marked structures than unmarked structures.

74
 ◦ Learners are less inclined to transfer marked structures from the L1 to
the L2.

There is some individual variation in ordered development, but it does not

affect the overall universal patterns of ordered development we see in
learners.

The general idea is that language acquisition is not completely idiosyncratic

such that each and every learner does something completely different from other
learners. There is something universal about language acquisition. That is, when it
comes to the formal features of language (i.e., the purely grammatical aspects of
language), learners seem to be guided and constrained by “something.” That
something is, most likely, not external to the learner; it is internal to the learner. In
other words, something inside the learner’s mind, some kind of mechanism or
device (possibly mechanisms or devices) is pushing learners along one path as
opposed to another. Something internal to the learner is causing -ing to be
acquired before past tense in English. Something internal to the learner is causing
Stage 2 of negation before Stages 3 and 4. To go back to our example of
resumptive pronouns for a minute, it’s clear that resumptive pronouns in relative
clauses are due to something other than the languages in question. English doesn’t
have them, so learners couldn’t have been taught them or couldn’t have heard
them. Arabic-speaking learners of English might transfer resumptive pronouns
into English, but Spanish-speaking won’t do this because Spanish doesn’t have
resumptive pronouns either. That both groups of learners produce or accept
relative clauses with resumptive pronouns suggests that something independent of
the L1 and independent of the language they are exposed to drives language
acquisition. Something internal to the learner is causing the appearance of
resumptive pronouns in relative clauses or for subject relative clauses to appear
before oblique relative clauses. We will return to the issue of constraints in the
next section.
In addition to the above, it is worth underscoring two related aspects of
acquisition. The first is that, as the morpheme studies and developmental

75
sequence studies show, acquisition is piecemeal. Many people think, for example,
that a person first acquires the present tense, then the past tense, then the future
tense, and so on. In fact, this is codified in a lot of language-teaching materials –
especially for modern languages such as Spanish, French, Russian, and Arabic.
However, as we saw, this is not the case. At a given point in time, a learner may
be “sketching in” pieces of language related to the present tense while also
“sketching in” pieces of language related to the past tense. The second point
worth underscoring is that acquisition is much slower than most people assume or
admit. As pointed out above, it can take years to pass through the developmental
sequences described earlier. And because of individual differences in rate of
acquisition, for some learners it can take many years. In short, when we look at
such things as morpheme orders, developmental sequences, U-shaped behavior,
and so on, we are not talking about learning a language in a month or two months.
In spite of the advertisements of some commercial “teach yourself” materials, it is
not possible to learn Spanish in just 30 days. At least not in terms of putting
together an underlying system.
A final observation about development is the role of L1 influence or transfer.
It is widely assumed among teachers and lay people that the L1 is the source of all
learning problems for the L2 acquirer. These ideas date way back and were
codified in behaviorist theory, which we reviewed in Chapter 1. Although there is
no doubt that the L1 does influence L2 acquisition in some way, it is also clear
that in many respects L1 influence is constrained. In some cases, we see evidence
of language transfer and in some cases we do not. In some cases it affects one
piece of the puzzle but not another piece of the puzzle.

76
Consider this …
Prior to the 1970s there was no empirical evidence to speak of on how L2
learners’ linguistic systems grew over time. The revelation of ordered
development, constrained L1 transfer, and other facts about L2 acquisition
struck a blow to behaviorism and the idea that learning language was
learning habits (see Chapter 1). What you must consider at this point is
that the research discussed in this chapter has been collected on classroom
learners, non-classroom learners, and learners who are in mixed
environments (i.e., they may be in a classroom learning for some part of
the day but then they leave the classroom and are in a community in
which the L2 is the dominant or only language). It seems, then, that
context is not a major factor affecting how language acquisition unfolds
over time. This is not to say it has no impact, but context does not seem to
play a major or determining role in ordered development. This has
consequences or implications for how people conceptualize language
instruction. What do you think these consequences and implications might
be?

Exemplary Study

Ellis, R. (1989). Are classroom and naturalistic acquisition the same?

Studies in Second Language Acquisition, 11, 305–328.

77
 Participants
39 beginning students of German as an L2

L1 = English, Spanish, Arabic, French, Mauritian Creole

location: London

the students were taught by different instructors using different

techniques and approaches

some students were fulfilling requirements, while others intended to

pursue more study of German after this initial course

78
 Major Target Item
German word order

previous research on immigrant non-classroom learners in Germany

revealed the following developmental order:

1. subject-verb-object: die kinder spielen mim ball ‘the children

are playing with the ball’

2. adverb preposing: da kinder spielen ‘there the children play’

(the adverb ‘there’ is placed at the beginning of the sentence
before everything else in this example)

3. verb separation: alle kinder muss die pause machen ‘all the
children must the pause make’ (modals and helping verbs, muss,
are separated from main verbs, machen)

4. inversion: dann hat sie wieder die knocht gebringt ‘then has she
again the bone brought’ (this begins to look like actual German
word order in which the inflected verb, hat, appears in the second
position of the sentence no matter what is in first position – in this
case an adverb; thus the subject appears after the verb, and the
main verb, in this case a past participle, appears at the end of the
sentence)

5. subordinate clause verb end: er sagte dass er nach hause kommt

‘he said that he to home came’ (this is standard German word
order; note that unlike simple or main clauses, the inflected verb
comes at the end of the subordinate clause)

79
 Materials and Procedure
participants were put into pairs with cut-up pictures, not knowing what
the other person was looking at; together they had to talk without
showing the picture in order to work out what was in the pictures and
what story the pictures told

at the end, one of the participants was required to tell the story (i.e.,
produce something like a short monologue)

each pair of participants performed three picture tasks at two intervals:

the first interval was after 11 weeks and the second interval was after 22
weeks

all pair work was recorded and subsequently transcribed

an analysis of the self-corrections made revealed minimal self-

corrections, suggesting the learners were engaged with the task
communicatively and not to produce correct sentences in German
(when they did correct, it was not for word order or sentence structure)

80
 Results
Only 17 learners produced enough sentences involving the different
components involved in German word order.

An analysis of production revealed an implicational order that looked

like the naturalistic order described above.

Thus classroom learners of German first produced adverb preposing

followed by verb separation followed by inversion.

81
 Conclusion
Both classroom and non-classroom learners seem to follow similar
developmental paths and are constrained by both the nature of language
and learner-internal mechanisms (see below).

82
 Knowing More Than What You Are Exposed To
Another interesting but often overlooked feature of L2 acquisition is that, just like
L1 learners, people come to know more about the L2 than what they are exposed
to. What is more, they come to know things about language that are impossible to
know given the language they are exposed to. Let’s look at a few examples to see
what this means.
In English, we can contract such things as want to to wanna, going to to
gonna, and I have to I’ve, as in the following examples.

(1) Who do you want to invite to the party? → Who do you wanna invite to
the party?

(2) I’m going to try a new look. → I’m gonna try a new look.

(3) I have decided that doesn’t work. → I’ve decided that doesn’t work.

At the same time, the following contracted sentences are impossible in English.
(Remember that an asterisk means a sentence is not possible.)

(4) Who do you want to tell Alex the news? → *Who do you wanna tell
Alex the news?

(5) I’m going to the store now. → *I’m gonna the store now.

(6) Should I have said that? → *Should I’ve said that?

Every English speaker knows that (4)–(6) are bad sentences because the
contractions “sound odd.” But nowhere are they taught this. What is more, these
are not even mistakes that children would ever make when learning English as an
L1.
Now let’s look at an example from Spanish. Spanish allows null subjects.
Null subjects are “hidden” subjects, meaning a sentence is perfectly grammatical
without a subject or subject pronoun in it. Here are some examples.

(7) Escribo un libro/Yo escribo un libro. ‘I’m writing a book.’

83
 (8) ¿Qué quería? /¿Qué quería él? ‘What did he want?’

What is more, null subjects are required in certain instances, such as time
expressions, weather expressions, “existential” statements, and when a subject is
unknown.

(9) Llueve. ‘It’s raining’ (*Ello llueve is ungrammatical in almost all

dialects.)

(10) Es la una. ‘It’s one o’clock.’ (*Ello es la una is ungrammatical in

almost all dialects.)

(11) Hay café en la cocina. ‘There’s coffee in the kitchen.’ (*Allí hay café en
la cocina is ungrammatical in all dialects.)

(12) ¡Me robaron! ‘They robbed me!’ (But I don’t know who the ‘they’ are;
this is more like ‘I was robbed.’ Here Ellos me robaron would be incorrect.)

Every speaker of Spanish knows when null subjects are possible and when they
are required. Of particular interest is (12), that people come to know that overt-
subject pronouns in Spanish must have antecedents. That is, ellos in (12) is only
possible if the speaker knows exactly who the robbers are (e.g., John and Paul, the
boys down the street, two men the speaker met at a bar). Again, children do not
show evidence of making mistakes with sentences in which null subjects are
required.
In short, people come to know not just what is possible in a given language
but also what is impossible. No one tells them what is impossible and, as
indicated, people just don’t make mistakes with such impossible structures during
L1 acquisition.
The same is true in L2 acquisition: Learners come to know more than what
they are exposed to or taught. This is well known in the research. L2 learners of
English come to know when contractions are possible, and they know this not
because they are taught. And like children learning English as an L1, L2 learners
don’t make such mistakes. The same is true for the examples in Spanish. It is
documented that L2 learners come to know when null subjects are required (i.e.,

84
when “overt” subject pronouns are not allowed) just like L1 learners do. Again,
no one teaches an L2 learner of Spanish how null and overt subjects work, and
certainly no one teaches them that overt pronouns must have an antecedent.
This particular situation in acquisition has been termed the poverty of the
stimulus. This term means that the stimulus (language) learners are exposed to
does not contain any obvious data in it about the impossibility of a particular
structure. What is important to keep in mind is that learners are only exposed to
possible sentences and structures. Somehow, something inside their minds must
“infer” what is impossible. And just because something hasn’t been encountered
in the language the learner has been exposed to doesn’t mean it’s impossible.
Perhaps the learner just hasn’t encountered it yet and it is possible.
In addition to impossibility, learners may come to know that certain things
are possible even when they haven’t encountered them previously. A classic
example is ambiguity, as in John said he broke his leg yesterday. This sentence
can be understood in two ways: John made the announcement yesterday
(Yesterday John said he broke his leg) or that the leg breaking occurred yesterday
(Yesterday John broke his leg). Hidden behind this ambiguity are two distinct
syntactic structures for the sentence, and these two distinct structures permit two
different interpretations of the sentence. The question is, how do people know
this? They aren’t taught this and there is nothing in the language they encounter
that would tell them there are two interpretations. They just know.

Learners obtain knowledge about grammaticality and ungrammaticality or

possibility and impossibility that goes beyond just the grammatical and
possible structures they hear or see in the language that surrounds them.

In short, a significant part of L2 development, then, is projecting beyond the

data learners are exposed to (we will see another example in the section below on
constraints). They obtain knowledge about grammaticality/ungrammaticality or
possibility/impossibility that goes beyond just the grammatical and possible

85
structures they hear or see in the language that surrounds them. No one teaches
them. They just seem to know.

86
Consider this …
What we look at in L2 acquisition and what we see as “patterns” in what
learners do may depend on the kind of data we get. For example, research
on acquisition orders and developmental sequences requires spontaneous
or semi-spontaneous oral data. The research is interested in what learners
do. Research on the poverty of the stimulus, however, cannot rely on oral
data – at least not on oral data alone. This research is interested in what
learners unconsciously know. That is, to find out what learners believe are
impossible sentences in a language, their underlying knowledge has to be
probed. This is because what learners produce would, presumably, be
what is possible at a given point in time in their developing systems;
learners don’t produce what is impossible in their linguistic systems. Why
can’t the researcher simply infer that if the learner doesn’t produce
something while speaking, that thing must be impossible in the learner’s
system?

87
 Constraints on Acquisition
Once again allowing for certain (limited) individual variation, we saw earlier that
two adjectives that describe acquisition are ordered and constrained. It should be
clear what we mean by ordered. What do we mean by constrained and constrained
by what?
The term constrained means that a process or action is compelled or forced
along a particular path. In L2 research, we use the term constrained to mean that
development over time is pushed along a particular route or path that the learner
cannot consciously control; nor can it be controlled by exterior forces such as
instruction. It is now accepted that language acquisition is constrained; however,
it is not always clear what the source of the constraints is.
One kind of constraint comes from markedness, as we saw earlier. Learners’
linguistic systems are constrained by markedness in that more marked structures
are more difficult to acquire than less marked structures. In addition, learners are
less likely to transfer marked structures from the L1 compared to unmarked
structures during L2 acquisition. We can say, then, that markedness guides or
constrains language development in particular ways. Acquisition is also
constrained by universal aspects of language. In generative theory, for example,
language acquisition is said to be constrained by Universal Grammar. Universal
Grammar or UG is a set of universal features and principles common to all
languages. No language may fall outside the scope of the features contained
within UG and no language may violate the principles of UG.
Let’s begin with a basic notion within UG. Sentences are said to consist of
hidden structure involving phrases. Phrases are of two types: lexical phrases and
functional phrases. Lexical phrases are the types of phrases many of us are
familiar with, such as verb phrases (e.g., eats veggies) and prepositional phrases
(e.g., in the kitchen). Functional phrases are such things as tense phrases and
complementizer phrases. Tense phrases are hidden parts of sentences that contain

88
information about past, present, or future, for example. Complementizer phrases
are hidden phrases that contain information about what type of sentence or clause
we are dealing with, such as declarative versus question versus embedded clause.
We cannot easily illustrate a functional phrase here because functional phrases
often do not contain lexical material, although lexical material can be inserted into
them, as we will see in a minute. UG says that all languages have such phrases.
What is more, UG says that all such phrases consist of a head (e.g., in a verb
phrase, the verb is the head of the verb phrase; in a prepositional phrase, the
preposition is the head of the phrase) and a complement (e.g., in a verb phrase, the
object of the verb is a complement; in a prepositional phrase, a noun or noun
phrase tends to be the complement of a preposition). So the verb phrase eats
veggies consists of the head eats and the complement veggies. The prepositional
phrase in the kitchen consists of the head in and the complement the kitchen.
Again, all languages must have such underlying structure. What is more, a
consequence of phrase structure is that there are only two possible orders: head +
complement (like English) and complement + head (like Japanese). Thus, whereas
English says eats veggies, Japanese says something like veggies eats. And UG
likes things to be uniform such that all phrases, lexical and functional, in a
language follow the same order. So all English phrases are head + complement
and all Japanese phrases are complement + head.
Now let’s look at a functional phrase. To form a yes/no question in English,
we insert a form of do as in Do you eat veggies? Do is the head of the
complementizer phrase, and note that the standard head + complement order is
followed: do + you eat veggies. Japanese uses the particle ka to indicate yes/no
questions and, following its standard phrase structure, the equivalent English
question would be you veggies eat ka? That is, the complement precedes the head:
you veggies eat + ka. Research has shown that both child L1 learners of Japanese
and adult English speakers of Japanese as an L2 very quickly (after the most
minimal exposure) determine that Japanese has complement + head word order.
Children demonstrate this from the first time they put two words together, and

89
adult L2 learners know this after about 30 minutes of exposure to basic subject-
object-verb sentences. They do not need to hear and see every single instance of a
phrase type to know that Japanese is complement + head. Mere exposure to
simple sentences with verb phrases “tells” the internal mechanisms that in
Japanese the head of a phrase follows the complement. Bill VanPatten and Megan
Smith demonstrated this in a study published in 2015. After exposure to just 100
simple sentences in Japanese containing verb phrases and prepositional phrases
(which, technically, in Japanese we would call postpositional phrases), 9 out of
the 49 participants demonstrated knowledge of Japanese word order in all the
domains they were tested on, including some they were not exposed to (i.e.,
yes/no questions, embedded clauses). Another 20 participants demonstrated
knowledge of Japanese word order in some domains they were not exposed to
(i.e., yes/no questions). Something was pushing these learners to “unify” phrase
structure and assume that all underlying phrases in Japanese were complement +
head – even for phrase types they had not been exposed to – another example of
the poverty of the stimulus; learners had come to know something about the
language they were learning when they had not been exposed to it yet.
In another approach to universals referred to as the typological approach,
languages are viewed as having either categorical universals or strong tendencies
regarding structure and formal features of language. The name most associated
with this framework is Joseph H. Greenberg. In 1963 he published a set of
linguistic universals, one of which is Universal 41: If in a language the verb
follows both the nominal subject and nominal object as the dominant order, the
language almost always has a case system. In other words, if a language has
dominant subject-object-verb or SOV word order, it is also likely to have case
endings indicating subjects and objects. The same is not true for subject-verb-
object or SVO languages; these languages tend not to have case endings. Indeed
this is borne out in the literature. Matthew Dryer has surveyed over 500
languages. He found that of the 253 SOV languages he studied, 72 percent case-
marked nouns in some way. For the 190 SVO languages he studied, only 14

90
percent case-marked their nouns. This is a very strong tendency. Bill VanPatten
and Megan Smith (in a different study) show how this kind of universal guides
and constrains language acquisition in adults. They studied 50 participants with
no prior knowledge of or exposure to case marking or SOV word order. They
divided them into two groups to learn Latin from scratch (Latin is an SOV
language with case marking). One group was exposed to only SOV sentences in
Latin while the other was exposed to only SVO sentences (all sentences were the
same for each group except the word order). What they found after exposure to
just 100 simple sentences was that the SOV group demonstrated a very strong
sensitivity to case-marking violations (85 percent) while the SVO was at chance
(around 55 percent). Thus, the consistent exposure to SOV somehow guided the
learners’ internal mechanisms to search for case marking during exposure much
better than for those in the SVO exposure. Note that in this experiment the
participants were not taught anything about Latin or case marking. They simply
heard sentences while looking at pictures. Taken together, the VanPatten and
Smith studies with beginning learners indicate that something about the way the
mind processes language constrains acquisition and guides learners’ internal
systems in the development of the implicit system. This process is independent of
instruction, explicit knowledge of the structure in question, the L1, and the L2
(Japanese and Latin mark case differently).
Other theories impose other kinds of constraints on acquisition. In usage-
based models of acquisition, acquisition is constrained by a number of factors
working together, two of which are frequency in the input (input is discussed in
the next chapter) and blocking. In this framework, the human mind contains what
we might call a “frequency tabulator.” This tabulator keeps track of the relative
frequency of things that learners encounter in the language that surrounds them.
More frequent things build up faster than others, putting a kind of constraint on
when things emerge in the learner’s linguistic system; that is, less frequent things
don’t make it into the linguistic system as fast as more frequent things. This is as
true of vocabulary as it is for grammatical structure and formal features of

91
language. To be sure, other factors may combine to attenuate the role of
frequency. One of those factors is salience. Salience is not a well-defined term in
the field of L2 research but it is often understood to mean something like “the
degree to which something stands out compared to something else.” So in a herd
of elephants, a lone zebra would be salient. It would stick out. In language,
salience is tied to such things as where something occurs in a sentence (e.g., the
beginnings of sentences seem to be more salient positions than the middle parts of
sentences). Syllables with strong stress are more salient than syllables with weak
stress. There are other attributes to salience, and the idea is that salience may
serve as an intervening factor to attenuate frequency (e.g., non-salient items that
are frequent may take longer to acquire than salient frequent items). In spite of
these caveats, the “frequency tabulator” in the mind is part of the mechanism that
guides and constrains acquisition.
Blocking refers to the way that previous knowledge affects how incoming
data are handled. The learning mechanism might not let something into the
linguistic system if something previously learned “seems to already do the job.”
The previously learned knowledge could be something from the L2 or something
from the L1. As one example, Nick Ellis and Nuria Sagarra, in a 2010 publication,
showed how L2 learners of Spanish (with English as L1) relied on previously
learned adverbs of time (e.g., yesterday, last week) to determine if something was
in the past. This reliance blocked their processing of verb endings that also
indicated past tense. So a previously learned “cue” (in this case adverbs of time)
caused another cue (in this case verb endings) to be ignored. They also showed
that Chinese L1 speakers showed difficulty with past-tense endings in English as
an L2, presumably because their previous knowledge of Chinese “blocked” their
processing of English past-tense verb endings; Chinese has no verb endings to
indicate any kind of tense.

Various approaches indicate that language acquisition is constrained by

universals, L1 knowledge or already acquired L2 knowledge, frequency,

92
 and other factors.

In short, various approaches clearly indicate that language acquisition is

constrained by universals, markedness, L1 knowledge or already acquired L2
knowledge (blocking), frequency, and other factors. By and large, learners seem
to be in control of the acquisition of formal properties of language acquisition
(e.g., morphemes, syntax), albeit unconsciously. Internal constraints guide and
nudge acquisition down particular paths and not others.

93
 Back to Variation for a Moment
Some scholars have looked at variation in L2 development. By variation we
don’t mean differences between learners but what learners might do at a given
stage of development. For example, at Stage 2 of negation in English (see above)
the learner is said to be at the stage of subject + negator + phrase, as in I no like
and He don’t drink beer. Variation occurs in this stage for a variety of reasons.
First, stages aren’t so neat that the learner leaves one stage and is 100 percent in
the next stage with the following stage being 100 percent absent. Instead, the
learner is largely in Stage 2 with some remnants of Stage 1 and some aspects of
Stage 3 beginning to creep in. So even though we categorize the learner as being
at Stage 2, that learner’s production is not entirely made up of Stage 2-like
structures. In addition, at Stage 2 the learner may freely interchange no and don’t
as negators, each meaning the exact same thing, with the exact same grammatical
properties. But as the learner switches between He no like and He don’t like, it
seems there is variation in the choice of the negator.
Likewise in the acquisition of morphemes we see variation. A learner in the
stages of acquiring the past tense may sometimes use the past-tense ending
(Yesterday he called me) and may sometimes leave it out (Yesterday he call me).
At the same time, that learner may sometimes produce nativelike irregular past-
tense verbs (went) and sometimes non-nativelike irregulars (wented).
The source of variation is not clear. What is to be noted is that variation as
described here is well documented in child L1 acquisition as well. So it seems that
as learners move through a stage of acquisition, regardless of context, they will
exhibit variation in control over something in their spontaneous and
communicative production.
To be sure, there is some variation across learners as well. Let’s go back to
Stage 2 of negation in English for a moment. Some learners may use almost
exclusively no. Some may use almost exclusively don’t. Others may mix. Again

94
the source of this variation is not clear, and we will discuss individual differences
in Chapter 8.
Does variation undermine the ordered development and stage-like
development described in this chapter? Not at all. Learners seem to vary within a
range of possibilities; they do not seem to create completely idiosyncratic
linguistic systems that do not resemble each other in some way. We can look at a
metaphor using snowflakes. It is said that each snowflake is different. This is true
if we stand back and look from afar. But a close inspection of snowflakes reveals
the following. First, all snowflakes are six sided. There are no eight-sided, ten-
sided, four-sided or X-sided snowflakes. And there is a reason for this. Digging a
little deeper, we see that the crystalline structure underlying snowflake formation
follows a universal structure. These universal structures may adhere to each other
to give each snowflake a unique appearance, but underlyingly each snowflake is
alike. In language acquisition, linguists look beyond the immediate surface
manifestations of language to discover the underlying structure of a learner’s
linguistic system at a given time. For this reason, then, the learner who says He no
like and the learner who says He don’t like seem to be doing different things but
their underlying grammars are the same: subject + negator + phrase.

What about Social Factors?

Social factors such as identity and who we want to talk with are powerful
factors in communication. As such, they influence the kind of language
we are exposed to and thus the linguistic data we are exposed to. This in
turn influences the way we talk or behave with language. And to be sure,
if social factors affect who we talk to and how much, this affects the
language we are exposed to and this may accelerate acquisition or it may
negatively affect the rate. Imagine, for example, the learner abroad who
fiercely clings to his native identity compared to the learner who is open
and almost adopts a new identity in the second culture. Such situations can

95
and do affect learning overall. However, they do not seem to affect such
things as ordered development and internal constraints on language.
Social factors may influence some of the variation we see in learner
performance. Imagine the learner who, consciously or unconsciously,
perceives that social context X is different from social context Y and
adjusts his or her communication accordingly. It is not clear from the
research, however, that such adjustments affect the grammatical
components of language in the implicit system. Clearly social factors
affect tone, word choice, possibly sentence length, and other “stylistic”
aspects of language – all very important aspects of communication. But
the focus here is on the actual linguistic (formal) system. It is not clear at
all whether social factors affect the variation between He no like and He
don’t like. That is, it is not clear whether the use of one or the other of
these negative sentences during Stage 2 of negation development indicates
the influence of social factors such as perceptions of social context.

96
 Recap
In the present chapter, we touched on the following major ideas and issues.

The acquisition of formal features of language (grammatical aspects of

language) is ordered.

Learners show evidence of acquiring morphemes in particular orders over

time.

The acquisition of a particular structure (e.g., negation) shows stage-like

development.

U-shaped learning (i.e., starting well with something, getting worse, and
then getting better again) has been found for some features of language.

The acquisition of formal features of language is constrained. Such things

as markedness, Universal Grammar, and perhaps general learning
mechanisms all work to push and guide acquisition in particular
directions.

The role of the L1 is also constrained.

Variation is present in learners. That is, when in a given stage of the

acquisition of a particular structure, they may show variation in their
production of that structure. And learners may vary one from the other in
the degree and type of variation within a stage. However, the ordered and
constrained development of the formal features of language is not affected
by variation.

97
 References and Readings

Andersen, R. A. (1983). Transfer to somewhere. In S. Gass & L. Selinker (Eds.),

Language transfer in language learning (pp. 177–201). Rowley, MA: Newbury
House.

Dryer, M. S. (2002). Case distinctions, rich verb agreement, and word order type.
Theoretical Linguistics, 28, 151–157.

Dulay, H. & Burt, M. (1974). Natural sequences in child second language

acquisition. Language Learning, 24, 37–53.

Eckman, F. (1977). Markedness and the contrastive analysis hypothesis.

Language Learning, 27, 315–330.

Eckman, F. (2004). Universals, innateness and explanation in second language

acquisition. Studies in Language, 28, 682–703.

Ellis, N. & Sagarra, N. (2010). The bounds of second language acquisition:

Blocking and learned attention. Studies in Second Language Acquisition, 32,
553–580.

Ellis, N. & Wulff, S. (2015). Usage-based approaches to SLA. In B. VanPatten &

J. Williams (Eds.), Theories in second language acquisition (pp. 75–93). New
York: Routledge.

Greenberg, J. H. (1963). Some universals of grammar with particular reference to

the order of meaningful elements. In J. H. Greenberg (Ed.), Universals of
language (pp. 73–113). Cambridge, MA: MIT Press.

Hatch, E. M. (Ed.) (1978). Second language acquisition: A book of readings (pp.

207–230). Rowley, MA: Newbury House.

98
Kellerman, E. (1985). If at first you succeed … In S. M. Gass & C. Madden
(Eds.), Input in second language acquisition (pp. 345–353). Rowley, MA:
Newbury House.

Krashen, S. D. (1982). Principles and practice in second language acquisition.

New York: Pergamon Press.

Krashen, S. D., Houck, N., Giunchi, P., Birnbaum, R., Butler, J., & Srei, J.
(1977). Difficulty order for grammatical morphemes for adult second language
performers using free speech. TESOL Quarterly, 11, 338–341.

Luk, Z. P. & Shirai, Y. (2009). Is the acquisition of grammatical morphemes

impervious to L1 knowledge? Evidence from the acquisition of plural -s, articles,
and possessive -s. Language Learning, 59, 721–754.

Murakami, A. & Alexopoulou, T. (2016). L1 influence on the acquisition order of

English grammatical morphemes: A learner corpus study. Studies in Second
Language Acquisition, 38, 365–401.

VanPatten, B. (1984). Morphemes and processing strategies. In F. Eckman, L.

Bell, & D. Nelson (Eds.), Universals of second language acquisition (pp. 88–98).
Cambridge, MA: Newbury House.

VanPatten, B. (1987). Classroom learners’ acquisition of ser and estar:

Accounting for the data. In B. VanPatten, T. R. Dvorak & J. F. Lee (Eds.),
Foreign language learning: A research perspective (pp. 61–76). Cambridge, MA:
Newbury House.

VanPatten, B. & Smith, M. (2015). Aptitude as grammatical sensitivity and the

initial stages of learning Japanese as an L2: Parametric variation and case
marking. Studies in Second Language Acquisition, 37, 135–165.

White, L. (2003). Second language acquisition and universal grammar.

Cambridge: Cambridge University Press.

99
Wolfram, W. (1991). Interlanguage variation: A review article. Applied
Linguistics, 12, 102–106.

100
 Following Up
1. On your own, you should be able to succinctly describe and offer
examples of each of the following:

ordered development

U-shaped behavior

constraints on development

variation in staged development.

2. Because of the similarities between ordered development in both English

as an L1 and English as an L2 (e.g., morpheme orders, developmental
sequences) and because of the way in which development seems to be
constrained similarly in both acquisition situations, some researchers
concluded early on that L1 acquisition and L2 acquisition were
fundamentally the same. This was termed the “L1 = L2 hypothesis.” Conduct
a search on this hypothesis and report (1) the evidence behind the hypothesis
at the time and (2) any criticisms of that hypothesis. Add your own
conclusion about what the evidence suggests.

3. Although we will review the effects of instruction on acquisition in a later

chapter, review the Exemplary Study in this chapter (R. Ellis, “Are
classroom and naturalistic acquisition the same?”). In that study, he found
that learners in a classroom showed the same developmental trajectory or
path as learners without any classroom experience. What is your reaction to
this? Do you expect that instruction would “override” ordered development?

4. In this chapter, we have minimized the role of the L1 (i.e., transfer) in L2

development. We do not mean to imply that L1 transfer is non-existent, but
instead that it is constrained. Read either Andersen, “Transfer to somewhere”
or Eckman, “Markedness and the contrastive analysis hypothesis” (see

101
References and Readings) and summarize in 1,000 words or less (or the
equivalent in a PowerPoint presentation!) how L1 transfer is said to be
constrained. Note: Both readings contain technical jargon here and there.

5. One of Greenberg’s universals is this: In languages with dominant order

VSO (i.e., verb subject object, such as Irish, Classical Arabic), an inflected
auxiliary always precedes the main verb. In languages with dominant order
SOV (e.g., subject-object-verb, such as Japanese, Turkish), an inflected
auxiliary always follows the main verb. What predictions would you make
about acquisition based on this universal? In other words, how would
acquisition be constrained by this universal? Do you expect learners of a
VSO language to have consistent problems with word order when learning
an SOV language? Do you expect learners of an SOV language to have
consistent problems with word order when learning a VSO language? Or do
you think something internal to the learner will push that learner along a
particular path because of Greenberg’s universal?

6. Review the poverty of the stimulus (POS) situation described in this

chapter. See if you can summarize in simple terms what the POS is. In your
answer, pay particular attention to ungrammaticality; that is, how learners
come to know what is impossible or ungrammatical in a language. If you
need to, conduct an internet search on “poverty of the stimulus in language
acquisition” to see more examples.

102
 3
What Are the Roles of Input and
Output?
◈

As L2 research quickly emerged in the 1970s and early 1980s, scholars pondered
the ordered development we reviewed in Chapter 2. Clearly something was going
on in the minds and brains of L2 learners regardless of context and regardless of
language being learned. Current theories in the field of SLA argue about what the
learner brings to the task of acquiring another language and how this imposes
order on acquisition. But there is consensus in the field about something very
basic: what the data are that learners use to create a linguistic system. Most
teachers and classroom learners assume the data to be the rules they learn in
textbooks and the practice they engage in (see Chapter 6 for discussion of the role
of instruction). But we see ordered development in and out of classrooms and they
look the same. The reader may recall the study by Rod Ellis presented in Chapter
2 in which the ordered acquisition of German sentence structure is identical for
learners in classrooms in the United Kingdom and for learners of German not in
classrooms in Germany. Non-classroom learners don’t get rules and don’t get the
practice that classroom learners get. How is this possible?
As scholars began to peel back the layers of classroom and non-classroom
comparisons and the nature of ordered development, they zeroed in on the
hypothesis that the data for acquisition must be something other than textbook
rules and practice. What scholars concluded was that the data must be what is

103
now called input. Because input is a ubiquitous term in common everyday
language (e.g., “I’d like your input on this,” “Where’s the video input slot in the
back of the TV?”) we will need to define input as it is used in the acquisition
literature (for both L1 and L2). So we begin with a definition of input followed by
discussion of its role and necessity in acquiring a language.

104
 The Nature of Input
Input refers to the language (in spoken, written, or signed forms) the learner is
exposed to in communicative contexts. It is language that learners hear, read, or
see that carries some sort of message. For example, in child L1 acquisition when a
parent says to a two-year-old “Get your teddy bear, honey. We’re going to
Grandma’s now,” this stretch of speech is input to that child because the parent is
communicating a message that the child is supposed to respond to (e.g., by going
to get the teddy bear). In an L2 classroom in Denmark, when the instructor says to
a first-year learner in French, “Where’s your homework? Did you do your
homework?” this stretch of speech is input to that learner because the instructor is
communicating a message that the learner is supposed to respond to (e.g., by
saying “I don’t have it” or pulling out the homework and showing it). In a
Starbucks in Chicago, the clerk might ask the L2 learner of English at the counter
“What size would you like? We have tall, grande, and venti” while showing the
cups. This stretch of speech is input to that learner because the clerk is
communicating a message that the learner is supposed to respond to (e.g., by
pointing or saying “tall”).
What all of the previous scenarios have in common – whether L1, L2
classroom, or L2 non-classroom – is that input is language embedded in some
kind of communicative event. Learners are actively engaged in trying to
comprehend meaning. Because the focus of the learner with input is on
comprehension, input is not limited to oral or interactive language. As alluded to
earlier, input can be written (street signs, ingredients on a label, stories,
newspaper articles, websites), it can be non-interactive (movies, radio, songs), and
of course it can be signed when communicating with hearing-impaired persons.
Scholars have distinguished different kinds of input. There is input in general
(any language that is embedded in communication intended for a learner to
comprehend) but there are also comprehensible input and modified input.

105
Comprehensible input is usually associated with the scholar Stephen Krashen and
is a construct he developed in the late 1970s. In essence, Krashen claimed that for
input to be useful for language acquisition it must be comprehensible. After all, if
language is not comprehensible in some way, it is mere noise. Lurking behind this
idea is that during the act of comprehension, learners are engaged in mapping
meaning onto form. The internal mechanisms are working on the data found in the
input to create language in the head. The mechanisms can’t do this if nothing in
the speech stream is comprehensible.
Modified input is a term first coined by Michael Long in the early 1980s and
refers to what speakers do when they encounter a communication problem with a
language learner. For example, a speaker might say to a learner “Do you have any
siblings?” When the learner looks puzzled or attempts to say “siblings” with
rising intonation to show she doesn’t understand the word, the speaker might then
say, “Yes. Siblings. Brothers and sisters. Do you have any brothers and sisters?”
The speaker has just modified the original stream of speech and elaborated on it in
an attempt to help the learner comprehend. As in the case of comprehensible
input, modified input is related to the idea that acquisition happens when learners
comprehend messages coded in the input.
Evelyn Hatch, among others, has attempted to describe what happens to
modified input or input made comprehensible to learners. In her 1983 essay,
Hatch compares language addressed to learners with that addressed to natives.
She discusses such things as slower rate and more pauses in the input. She also
identifies the use of higher-frequency vocabulary and fewer uses of pronouns in
addition to more simplified syntax and shorter sentences. In addition, Hatch
describes how input to learners contains more repetitions and restatements, how
the discourse is adjusted, for example by offering options (e.g., “Do you want tall,
grande, or venti?”) rather than leaving things completely open-ended (e.g., “What
size do you want?”). This is just a partial description of what Hatch outlines in
detail about how speakers go about modifying input for learners.

106
 The definition of input as language embedded in a communicative event
(what we might call communicatively embedded input or CEI) suggests, then,
that acquisition is a by-product of learners actively trying to comprehend
language. Acquisition is not a product of learning textbook rules and practicing
them. Thus, textbook rules and lists of verbs and their conjugations are not input
for acquisition. Again, we will discuss the role of instruction and any effects it has
on acquisition in Chapter 6. For now, we want to be sure we delimit and make
clear just what is meant by input. For this reason, we think the term
communicatively embedded input is a good way to think about input. As a
reminder to the reader, communication has a definition used in L2 research and
teaching. First offered by Sandra Savignon and then elaborated on by others and
more recently by Bill VanPatten in his 2017 book, While We’re on the Topic,
communication is defined as the expression and interpretation of meaning in a
given context with a given purpose. From this definition, it falls out that CEI falls
on the side of the interpretation of meaning. We will touch on the expression of
meaning later in this chapter.

Input as language embedded in a communicative event suggests that

acquisition is a by-product of learners actively trying to comprehend
language.

107
 Consider this …
Input must be communicatively embedded and it must, in some way, be
comprehensible for it to be of use for language acquisition. If such data
are required for language acquisition (as scholars have concluded), the
outcome of this claim is that whatever language-creation mechanisms
exist in the mind of the student can only operate on one kind of data. What
general implications does this have for language teaching? What general
implications are there for comparing L1 and L2 acquisition?

Learners are exposed to vast amounts of input over time. The child L1
acquirer at the age of 1.5 years might hear hundreds of sentences during a typical
day. The L2 learner might hear, see, or read hundreds of sentences in a given
week. Yet acquisition is slow. The child L1 learner does not get a fully developed
adult-like system for years. For the L2 learner, given limited contact with the L2,
even more time (in terms of years) is involved. Why is acquisition so slow if input
is so abundant? In a seminal 1967 essay, S. Pit Corder coined the construct
intake. Intake is that subset of the input that a learner can actually make use of at
a given time. Scholars vary slightly on the exact definition of intake, but one
definition we think works best is this: Intake is that subset of the input for which a
learner can connect form and meaning during real-time comprehension. For
example, in the earliest stages of learning English, a learner might hear “The
professor gives way too much homework.” Using context and other non-linguistic
cues, the learner attaches meaning to “professor,” “give,” and “homework.” The
rest is mere noise. It can’t form part of intake. Later in acquisition, the learner
might hear the same sentence and this time process not just “professor,” “give,”
and “homework” but also “too much.” Later, the learner might hear the same
sentence but this time process not only those same words but also “the.” In this
simple example, we see that something internal to the learner constrains what part

108
of the input the learner can actually process while trying to comprehend a speech
stream. While the outside world largely (but not exclusively) controls the kind of
input learners receive, learners are the ones who control the intake – what actually
gets processed at any given point in time and becomes available for acquisition.
And to be sure, the internal mechanisms and constraints on acquisition discussed
in Chapter 2 may further narrow what actually becomes part of the developing
linguistic system.

While the outside world largely controls the kind of input learners receive,
learners are the ones who control the intake – what actually gets
processed.

Interestingly, while intake is a concept accepted by most if not all scholars of

SLA, little research has been conducted to determine what factors and what
mechanisms are involved in the “conversion of input to intake.” Some scholars
have considered the role of working memory – just how much information people
can process at any given moment – and how that affects what parts of the input
learners “get.” Others have looked at linguistic constraints such that learners can’t
process B until they can already process A (e.g., in our example, learners tend not
to process determiners like the and a/an until they can readily process nouns like
professor and dog). And to be sure, there are a few other ideas about constraints
on the processing of input and we have listed some suggested readings in the
conclusion of this chapter. The point is that scholars know that learners don’t
process all of the input they are exposed to at a given point; they just have not
converged on clear explanations of why.
We have mentioned the word process several times and also mentioned
form–meaning connections. Before continuing, it might be a good idea to touch
on the complexity of these notions. Above we used simple vocabulary items like
linking professor to its intended meaning and later the professor to its intended
meaning. One reason the creation of intake is so complex is that there is a lot of

109
hidden “stuff” going on in linking form to meaning. The simple word the, for
example, has a number of features associated with it that must get “tagged” and
incorporated into the word for it to be fully represented in the learner’s
mind/brain. For example, it must be tagged with the feature “D” which stands for
“determiner.” This feature restricts where the word can be used in a sentence; that
is, it can only occur where determiners are allowed. Without this feature attached
to it, the word could show up anywhere in a sentence! But we know it can’t. So
the learner’s mechanisms have to tag it with this syntactic feature. The learner’s
internal mechanisms must also tag it with the feature “specific.” This abstract
feature is what distinguishes the from a/an as well as the lack of any determiner in
sentences such as Dogs make good pets (as opposed to The dogs make good pets).
Another abstract feature that has to get tagged with the is “known to the speaker
and hearer.” This is the feature that allows the person to understand the difference
between The dogs sure are good with the kids and Dogs sure are good with kids.
In the first sentence, the entities (particular dogs and kids) are known to the
speaker and the hearer. We can’t say that for the second sentence. So a simple
morpheme such as the is not so simple as we might first think when we begin to
peel back the linguistic layers surrounding it. It could be, and probably is, that
intake is partly constrained by all the features and behind-the-scenes “stuff” that
have to be mapped onto words and parts of words. Learners just can’t tag
everything at once. And imagine the learner’s mechanisms trying to process an
entire sentence such as The story Anne liked is very good during the earliest stages
of language acquisition.
So far, we have touched on the idea that learners are in control of intake, that
mechanisms internal to their minds constrain what parts of the speech stream get
processed at any given point in time. Some scholars have focused on factors
external to the learner that affect the “conversion of input to intake.” Scholars like
Brian MacWhinney, Nick Ellis, and others focus on factors such as frequency.
More frequent things in the input tend to make it into intake before less frequent
or infrequent things. In ordered development, we see that -ing is much more

110
frequent in input than third-person -s. This might be one (and only one) reason
why -ing is acquired before third-person -s. Another factor is reliability. Does a
particular thing always serve the same function and have the same meaning in the
input? If so, it is highly reliable and is more likely to make it into intake compared
to something that is less reliable (e.g., has multiple functions or meanings). In the
end, because of the complexity of language acquisition, it is likely that factors
both internal and external to learners work in tandem in the creation of intake
from input.

111
Consider this …
Input that learners of languages like Spanish and French receive contains
abundant examples of gender of nouns as well as gender agreement with
adjectives. For example la casa/la maison ‘the house,’ el libro/le livre ‘the
book’ and thus la casa blanca/la maison blanche ‘the white house’ and el
libro blanco/le livre blanc ‘the white book.’ Yet we know that learners
first “get” masculine and singular and only later “get” feminine. What do
you think is happening during the processing of input by learners of these
languages? Do you think somehow they are not processing gender and
gender agreement? Do they skip it in the input?

112
 The Necessity of Input
As we said earlier in this chapter, the role of input in acquisition began as a
hypothesis in the 1970s as researchers struggled to make sense of the ordered
development they were seeing in and out of classrooms and by all ages of
learners. Since the early 1980s, the role of input as a fundamental ingredient in
language acquisition has moved from hypothesis to accepted fact. Learners don’t
build linguistic systems in their minds without CEI. Even those who claim “Well I
learned all the grammar and stuff and this is how I learned language” would be
hard pressed to discount the role of input as they consider what else they did other
than learn rules and memorize verb forms (e.g., read, hear instructors talk, study
abroad, watch movies, seek out speakers of the language). No advanced speaker
of a language has become advanced without extensive interaction with CEI. All
current mainstream theories in both L1 and L2 research accept input as the data
for language acquisition.
The question for us, then, is why is input necessary? Why can’t people learn
a language simply by memorizing rules and verb forms and then practicing them?
The answer is relatively simple: That’s how the mind/brain is wired. Regardless
of one’s theoretical position (and we will explore two of the more widely used
theories in a moment), learners are not blank slates when it comes to language
acquisition. Learners come equipped with internal mechanisms that can only act
on communicatively embedded input when it comes to creating a linguistic
system. There are two broad categories of theories that we can briefly examine to
understand this idea. Those categories are the nativist and the non-nativist
theories.
The most well-known nativist theory is the generative tradition, based on
the work of Noam Chomsky. In the generative tradition, people are genetically
endowed to learn language (as opposed to bees, dogs, and chimpanzees).
Language is defined as a complex and abstract system that, for example, allows a

113
learner of Spanish to know very early on in acquisition that explicit subject
pronouns (‘he,’ ‘she,’ ‘it’) must have antecedents whereas null-subject pronouns
(subject pronouns are not required in Spanish) do not require antecedents. Thus,
in Spanish, both Ella habla and Habla are grammatical and mean ‘She speaks.’
Very early on, even if the learner can barely eke out a sentence in Spanish, that
learner knows that Ellos me robaron ‘They robbed me’ means specific people
committed a robbery and that the speaker knows who they are. The subject
pronoun ellos must have an antecedent. That same learner also knows that Me
robaron ‘They robbed me’ can mean the perpetrators are unknown and there is no
antecedent. Thus only the second sentence can be used to mean “I was robbed”
and not the first. This underlying knowledge is linked to another aspect of Spanish
that learners know early on and that is that when there is no antecedent, null-
subject pronouns are required, as in weather expressions (Está lloviendo ‘It’s
raining’), time expressions (Es la una ‘It’s one o’clock’), impersonal expressions
(Es imposible ‘It’s impossible’), and existential statements (Hay café ‘There’s
coffee’). In these examples, English requires what are called expletive subjects
(‘it’ and ‘there’), but in Spanish such subject pronouns are prohibited, hence the
lack of any expletive subject pronouns in those sentences. What the theory says is
that these properties of null and explicit subject pronouns are part of Universal
Grammar (UG) and available to all learners of language before they ever hear a
word of a language (both L1 and L2). UG shapes and constrains the kinds of
unconscious “hypotheses” the learner can make about how languages work. This
is why the syntactic properties of null and explicit subject pronouns come in so
quickly in language acquisition. In this view, language is what is called modular:
It is special and is learned with mechanisms that are particular to language (i.e.,
UG and language processing mechanisms). Because the linguistic system is
modular, it can only operate on the kind of data it regulates: language embedded
in communicative contexts. It can’t operate on explanations or rules provided by a
textbook. And for the record, the properties of null and explicit subject pronouns
are never taught to either L1 or L2 learners yet both types of learners quickly

114
“get” these properties. What happens to learners is that as soon as they begin to
interact with communicatively embedded input from the environment, “UG goes
to work” to determine if the language allows both null and explicit subject
pronouns or only explicit subject pronouns. Once it does this, the relevant
properties are “kicked in” without the learner even knowing.
An example of a non-nativist tradition is usage-based approaches. Usage-
based approaches agree that what winds up in the head is a complex and abstract
array of “stuff” that we call language, but usage-based approaches do not posit a
role for something like UG. Instead, usage-based approaches claim that general
learning mechanisms (the same kind that, for instance, allow a child to know that
a pelican is a bird and not a rodent when seeing a pelican for the first time) are
used in language acquisition. However, to create a linguistic system in the
mind/brain, those general learning mechanisms need access to basic linguistic
data for all the “categories” and “patterns” associated with language. In short, the
learning mechanisms need communicatively embedded input for language to
grow, examples of birds embedded in real-life events for the notion of “bird” to
grow, and so on. Each area of abstract knowledge in the head of a learner is
dependent on an interaction of real-world examples with general learning
mechanisms.
The point in this brief exposition is that differing theories about what
underlies language acquisition converge on what the basic data are for acquisition.
Whether you believe that language is special and therefore language acquisition is
special or whether you believe that language acquisition is the result of the same
processes as something like category learning (e.g., a bird is not a rodent), you
would still believe that acquisition happens as a result of internal mechanisms
operating on CEI: communicatively embedded input.

Differing theories about what underlies language acquisition converge on

what the basic data are for acquisition: communicatively embedded input.

115
 In Chapter 5 we will explore the issue of whether nativelikeness is possible
in language acquisition. For now, let’s accept a general observation that many L2
speakers are non-nativelike. They have accents. They sometimes produce
sentences or use words that don’t sound the way native speakers make sentences
or use words. Such observations have led some scholars to ask whether CEI is
sufficient for L2 acquisition. It is clearly sufficient for L1 learners or they’d never
become native speakers. But because L2 acquisition appears to be “incomplete”
for many learners, some scholars have made the claim that while input is
necessary, it isn’t sufficient. This has led them to posit a role for instruction, for
example – the idea being that input needs to be supplemented by some kind of
explicit focus on the formal properties of language. Again, we will take up the
role and effects of instruction in Chapter 6, but as a preview we say here that the
effects of instruction on language acquisition tend to be exaggerated. But even if
there are positive and beneficial effects of instruction, these results do not
necessarily suggest that input is not sufficient for acquisition. Let’s look at the
logic of this position for a moment.
One problem with the input-is-not-sufficient claim is that there are learners
who become nativelike (Chapter 5). If non-nativeness is a central criterion for
claiming that input is not sufficient, then those learners who do become nativelike
pose a problem for the argument – especially those that do so when all they have
is exposure to input over time. What is more, being nativelike is not an either/or
proposition. A learner could be nativelike in some parts of the language and not
nativelike in others (again, see Chapter 5). This situation, too, poses problems for
the fundamental claim of the input-is-not-sufficient argument. But let’s just
assume for the moment that L2 learners generally don’t become nativelike. It
could be that this is just the way it is and that nothing can make an L2 learner
nativelike. Although we don’t necessarily agree with this position, it is a possible
claim. If so, then the input-is-not-sufficient argument goes out the window: Input
is sufficient and it’s just that L2 acquisition tends to result in non-nativeness for

116
reasons that researchers have yet to articulate in a convincing way. Again, these
issues are taken up in Chapter 5.
Another problem with the input-is-not-sufficient claim is that it tends to
focus on time on task based on years. In a later chapter we will demonstrate that
the way to view the time on task for acquisition is hours and not years. For
example, by the age of five, a child L1 learner has amassed thousands and
thousands of hours of interaction with CEI. An adult, on the other hand, who
begins learning Japanese in the classroom does not – and cannot – match the
thousands of hours of interaction with input during a five-year span. What this
observation suggests is that it’s not that input is not sufficient, it’s that most L2
learners just don’t spend the time interacting with input that first-language
learners do. One final problem with the input-is-not-sufficient argument is that the
research on instruction – teaching learners language explicitly to help them along
– has not borne the fruit of its efforts. It’s not clear that instruction really does
anything. We will take up that matter in Chapter 6.
In short, although it could be the case that some kind of evidence or research
may someday show us that input is not sufficient, at this point the argument is
speculative at best. It is based on the observation that many L2 learners do not
become nativelike. Our position is that the default hypothesis should be that input
is both necessary and sufficient and that one goal of L2 research is to determine
why most L2 learners don’t become nativelike or why some parts of the linguistic
system don’t become nativelike. Once more, some of this research is discussed in
Chapter 5.

117
Consider this …
If we take the default position that communicatively embedded input is
both necessary and sufficient for language acquisition, what do you think
might be the reasons for L2 learners tending to be non-native in the long
run compared to L1 learners? Here are some possibilities for you to
ponder: motivation, the learner already has an L1, quality of input,
working memory deteriorates with age, social factors.

118
 Output
There is the claim floating around in some teaching circles and in the general
public that you have to speak in order to learn a language. The idea is that you
can’t learn Arabic or Spanish unless you practice speaking these languages. What
does the research say, if anything, about this position? Before we explore the role
of speaking (or signing or writing) let’s be sure what we mean by speaking,
referred to as output in the L2 literature.
Output is generally defined as any attempt by an L2 learner to produce
language in spoken, written, or signed form – although most L2 research on
output deals with oral output. When learners produce output, they are not engaged
in comprehension as they are with input. They are doing something else. One kind
of output is what we might call “practice,” in the traditional sense of the term.
This is the kind of output that many language learners encounter. Learners repeat
something after the instructor (or the voice on the software program they are
using). Learners translate a sentence from the L1 to the L2. Learners are asked to
make a sentence in the past tense to show they can make the past tense. Learners
are asked to write full sentences as answers to particular prompts (e.g., “Where do
you live?” “I live in London.”). Learners are asked to describe their daily routines
using reflexive verbs in Spanish or pronominal verbs in French. What all of these
output events have in common is that learners are producing language for the sake
of producing language. This is output as practice.
The other major kind of output can be termed communicative output or
communicatively embedded output. The reader may recall our definition of
communication from earlier: the expression and interpretation of meaning in a
given context for a given purpose. Context is defined by participants and setting,
and purpose refers to why meaning is exchanged (e.g., social reasons, to learn
something, to remember something later, as when we repeat a phone number). So
communicative output refers to the kind of language learners produce in order to

119
express meaning to someone else in a given context for a given purpose. When a
learner approaches someone at a bus stop and asks, “Does bus number 25 come
here?” that learner is engaged in communicative output. When a learner says to an
instructor “Professor, I don’t understand this word” and points to a word on a
page, that learner is engaged in communicative output. When a learner witnesses
an accident and the police interview that learner-witness, the learner engages in
communicative output during the interchange. What all of these scenarios have in
common is communicative purpose and not practice; that is, learners are speaking
not because they are practicing sentence structure, verb forms, or vocabulary –
they are speaking because some kind of meaning-making is relevant to the task or
situation at hand. Those learners need information, are seeking help, or are asked
to supply some kind of information for a given task, for example.
So the question that confronts L2 research is whether output as practice,
communicative output, or both play a role in L2 acquisition. If so, what is that
role? Because output as practice is largely confined to classrooms and explicit
learning environments, we believe any role it plays is part of instruction. We thus
will not address the role of practice here and direct the reader to Chapter 6 on the
effects of instruction on acquisition. Our focus in this chapter will be on the role
of communicative output.
There are three possible ways to view the role of communicative output in
L2 acquisition. The first is that it is necessary. The second is that it is not
necessary but is somehow beneficial. The third position is that it is neither
necessary nor beneficial; in essence, it does not affect L2 acquisition. Before
exploring these positions, let’s be clear what we mean by acquisition: the creation
of a linguistic system in the mind/brain of the learner. The question here is the
role of communicative output in how language grows inside the learner. We are
not focused on how learners make use of language during communication, what is
called by any number of terms (not all meaning exactly the same thing): skill,
proficiency, interactional ability, communicative ability, among others.

120
 Output Is Necessary
Most theories and perspectives in contemporary L2 research do not hold that
communicative output is necessary for L2 acquisition. The one exception is skill
theory. Under skill theory, there is no concern for what grows in the mind/brain of
the learner but instead the focus is on how particular abilities to deploy language
in real time develop. A major proponent of skill theory is Robert DeKeyser.
Borrowing from research in cognitive and educational psychology, DeKeyser has
argued that some kind of output is necessary during acquisition so that declarative
knowledge (knowledge of) can become procedural knowledge (doing), which in
turn can become automatized (be done without thinking) for the purpose of
speaking (or signing). It is important to underscore here that DeKeyser and those
working within skill theory are not concerned with how language grows in the
mind/brain; they are concerned with the development of skill, as the name of the
theory clearly suggests. In a sense, then, it is not fair to claim or imply that skill
theorists believe that communicative output is necessary for a linguistic system to
develop; their position is that it is necessary for skill to develop.

Consider this …
In the 1970s, Sandra Savignon made the claim based on her research that
learners learn to communicate by engaging in communicative events. This
notion has been supported by research since then. In a sense, she was
suggesting that the skill of using language to express and interpret
meaning develops because learners are constantly put in situations in
which they have to express and interpret meaning. How do learners who
are at the lower levels of proficiency communicate? Can you describe
what they do? Here are some ideas to consider: How dependent are they
on the other speaker for the construction of the conversation? How easily
do they initiate topics or ask questions of the other speaker? How well do

121
 they elaborate in an answer to a question? Consider the following
fictitious example of a newly arrived exchange student to the United
States with minimal ability. B is the exchange student and A is a native
speaker at the school where B will be taking classes.

A: Are you new to the United States?

B: Sorry?

A: New. Are you new to the United States? Did you recently arrive?

B: Oh, yes.

A: When? When did you arrive?

B: Uh, two weeks.

A: Two weeks ago?

B: Yes. Two weeks.

In 1985, Merrill Swain launched what she called “The Output Hypothesis.”
Swain was interested in the outcomes of elementary school immersion programs
in Canada – programs in which learners didn’t have language classes per se but
instead learned language by studying content matter such as history and science.
What Swain and her colleagues documented was that learners in immersion
programs were far from being nativelike in their abilities. While better than
learners in traditional language programs, they were far from being “perfect” in
using the L2. This situation caused Swain to ponder what was going on in
immersion classes and what the learners were doing. She concluded that while
communicatively embedded input was necessary or essential to L2 acquisition,
there was something missing (i.e., input is necessary but not sufficient). What was
missing was communicative output: the constant opportunity to express one’s
own meaning in the L2. Swain argued that the role of such output was threefold:
(1) to allow the learner to develop skill (again, we will touch on this later); (2) the

122
opportunity to try out “hypotheses” about how the language works while making
meaning; and (3) having to make meaning in the L2 could trigger the learner to
focus more on how to say something (what Swain called moving from “semantic
processing to syntactic processing”). Later, in 1998, Swain would also discuss
how having to make output has a metalinguistic function in getting learners to talk
about language.
Since Swain’s hypothesis, a good deal of research has been conducted in and
out of classrooms to examine the role of output. Unfortunately, some of the
research has strayed from Swain’s original intent of communicative output such
that some scholars lump output as practice along with communicative output. The
research that best captures Swain’s original intent would be that research that
focuses on actual interactions between L2 speakers and others in which meaning-
making is at the center of what the interlocutors are doing. However, it is very
difficult to isolate variables in such research and to actually show that
communicative output is necessary for language growth. As we will see, such
research falls more into the camp that communicative output is beneficial.

123
 Output Plays Little to No Role
At the extreme end of output is necessary is the position that output plays
(essentially) no role in the acquisition of language. In other words, output has no
effect on linguistic development itself. It is difficult to find scholars who would
agree with this position wholesale. We can, of course, find scholars who would
agree with the idea that output as practice does little to nothing for the
development of the linguistic system; that is, that you don’t internalize a linguistic
system by practicing. We might even find that a majority of scholars agree with
this position in one way or another. But that is not the focus of this chapter. We
are focused on communicative output – when learners speak to express meaning
and interact with someone else. So we conclude this very brief subsection by
stating that the idea that communicative output plays absolutely no role (either
direct or indirect) in the development of a linguistic system is the least tenable
position of the three outlined in this chapter.

124
 Output Is Beneficial
In between the positions that output is necessary and output does little to nothing
for acquisition is the position that communicative output is somehow beneficial
but not necessary. By engaging in interactions with others, learners benefit in
some way. Most scholars would agree with this, but there are different positions
on just what parts of acquisition output might positively affect.
One position is that when learners engage in interactions with other speakers,
their participation as meaning-makers creates more input for them – and possibly
better input in that the input might be adjusted for their level. We caught a
glimpse of this in the section on input when we reviewed the concept of modified
input. Michael Long, Stephen Krashen, and others have tended to align with this
position. As we said earlier in this chapter, how a learner responds to another
speaker may trigger adjustments in how that other speaker talks. That speaker
might slow down, repeat, rephrase, pause more, and so on (see the discussion
above of Evelyn Hatch’s description of input to non-native speakers). Michael
Long has argued that such adjustments can increase comprehension. When
comprehension is increased, acquisition is fostered. That is, modifications
increase what Stephen Krashen has called comprehensible input. What is more,
by engaging with another speaker, that learner is getting that person to talk more,
which simply increases the amount of input as well. So the idea here is that a
learner engaged in making meaning through communicative output gets more
conversational input from other speakers. The more input, the better – and the
more the input is comprehensible, the better.
Another position is similar to one of Swain’s ideas, namely that by engaging
in communicative output, a learner can be pushed to be more precise and that this
may trigger some kind of learning (although it is not made clear in the literature
just what that learning is or the underlying psycholinguistic processes involved).
Unlike the position that such processes are necessary, those in this camp tend to
claim it is beneficial or facilitative. Names associated with this position include

125
Susan M. Gass, Alison Mackey, and others. Many in this research tradition claim
that when learners produce output, they can then somehow “check it” against
input to see where there are gaps in their knowledge. This is called noticing the
gap. Here is a sample interchange from Mackey’s research that shows how,
during the meaning-making process, the learner comes upon a new and more
appropriate way to talk about a particular concept (see the 2012 reading at the end
of this chapter for the source). Note that NS stands for native speaker and NNS
stands for non-native speaker.

NNS: And in hand in hand have a bigger glass to see.

NS: It’ err. You mean, something in his hand?

NNS: Like spectacle. For older person.

NS: Mmmm, sorry I don’t follow, it’s what?

NNS: In hand have he have has a glass for looking through for make the print
bigger to see, to see the print, for magnify.

NS: He has some glasses?

NNS: Magnify glasses he has magnifying glasses.

NS: Oh aha I see a magnifying glass, right that’s a good one, ok.

In this interchange, we see that the learner does not have the word
magnifying glass at the outset. Through the course of interaction, the NS keeps
giving indications that the NNS is not being clear. This pushes the NNS to keep
talking until in the end the NNS hits upon the word magnifying glass and there is
an “aha moment” for the NS. Communication is successful. What this interchange
exemplifies is a classic negotiation of meaning scenario. Although
communication is defined as the expression and interpretation of meaning in a
given context for a given purpose, communication is not always successful – in an
L1 or an L2. When it is clear to one or more of the parties that something is not
understood or not understood as it was intended, the interlocutors may enter into a
negotiation phase. In this sample interchange, negotiation is first seen in the NS

126
query “You mean, something in his hand?” The NS is seeking confirmation about
what he or she thinks the other person is saying. Such negotiations may push
learners to search for alternative ways to say something, perhaps to be more
precise. Or they may push the other speaker to say something like “Oh! You mean
a magnifying glass!” In either case, the idea in such interactions is that learners
may benefit by learning something about language because of their own
participation in the conversation. They “notice the gap” in what they want to say
and whether or not they can say it.
The astute reader may have noticed that in the above interchange, the focus
was on a lexical item. That is, the problem in communication was a particular
word the NNS did not know, and what the learner arrived at in the end was a new
word: magnifying glass. But what about formal parts of the linguistic system such
as word endings, sentence structure, the sound system, and so on? In 2003, Susan
Gass stated that interaction might only have an effect on what she termed “low-
level phenomena” such as some aspects of pronunciation and the basic meanings
of lexical items. At the time of her writing, there was no research to claim that
interaction affected the formal parts of language (except see the Exemplary Study
in this chapter). We do not see that such evidence has emerged since 2003, at least
not compelling evidence. Interaction seems to affect precisely what Gass
suggested it might: non-formal phenomena that affect meaning-making and
successful communication. To put it in “notice the gap” terms, the research does
not show very strong evidence that learners notice gaps between their sentence
structure or word formation, for example, and that of their interlocutors. As we
have seen elsewhere in this book, acquisition of the formal elements of language
is constrained in multiple ways, so it may not be so surprising that the formal
elements of language are the last to receive any “notice the gap” treatment by
learners during interaction.
To be sure, some of those involved in interaction research would claim that
the research has shown effects for sentence structure and more formal elements of
language. The problem with this conclusion is that the research relies on what is

127
called uptake as well as explicit learning. Uptake refers to that moment in an
interaction when a learner repeats something he or she just heard. This
interchange comes from Bill VanPatten, who has reported this interchange before;
it is one he overheard in the locker room after his men’s tennis league was done
with its doubles matches for that week. Bob, the native speaker, is asking Tom,
the non-native, about his doubles partner.

BOB: So where’s Dave?

TOM: He vacation.

BOB: He’s on vacation?

TOM: Yes. On vacation.

BOB: Lucky guy.

In this interchange, Tom demonstrates that he does not have the preposition
phrase on vacation in his repertoire yet. When Bob confirms what he hears
through what is called a conversational recast (saying what the learner said but in
a nativelike way), Tom repeats what Bob says to mean “yes, that’s what I meant.”
When Tom uses on vacation at this moment, this is called uptake. Uptake is not
the same as intake in that researchers do not know if what is “taken up” during the
conversation actually goes anywhere (i.e., is “taken in”). Most research claiming
that interaction affects formal elements of language tend to rely on uptake data.
As researchers such as Susan Gass and María Pilar de García Mayo have
suggested, however, these kinds of “snapshot” studies are unable to speak to
whether anything is actually acquired and if there are long-term effects of a
particular interaction or even sets of interactions. In other words, the theoretical
status of uptake and how it interacts with acquisition is unclear. What is more, it
would be almost impossible to design research to show a connection between
uptake and acquisition in spite of some attempts to do so. Most research on uptake
during interactions is based “on the moment” and not on long-term effects (e.g.,
going back months later to see if learners now have something in their linguistic
systems that they didn’t have before).

128
Exemplary Study

Mackey, A. (1999). Input, interaction, and second language development?

An empirical study of question formation in ESL. Studies in Second
Language Acquisition, 21, 557–587.

Participants
34 learners of ESL at different developmental levels

L1 included Korean, Mandarin, Cantonese, Indonesian, Thai, Japanese,

Spanish

location: Sydney (Australia)

Major Target Item

wh-question formation in English

previous research on ESL has revealed the following developmental

stages

1. subject-verb-object: It’s a monster? Your cat is black? I draw a

house here?

2. fronting of wh-/do/Q word: Where the cats are? Does in this

picture there is a cat?

3. pseudo-inversion: Have you got a cat? Had you drawn the cat?

4. auxiliary verb in second position: What do you have? What have

you got in your picture?

5. non-inversion in subordinate clause: Can you tell me where the

cat is?

tag questions: It’s on the wall, isn’t it?

129
 Neg Q: Doesn’t your cat look black?

Materials and Procedure

participants engaged in tasks such as story completion, picture
sequencing (discovering the order of a pictures in a story), spot-the-
difference, picture drawing

participants were divided into five groups: interactionally modified

input through tasks (n = 7); developmentally unready (n = 7); observers
(unlike the first two groups, they watched but did not participate in
interactions) (n = 7); premodified scripted group (a speaker read a script
but no meaning was negotiated) (n = 6); control (n = 7)

participants engaged in tasks for 10–15 minutes per day during week 1,
the same one week later, and the same three weeks later

spot-the-difference tasks were used as a pre-test and a post-test

instrument

pre- and post-test tasks were coded for developmental stage of the
participants and for different stages of questions produced

Results
A majority of participants showed an increase in stage development
except for the control and scripted group.

The changes in production of question types for the other groups all
proved to be significant (e.g., these participants produced more
questions from stages 4 and 5, see above).

Conclusion
Conversational interaction seems to facilitate L2 development.

130
 Finally, there is an underlying assumption in the research on interaction and
learner output that has gone largely unstated – and that assumption is that
somehow explicit learning is linked to acquisition. Explicit learning refers to
consciously focusing on something in the language. We will touch on the
potential role of explicit learning in Chapter 7, but here we will simply state that
there is no clear link between explicit learning and acquisition. This creates
questions for research on interaction and output because so much of the research
reports a very explicit focus on surface features of language. Returning to the
example from Alison Mackey cited above, it should be obvious to the reader that
there is an explicit focus on the right word magnifying glass. But let’s look at
another example regarding a surface feature of language. This example comes
from research by María Pilar García Mayo published in 2002 and is exemplary in
that we find many such examples, partly in response to Merrill Swain’s claim that
output serves a metalinguistic function.

L E A R N E R 1 : men are less incline … it has to be an adjective … inclined to

confess, you are inclined to do something

L E A R N E R 2 : to confession …

L E A R N E R 1 : to confess

L E A R N E R 2 : but after a preposition …

L E A R N E R 1 : to confess … what?

L E A R N E R 2 : to is a preposition

L E A R N E R 1 : yeah …

L E A R N E R 2 : so it should be followed by ing

L E A R N E R 1 : inclined to confessing

L E A R N E R 2 : yeah

L E A R N E R 1 : no, because to is part of the second verb … inclined to confess …

yeah

131
L E A R N E R 2 : ok I trust you

This interchange typifies what researchers call a language-related episode

in which learners engage in explicit discussion about language itself and how it
works – almost like a mini-lesson. As the reader can see, there is a very explicit
focus here, with use of words such as adjective, preposition, part of the second
verb, and so on. While teachers may delight in such an interchange and
researchers may find it interesting, the claim that such interchanges foster
acquisition depends on some kind of hypothesis within a theory of language
acquisition that such explicit learning can “turn into” or “foster” the development
of an implicit underlying linguistic system, what we have called in this book
mental representation. As we will see in Chapter 7, it is not at all clear that such a
connection between explicit learning and acquisition can be made.
With the above said, there is one seemingly clear piece of evidence that
interaction might affect the development of formal elements at the level of the
sentence. This evidence is illustrated in the Exemplary Study in this chapter, the
study by Alison Mackey published in 1999. However, we note here that although
we find this study intriguing and hopeful in terms of what it suggests about the
effects of interaction, it is an isolated study and short term in nature (see
comments in Chapter 6 about the problems of short-term studies). From the
excerpts provided by Mackey, the interactions suggest that learners who are
participating in the tasks are getting higher-quality and perhaps “targeted” input.
Here’s an example from Mackey’s 1999 study.

NNS: What the animal do?

NS: They aren’t there, there are no bears.

NNS: Your picture have this sad girl?

NS: Yes, what do you have in your picture?

NNS: What my picture have to make her crying? I don’t know your picture.

NS: Yeah ok, I mean what does your picture show? What’s the sign?

132
NNS: No sign? … No, ok, What the mother say to the girl for her crying?

NS: It’s the sign “no bears” that’s making her cry. What does your sign say?

NNS: The sign? Why the girl cry?

This sample comes from one of the treatment interactions (not the pre- or
post-tests). We have highlighted here what the native speaker is doing during the
task. Notice the consistent and constant use of what questions. If we assume that
this kind of interaction happened consistently during all phases of the study, we
might easily conclude that the learner was getting modified input that repeatedly
used this kind of question. We also note that this is a stage 5-type question and
production of stage 4 and 5 questions was of particular interest to Mackey. Two
things could have occurred in this study (and others like it): (1) the learner got
mini-massive doses of wh-questions in the input that pushed acquisition along; (2)
learners began “copying” wh-questions from what they were hearing and using
them for communication without the internal system having actually progressed in
terms of the underlying complexities regarding the features and operations
required to make wh-questions. In other words, they could have explicitly learned
something and then applied it during interactions in this particular study yet their
underlying mental representation for how wh-questions actually work has not
changed. This possibility becomes greater when we look at another treatment
interchange provided by Mackey.

NNS: You have girl too?

NS: Yeah I do, how many girls in yours?

NNS: I have a one girl my picture. How many girl?

NS: I’ve got three girls in my picture.

NNS: Ok two difference. Another one. How many?

NS: How many what?

NNS: How many bird birds in you picture?

133
NS: Birds? There are lots of birds …

NNS: Ok, same. How many tree?

( T W O L I N E S L A T E R ) N N S : How many flower in your picture?

( T W O L I N E S L A T E R ) N N S : In your picture, how many blue sky?

( T W O L I N E S L A T E R ) N N S : How many small animal?

( T W O L I N E S L A T E R ) N N S : How many sun?

( S E V E R A L L I N E S L A T E R ) N N S : How many lines … sun lines?

In this interchange, it seems that the NNS picked up on the “how many” to
ask about numbers after the NS says “How many girls in yours?” and then a bit
later “How many what?” From then on, the NNS keeps using how many +
sentence, almost in a formulaic way. Although this may be a productive and
useful communicative strategy, we don’t know from this if there is any change in
the underlying syntax of how wh-operators work in English. It would remain to be
seen what their spontaneous non-experimental speech would look like, say,
outside the classroom and outside the experiment as well as what their underlying
competence looked like when probed with non-speaking tasks.
Returning to Susan Gass’s 2003 claim, it remains to be seen whether
interaction research and thus the contribution of learners’ communicative output
actually affects the development of the formal features of language in the
mind/brain. As we hinted earlier, when learners are engaged in interactions or use
output to communicate meaning, they develop important communication skills.
But that is not the focus of our discussion here. For now, we end this section
saying that one of the clearly beneficial roles of learners engaging in meaning-
making via output is that it gets them more input. Given the critical role of input
in language acquisition and given that learners may incidentally push other
speakers to modify their input, this is not a trivial benefit.

134
It remains to be seen whether interaction research and thus the
contribution of learners’ communicative output actually affects the
development of formal features of language in the mind/brain.

What about Social Factors?

Social factors must clearly impact how learners interact (or don’t) with
speakers of the language they are learning. And part of these social factors
may be cultural conditioning. For example, if interaction pushes learners
to “notice the gap” as suggested above, how do learners react to such
things as other speakers asking for clarification or even providing
feedback? Do they see it as challenging or do they see it as helpful and a
normal part of the communication process? And are learners more
accepting of questions and feedback from certain kinds of people as
opposed to others? Scholars like Jenefer Philp and Alison Mackey, along
with Elaine Tarone, among others, suggest that the identity and social
roles of interlocutors in L2 situations may indeed play a role in
interactions and how learners perceive what the other speaker is doing.
This would, of course, somehow affect the degree to which learners
actually learn from interactions and perhaps what they learn.

135
 Recap
In this chapter we have reviewed the following major ideas.

the nature of input as communicatively embedded input as well as the

constructs of comprehensible input and modified input

the fundamental and essential role that input plays in the acquisition of a
linguistic system

the nature of communicative output as part of meaning-making

three perspectives on the role of output in language acquisition:

◦ two not widely accepted positions, that output is necessary or that

output contributes little or nothing to language acquisition

◦ that output embedded in interactions is beneficial to language

acquisition

the benefits of interaction and output are short term/in the moment and
tend to rely on episodes of explicit learning (such as uptake), and it is not
clear what aspects of language as part of mental representation are actually
affected by interaction

interaction and output can and often do result in modified and more
comprehensible input, which in turn can foster language acquisition given
the fundamental role of input.

136
 References and Readings

DeKeyser, R. M. (2015). Skill acquisition theory. In B. VanPatten & J. Williams

(Eds.), Theories in second language acquisition: An introduction. 2nd edition (pp.
94–112). Mahwah, NJ: Lawrence Erlbaum.

Ellis, N. C. (2012). Frequency-based accounts of SLA. In S. M. Gass & A.

Mackey (Eds.), Handbook of second language acquisition (pp. 193–210). New
York: Routledge.

García Mayo, M. P. (2002). Interaction in advanced EFL pedagogy: A

comparison of form-focused activities. International Journal of Educational
Research, 37, 323–341.

García Mayo, M. P. & Alcón Soler, E. (2013). Negotiated input and

output/interaction. In J. Herschensohn & M. Young-Scholten (Eds.), The
Cambridge handbook of second language acquisition (pp. 209–229). Cambridge:
Cambridge University Press.

Gass, S. M. (1998). Input, interaction, and the second language learner. Mahwah,
NJ: Lawrence Erlbaum Associates.

Gass, S. M. (2003). Input and interaction. In C. Doughty & M. H. Long (Eds.),

The handbook of second language acquisition (pp. 224–255). Oxford: Blackwell.

Gass, S. M. & Mackey, A. (2015). Input, interaction and output in second

language acquisition. In B. VanPatten & J. Williams (Eds.), Theories in second
language acquisition. 2nd edition (pp. 180–206). New York: Routledge.

Hatch, E. M. (1983). Simplified input and second language acquisition. In R. W.

Andersen (Ed.), Pidginization and creolization as language acquisition (pp.
64–86). Cambridge, MA: Newbury House.

137
Krashen, S. (1982). Second language acquisition and second language learning.
Oxford: Pergamon.

Krashen, S. (2009). The comprehension hypothesis extended. In T. Piske & M.

Young-Scholten (Eds.), Input matters (pp. 81–94). Bristol: Multilingual Matters.

Leeser, M. J. (2008). Pushed output, noticing, and development of past-tense

morphology in content-based instruction. Canadian Modern Language Review/La
Revue canadienne des langues vivantes, 65, 195–220.

Long, M. H. (1983). Linguistic and conversational adjustments to non-native

speakers. Studies in Second Language Acquisition, 5, 177–193.

Long, M. H. (1985). Input and second language acquisition theory. In S. M. Gass

& C. G. Madden (Eds.), Input in second language acquisition (pp. 377–393).
Rowley, MA: Newbury House.

Mackey, A. (1999). Input, interaction, and second language development: An

empirical study of question formation in ESL. Studies in Second Language
Acquisition, 21, 557–587.

Mackey, A. (2012). Input, interaction and corrective feedback in second language

learning. Oxford: Oxford University Press.

MacWhinney, B. (1987). The competition model. In B. MacWhinney (Ed.),

Mechanisms of language acquisition (pp. 249–308). Hillsdale, NJ: Lawrence
Erlbaum Associates.

Philp, J. & Mackey, A. (2010). Interaction research: What can socially informed
approaches offer to cognitivists (and vice versa)? In R. Batstone (Ed.),
Sociocognitive perspectives on language use and language learning (pp.
210–228). Oxford: Oxford University Press.

Savignon, S. (1998). Communicative competence: Theory and classroom

practice. 2nd edition. New York: McGraw-Hill.

138
Swain, M. (1985). Communicative competence: Some roles of comprehensible
input and comprehensible output in its development. In S. M. Gass & C. G.
Madden (Eds.), Input in second language acquisition (pp. 235–253). Rowley,
MA: Newbury House.

VanPatten, B. (2004). On the role(s) of input and output in making form–meaning

connections. In B. VanPatten, J. Williams, S. Rott, & M. Overstreet (Eds.),
Form–meaning connections in second language acquisition (pp. 29–47).
Mahwah, NJ: Erlbaum.

VanPatten, B. (2015). Input processing in adult SLA. In B. VanPatten & J.

Williams (Eds.), Theories in second language acquisition. 2nd edition (pp.
113–135). New York: Routledge.

139
 Following Up
1. Review the recap that concludes the formal part of this chapter. Can you
expand on each idea and explicate it in your own words?

2. Read Hatch, “Simplified input and second language acquisition” in

References and Readings. Create a check sheet of features of simplified input
that she discusses. Then collect 10 minutes of native-to-native speech of a
language you are familiar with and 10 minutes of that same language but
native (or teacher) to non-native (preferably novice or intermediate learner).
Which of Hatch’s characteristics can you isolate and exemplify? Prepare a
brief presentation on your findings.

3. In this chapter we have distinguished between output as practice and

output as part of communication (i.e., communicative output). Under this
distinction, output as part of communication has to have a purpose other than
practice; that is, learners don’t produce language because someone has asked
them to. They produce language because meaning-making is at the core of an
event. With someone else, review the following questions. What ideas
surface in your discussion?

a. What is the purpose of communication? Why do people communicate?

b. What does “meaning” mean?

c. If a teacher asks a student “What color is this pen?” and the student says
“It’s red,” is this an example of communication? How does it compare to
what people do outside of the classroom?

4. Underlying much of the discussion in this chapter is the notion of

communication: the expression and interpretation of meaning in a given
context for a given purpose. Both input and output must be embedded in
communication for either of them to be of use. What factors do you see
influencing a learner’s access to communicative events? With someone else

140
make a list and present it. Can you categorize or typify the factors in any way
(e.g., how many of them are related to “social” ideas and factors?)?

5. A reasonable conclusion to this chapter would state that communicatively

embedded input is necessary for language acquisition and the door remains
open that communicatively embedded output may play an indirect role in
acquisition. Yet many people, including teachers, believe that you have to
speak in order to “learn the grammar” or “learn the language.” Where do you
think this belief comes from? Is it a belief that you subscribe to? If so, come
back and visit this belief after reading this book and decide if it has changed
at all.

6. Read either Long, “Linguistic and conversational adjustments to non-

native speakers” or Leeser, “Pushed output, noticing, and development of
past-tense morphology in content-based instruction” (see References and
Readings) and prepare a summary similar to the summary of the Exemplary
Study in this chapter.

141
 4
What Is the Initial State?
◈

In his bestselling book, The Language Instinct, Steven Pinker makes the
compelling case that children do not begin language acquisition with a “blank
slate.” Prior to the revolution of linguistics in the 1960s and the work in child
language acquisition that was initiated around the same time (see Chapter 1),
many people believed that children were empty vessels upon which the
environment could make its imprint once they were born. As “blank slates” babies
came into the world with nothing, so to speak, and were at the mercy of the
environment for learning. In short, babies aren’t hard wired for anything.
Although there is no doubt that the environment shapes learning, we now know
that in a variety of domains, children seem to come into the world equipped with
special learning mechanisms that help them sort out the environmental stimuli
they are exposed to. One of those mechanisms is claimed to be Universal
Grammar, at least by those working in the Chomskyan tradition. (We touched on
Universal Grammar in Chapter 2 and will revisit it here.) We might say that the
“initial state” for child L1 acquisition is Universal Grammar and not a blank slate.
The question in this chapter is the nature of the initial state for L2 learners.
What do they begin the task of acquiring language with? Clearly, L2 learners
cannot be blank slates if it is the case that children are not. L2 learners must begin
the process of acquiring another language with something. But what is that
something? In this chapter we will explore three possibilities based on the
generative tradition:

142
 the initial state is Universal Grammar

the initial state is the L1

the initial state contains parts of the L1 but not all.

As we will see, even with the second two positions, Universal Grammar is
operative in the L2 context. That is, regardless of the starting point, Universal
Grammar is constraining acquisition all along the way as it does in L1 acquisition.
We will begin with a brief overview of Universal Grammar in order to provide
background for this chapter.

143
 Universal Grammar
As noted in Chapter 2, UG is part of the generative tradition in linguistics. The
nature of UG has changed somewhat as the theory has progressed over the years
(the construct has been around since the early 1960s) and in its current form the
characteristics of UG are the following:

it is human specific and innate (i.e., no other animals have UG)

it consists of a finite set of abstract features and principles (or constraints)

it guides and constrains language acquisition the minute a human begins to

hear (or see) language in a communicative context.

It should be clear to any reader that animals do not have language. They do
have systems of communication, to be sure, and some are quite complex. But
animals do not have such things as syntax, morphology, and so on (i.e., the formal
and abstract components of language). Even “signing apes” such as Koko the
gorilla were found to be able to communicate with signs but lacked the formal
system that governed it. So we won’t belabor here that UG (which underlies
language) is human specific.
So what does UG consist of? First, UG consists of a finite set of abstract
features related to language from which individual languages may choose. For
example, Case and Tense are abstract features. Languages like Classical Latin and
Modern Turkish select for these features. Once selected, UG dictates that certain
things must “play out” in the language regarding the features – namely, that the
feature has to have an overt expression (e.g., a marking on a noun or noun phrase
in some way, marking on a verb in some way) and that the feature has to get
“checked” during sentence computation to make sure the sentence is grammatical.
Languages like Chinese do not select for Case or Tense. It is the selection from
this finite set of features that causes languages to look different from each other
while sometimes looking similar.

144
 At the same time, UG consists of a finite set of principles or constraints that
all languages must obey. One such example is called phrase structure. The
principle that governs phrase structure states that all languages must contain
phrases and that these phrases consist of a head and a complement. Familiar to
most readers is something like a noun phrase. In English professor of linguistics is
a noun phrase. Professor is a noun and is the head (which gives the phrase its
name – noun phrase) and of linguistics is the complement. Phrases are very
important in how languages work, in part because although we can hear and see
something like a noun phrase, there are hidden phrases called functional phrases
such as a tense phrase. Even though we can’t hear or see these phrases, they are
well motivated theoretically and empirically. These hidden functional phrases
play important roles when it comes to abstract features such as Case, Tense,
Number, and others. Because phrases consist of a head and a complement, only
two orders are permitted: head + complement and complement + head. These two
options are what make, say, English have one kind of typical word order and
Japanese another. Thus, in English prepositional phrases are head + complement
(as all phrases are in English) and we get in the kitchen. In Japanese, phrase order
is the opposite: complement + head. So the equivalent of in the kitchen in
Japanese is kitchen in. The same is true for the hidden functional phrases. There is
a functional phrase called a complementizer phrase, with complementizer
referring to the head of an entire sentence (where a complement like that, which,
or who would appear if the sentence were embedded, for example). Because
English is head + complement, we get that acquisition is complex (as in ‘I know
that acquisition is complex’) and because Japanese is complement + head, we get
acquisition complex is that.
So the nature of phrase structure constrains the shape of languages. All
languages must have phrase structure; this is a “law” of UG. And because there
are only two orders available, languages wind up being either head + complement
or complement + head. There are other principles and constraints that work like
“laws” in languages. In Chapter 5 we will review the overt pronoun constraint

145
when we talk about how subject pronouns work and, given our limited space, we
will not offer more examples here.
The third feature of UG is that it constrains and guides acquisition. It does
this because it sets limits on what languages can do. At every stage of language
acquisition, the mental representation for language is evolving. Because it is still
language at every stage of evolution, it must conform to whatever the contents of
UG are. Thus a stage of language acquisition may not invent a new abstract
feature but must select it from the inventory in UG. Likewise, at every stage, the
developing language must obey the principles of UG. This does not mean that
during child L1 acquisition the child’s mental representation is adult-like; it
means that the mental representation is language-like. It obeys the properties of
language that all human languages must obey.

Universal Grammar constrains and guides acquisition. This does not mean
that during L1 acquisition the child’s representation is adult-like; it means
the representation is language-like.

The claim in child L1 acquisition is that the starting point of L1 acquisition is

UG. Children are not blank slates but instead come equipped with the contents of
UG that subsequently guide and constrain the unconscious hypotheses that they
can make as language unfolds in their minds/brains. From the UG perspective, the
ability to learn a language successfully is independent of intelligence, personality,
and social environment. Of course, UG is not the entire picture. Children learn the
language they are exposed to and it is the input data they hear or see that pushes
the mental representation to select this feature or that feature and for this principle
or that principle to be triggered.

146
 Consider this …
A common misunderstanding about Universal Grammar is that it
guarantees acquisition. What is the difference between guaranteeing
acquisition and constraining it? Can you think of other areas where
something constrains or guides development but doesn’t guarantee it?
Note: This will become important later in this chapter when we discuss L2
acquisition and will be especially important in Chapter 5.

Against this backdrop of UG and its role in language acquisition we will

review the three possible scenarios listed at the beginning of this chapter, namely,
whether UG is the starting point as it is in L1 acquisition, whether the starting
point is the L1, or whether the starting point is “partial transfer” of the L1.

147
 Universal Grammar as the Initial State
There are two basic perspectives on UG as the initial state. We will begin with
what is called No Transfer.

148
 No Transfer
The first position about the initial state we will review is the Initial Hypothesis
of Syntax as developed by Christer Platzack in 1996. The basic claim of this
hypothesis is that the initial state of acquisition for L1 and L2 is the same; that is,
that both L1 and L2 learners begin the process of acquisition with just UG. Thus,
L2 learners go about the business of building a mental representation of the L2
based on UG and the input they are exposed to. There is no influence from the L1
at the outset (or along the way). What this means, then, is that a Japanese speaker
acquiring English as an L2 does not unconsciously assume at the outset that
English selects for Tense as a feature (even though both Japanese and English do)
and does not unconsciously assume that English is complement + head. Instead,
the learner starts out with all the features from UG available (including Tense)
and only assumes that the language has phrases. From the input data, the learner
will determine whether to select Tense or not for English and whether the
language is head + complement or complement + head.
The Initial Hypothesis of Syntax has not received a lot of attention in the L2
literature. Platzack provided some support for his hypothesis in looking at the
acquisition of word order. He provided some evidence that learners of an L1 in
which the inflected verb is required to be in “second position” as in Swedish do
not seem to transfer this to learning an L2 which also requires the same (e.g., a
language like German). In German, there is a hidden feature that requires that
inflected verbs move out of their underlying position and to a spot higher up in a
sentence in order to get that hidden feature checked. (We will see what “higher
up” means and looks like a bit later.) This is why Swedish and German get word
orders such as ‘Every day take the employees a break’ whereas in English we
would say something like ‘Every day the employees take a break.’ Take has
moved from its original position to a position higher up, which in serial order
means closer to the front of the sentence. This is called V2 for short because the
verb is in “second position” in the sentence. What Platzack argued was that data

149
show that learners who speak a V2 language don’t transfer V2 order right away
into learning another V2 language. Instead, they might assume something like
English, where the verb stays in its original spot. According to Platzack, this
means that learners are open to options (provided by UG) and don’t assume L1
values for features at the outset. Again, they are operating much like child L1
learners.
Another piece of evidence that Platzack cites is that L2 learners tend to use
the order negation + verb at the outset of acquisition even when their L2 order is
verb + negation. The prediction here is that a speaker of French with post-verbal
negation (e.g., Je n’aime pas ça ‘I don’t like that’ where pas is the negator) would
first produce sentences such as ‘(I) no like that.’ Platzack cites evidence from
learners of Swedish and German, in which negation is post-verbal; yet in their
early production of language, learners produce negation + verb-type structures. In
essence, what Platzack claims is that if there is no transfer of the L1 at the
beginning, the initial state must be UG. We will evaluate this hypothesis after we
review the next one.

150
 The Full Access Hypothesis
The Full Access Hypothesis was posited by Suzanne Flynn and her colleagues in
the early to mid 1990s. Although the authors never really claim that UG is the
initial state, their hypothesis functions in a similar fashion to the Initial
Hypothesis of Syntax. Under the Full Access Hypothesis, L2 mental
representations are constrained by UG at every stage of development. What is
more, the authors explicitly claim that the L1 cannot be the initial starting point;
that is, the learner does not transfer all of the L1 features and values of those
features at the outset of acquiring an L2. If this is the case, then by default UG
must be the initial state, especially if it constrains acquisition at every stage –
including the beginning stage. Flynn and her colleagues do not rely on the same
kind of evidence that Platzack does for his Initial Hypothesis of Syntax. They
conducted experiments with learners using tasks such as elicited imitation and
grammaticality judgments. In elicited imitation, learners are given sentences to
repeat, and presumably how they repeat those sentences is reflective of their
underlying grammars. The sentences are, of course, not easy two- or three-word
sentences but more complex sentences (e.g., ‘The happy janitor does not want the
new television’ and ‘Which secret message does the young girl find in the
basket?’). When learners repeat them with alterations, Flynn and her colleagues
check to see what those alterations represent and how different they are from
when children learning English try to imitate the sentences. They claim their
results suggest that learners aren’t transferring properties of the L1 into L2
acquisition but instead their alterations reflect basic UG constraints on sentence
structure. To be sure, Flynn and her colleagues rely on other evidence as well, but
the idea here is that where Platzack tends to rely on production data gathered from
various research projects, Flynn and her colleagues tend to rely on experimental
data.
It is fair to say that the position that UG alone is the initial state of L2
acquisition is not the dominant “paradigm” for thinking about how learners begin

151
the process of acquisition. First, not much evidence has emerged from this
perspective and much more evidence has emerged from other perspectives.
Second, some scholars have been critical of the methodologies used and the
interpretation of the data. Indeed, Platzack admits at various points that there are
other interpretations of some of the data he examines. A major criticism of the
Flynn et al. research is that they tend not to use beginning learners, which means
that the data may not be the best to talk about the initial state. We turn our
attention now to the opposite position about the initial state.

152
The L1 Is the Initial State

153
 Full Transfer/Full Access
The major paradigm from the generative perspective is what is called Full
Transfer/Full Access, launched by Bonnie Schwartz and Rex Sprouse in the
1990s. Adherents of this position basically argue that the initial state of L2
acquisition is the L1 mental representation. That is, learners begin by
unconsciously assuming that the L2 contains all the features, operations, and so
on that make up the L1. So English L1 learners of Japanese would assume that
Japanese is just like English. While they might correctly assume that Tense is a
feature in Japanese, they would incorrectly assume that Japanese is head +
complement or that Japanese has the feature that moves wh-elements out of their
original position and into a higher position in the sentence – which it doesn’t. In
Japanese you would say something like “Bill what wants?” Remember that
Japanese is complement + head. Instead of wants what? in Japanese we would say
what wants? because what is the object of the verb wants and wants is the head of
the verb phrase. As another example, Spanish-speaking learners of English would
assume that English has a feature that moves verbs out of their place of origin,
which is what gives word orders like ¿Qué quiere Pablo? ‘lit: What wants Pablo’
for ‘What does Pablo want?’ and Alessandro lee a menudo la ciencia ficción ‘lit:
Alessandro reads often science fiction’ for ‘Alessandro often reads science
fiction.’). On the other hand, English speakers of Spanish L2 would assume the
opposite, that verbs don’t move and the word order would be like English.
So full transfer means that the initial state is made up of the L1 mental
representation for the formal components of language (and perhaps for all of
language). Learners start with the L1 and basically have to “rewrite” or
“overwrite” what is there as they gain experience with the language. The Full
Access part of the name means that UG is fully available during L2 development
and both constrains and guides acquisition especially when the L2 input suggests
that the L2 is different from the L1. So the English speaker learning Spanish who
hears questions like ¿Qué quiere Pablo? over and over in the input gets crucial
data that verbs move and that there is an underlying feature causing this. That

154
feature is part of UG, and how it operates is subsequently constrained by the
content of UG. Traditionally, this restructuring (i.e., switching something from an
L1 “value” to an L2 “value”) is called parameter resetting. A parameter refers to
the consequences of a particular value of a feature (e.g., a strong Tense feature
forces verbs to move out of the verb phrase whereas a weak Tense feature does
not) or to the consequences of a particular principle (e.g., the parameters for
phrase structure are head + complement and complement + head). For the reader
unfamiliar with what we mean by “move out of the verb phrase” we offer this
rough description. Alessandro lee a menudo la ciencia ficción ‘Alessandro often
reads science fiction’ “starts out” as something like Alessandro a menudo lee la
ciencia ficción. The strong Tense feature of Spanish forces the verb out of its verb
phrase so that it winds up in the Tense phrase under the Tense feature itself. A
linguist would capture it this way, with details excluded: [T P [T ] [A D V P a menudo]
[V P Alessandro [V lee] [N P la ciencia ficción]]] → [T P Alessandro [T lee] [A D V P a
menudo] [V P – [V –] [N P la ciencia ficción]]]. The dashes indicate the empty spots
left behind by movement.

Full Transfer means that the initial state is made up of the L1 mental
representation for the formal components of language. Learners start with
the L1 and basically “overwrite” what is there.

The main argument for Full Transfer or for the L1 as the initial state comes
from the observation that at all levels of acquisition (beginner, intermediate,
advanced) there seems to be L1 influence on learner performance. If there is L1
influence at the intermediate and even advanced levels, it is assumed that the L1
influence was there at the beginning.
Of course, the best evidence for Full Transfer would be to see L1 influence
in the earliest stages of development. Few studies along these lines exist, but one
study from the 1990s has been influential. That study comes from Belma
Haznedar. She investigated a child L2 learner with Turkish as L1. For the first

155
two months that child lived in a Turkish-only environment in the United
Kingdom. However, during the third month the child began nursery school with
English-speaking children. Haznedar was able to study the child in those initial
stages of development. Two pieces of evidence are relevant for Full Transfer at
the outset and both are a result of “headedness.” Let’s recall that languages can be
head + complement (like English) or complement + head (like Japanese). Turkish
is a complement + head language. Like Japanese it has object + verb order instead
of verb + object order for verb phrases. For negative phrases Turkish has
complement + negator instead of negator + complement. The child in Haznedar’s
study began producing basic sentences following Turkish word order. Here are
some examples:

I something eating (SOV, like Turkish, instead of SVO, like English)

Finish no (verb + negator, like Turkish, instead of negator + verb, like

English)

These examples suggest the learner assumed that English was complement +
head like Turkish. However, it did not take long for the relevant input to push the
child toward English head + complement direction, and within a month, Haznedar
reports examples such as:

You eating apple (SVO order, like English)

I not eat cornflake (negator + verb, like English)

Although the data are not abundant in that Haznedar does not report lots and
lots of examples, what she does report is consistent in that the child used Turkish
word order 100 percent of the time at the outset. This kind of evidence is strongly
suggestive that the child unconsciously assumed that English would be like
Turkish.

Consider this …

156
 The child in Haznedar’s study was learning English in a non-instructed
environment. In addition, the child was a pre-schooler. Do you think it is
possible to get this kind of free and spontaneous data from adult
immigrants? What about beginning classroom learners of an L2? Do you
think it would be a good idea to try to get different kinds of data?

As stated above, the idea is that if there is L1 transfer evident in intermediate

and more advanced stages, then it is assumed that the L1 was the starting point
and that transfer has simply persisted. The vast majority of research regarding
Full Transfer looks at non-beginners. There is considerable research on this so we
will offer just one typical study and encourage the reader to consult additional
readings. Boping Yuan conducted a study that tested whether the L1 influences
acquisition. This study is summarized in the Exemplary Study. He focused on
learners of Chinese as L2. What is critical about his study is that he used learners
from two different L1s: Japanese and English. For the target structure Yuan was
interested in, these two languages were necessary as their L1s functioned
differently. The target structure in question was the reflexive pronoun ziji in L2
Chinese.
As the summary in this chapter indicates, Yuan’s results suggest that L1
influence was present in that it helped the Japanese speakers but hindered the
English speakers. Once again, because the influence is seen at both intermediate
and advanced levels, L1 transfer is assumed to be there at the beginning.
As suggested earlier, there are many more studies such as Yuan’s suggesting
L1 transfer than there are studies that support no transfer as outlined in the
previous section. In the next section, we turn our attention to a third possibility for
the initial state.

Consider this …

157
If the initial state of L2 acquisition is the L1, this clearly makes L2
acquisition different from L1 acquisition. In what way? Does this mean
that the underlying processes of acquisition are different? Does it mean
that the roles of input and output are different (see Chapter 3)? Or does it
simply mean the starting point is different but everything else might be the
same?

Exemplary Study

Yuan, B. (1998). Interpretation of binding and orientation of the Chinese

reflexive ziji by English and Japanese speakers. Second Language
Research, 14, 324–340.

Participants
81 participants

24 Japanese speakers learning Chinese as L2

57 English speakers learning Chinese as L2

24 Chinese native speakers who served as a control group

Location: China

Major Target Items

In English, reflexive pronouns can only be interpreted as referring to
the closest possible antecedent. By closest possible antecedent we mean
an antecedent that is in the same clause. In the sample sentence (1)
below, the reflexive himself can only refer to Bill, which is in the same
clause. It is barred from referring to John. In Chinese, the equivalent
pronoun can be interpreted as referring to either Bill or John; it is not
restricted to only antecedents in the same clause. This is shown in (2).

158
(Note that the subscript letters i and j indicate what items can refer to,
with an asterisk meaning that co-reference is not possible.) Crucially,
Japanese functions just like Chinese; the reflexive pronoun is not
restricted to taking an antecedent only in its own clause.

(1) Johnj said that Billi trusts himselfi/*j. (English)

(2) Johnj renwei Billi xiangxin zijii/j (Chinese)John think Bill trust
self‘John thinks Bill trusts himself.’ Or ‘John thinks Bill trusts
him.’

Material and Procedure

The participants were given a proficiency test and the English learners
were split into two groups: intermediate and advanced. The Japanese
group as a whole was intermediate and not split into two groups.

The participants took a paper-and-pencil interpretation test to determine

how they were interpreting the pronoun ziji. The test included sentences
that were pragmatically biased for one antecedent over the other. If
learners were following a purely syntactic constraint on interpretation
of the pronoun, they would not be led astray by the pragmatic bias.

Results
The English-speaking and Japanese-speaking groups treated ziji
differently, even when the groups were matched for proficiency. That
is, the intermediate Japanese learners were better at interpreting ziji than
were their English-speaking counterparts.

The L1 English groups tended to interpret ziji as referring to a local

antecedent, even in contexts that favored another antecedent further
away. The English learners with advanced proficiency were more likely

159
to accept a long-distance reading for ziji, but they did so less
consistently than the L1 Japanese group.

In most cases, the Japanese L1 group – even though it was only at the
intermediate level – behaved like the Chinese L1 control group. The
English speakers tended to be different from the Chinese L1 control
group.

Conclusion
Because of the difference between the English groups and the Japanese
group, Yuan concluded that transfer was the main variable. It slowed
down the English speakers but helped the Japanese speakers. Because
these results were seen at the intermediate level, Yuan concludes that
transfer was present from the beginning.

In spite of transfer, enough advanced English speakers showed

similarity to the Chinese L1 group that Yuan concluded that UG was
still operative in adult L2 learners.

160
 Partial Transfer of the L1 in the Initial State
Situated between no transfer in L2 acquisition and full transfer of all features and
formal elements of the L1 at the outset lies the idea of partial transfer. We will
examine two proposals that gained wide circulation in the 1990s.

161
 Minimal Trees Hypothesis
In a series of publications in the mid to late 1990s, Anne Vainikka and Martha
Young-Scholten launched what is called the Minimal Trees Hypothesis. The use
of the word trees is purposeful because the idea behind the hypothesis is related to
how linguists visualize syntax as “syntactic trees.” To give you an example, here
is a basic sketch of any simple sentence (with much detail removed) as a linguist
would see it.

First, the letters refer to phrases and their elements. VP stands for verb
phrase and inside this VP we see its head (V) and its complement (XP). XP is a
shorthand way of saying various different kinds of phrases could be the
complement of V (a noun phrase, a prepositional phrase, even another sentence,
which would be a CP). Things such as VP, NP, PP, and so on are what linguists
call lexical phrases because the heads of the phrases are actual lexical items such
as verbs, nouns, prepositions, and so on. Above the VP are functional phrases.
These are the “hidden” parts of sentences where syntactic information is encoded
in the form of abstract features. TP stands for tense phrase (the head is a T
“element” and its complement is the VP). CP stands for complementizer phrase,

162
which basically means “clause/sentence”. C is the head of the phrase and TP is its
complement. The reader will note that there is also a Spec embedded in each
head. This abbreviation stands for specifier and, for right now, we will ignore
what it is. What the reader should see so far is that a sentence contains lexical
phrases and functional phrases. Something else the reader should note is that
phrases exist in a hierarchical order and are “stacked.” That is, sentences do not
consist of linear ordering (although we may hear and read them that way) but
instead have a hidden internal structure in which phrases build up, one upon the
other.
In syntactic theory, functional phrases may motivate structural operations
depending on the features embedded in them. For example, in languages like
Spanish and English, questions have a strong Q feature residing in the CP. This
motivates the movement of a wh-element up into the CP as illustrated below.
Languages like Chinese and Japanese lack a strong Q feature and so the wh-
elements don’t move out of their original positions.

163
 What Vainikka and Young-Scholten have argued is that learners transfer
only the lexical categories from the L1 at the outset, but not the functional
categories. What this means, then, is that learners begin L2 acquisition only with
VPs, NPs, PPs, and so on but not CPs, TPs, and other functional phrases. Their
underlying syntactic trees are minimal, hence the name. The sentence above
would be represented as something like this under the Minimal Trees Hypothesis:

Indeed, beginning L2 learners of English produce sentences such as this one,

suggesting that there is no movement and there are no functional categories (note
the lack of is for is reading because there is no TP in which the is can be
inserted).
Vainikka and Young-Scholten’s initial support for Minimal Trees came from
research on various groups of L1 learners acquiring German as an L2 in Germany.

164
Those L1s included Turkish, Korean, Spanish, and Italian. What the researchers
keyed in on was what we have seen repeatedly in this chapter: the order of
elements in a phrase. In particular, the order of heads and complements. German
is a complement + head language, as are Korean and Turkish. Spanish and Italian
are head + complement languages. Thus, Korean and Turkish basic word order for
verb phrases is object + verb whereas for Spanish and Italian it is verb + object.
Vainikka and Young-Scholten collected various kinds of oral data and noted that
the Turkish and Korean speakers readily produced OV order in German from the
outset while the Spanish and Italian speakers produced VO order and only later
switched to OV. What is more, they noticed an absence of functional words in
their speech (e.g., auxiliaries, modals) and a lack of wh-movement, suggesting
that functional phrases and features were not there, even though in Spanish and
Italian such features are. Their conclusion was that the lexical phrases transfer at
the beginning of L2 acquisition but not the functional phrases.

The Minimal Trees Hypothesis claims that lexical categories and phrases
transfer from the L1 at the outset but functional features and phrases do
not.

Consider this …
The absence of auxiliaries and modals and a lack of wh-movement suggest
to Vainikka and Young-Scholten that there is no CP in their participants’
early representations. But we know that with yes/no questions, do winds
up where a wh-element would go. Example: He wants to go? → Does he
want to go? Suppose an early-stage learner produces something like Do he
want go store? Would you take this as evidence that the learner has a CP
and therefore Minimal Trees is wrong? Hint: Consider looking at all the
instances of do used by the learner to see exactly how they are used, e.g.,
Do he can go? Do you like? Do I can eat this one?

165
 The Minimal Trees Hypothesis is appealing given the data examined, but as
a number of scholars have pointed out, we cannot rely exclusively on oral
production data. Why is this? It is possible that learners may know more than
what they can do. That is, their production does not necessarily match their
underlying competence. This is a given in the child L1 literature, in which
children are sometimes shown to have underlying abstract features in their syntax
that are not realized in their speech. The same is certainly possible in L2
acquisition. Just because a learner doesn’t produce, say, an auxiliary or show
evidence of wh-movement does not necessarily mean that TP and CP are absent
from their underlying competence or that they haven’t transferred the features
associated with these phrases from their L1 into the L2 at the outset. It could be,
for example, that learners just haven’t heard enough auxiliaries and modals for
them to be part of their lexicon. There are other methodological issues with how
Vainikka and Young-Scholten determined whether something is present or absent
in learner speech but we will not belabor those here. Readings and references at
the end of the chapter will allow the reader to explore these and other topics in
this chapter in more detail.

166
 Valueless Features Hypothesis
The Valueless Features Hypothesis, also called “weak transfer,” was proposed
by Lynn Eubank around the same time as the Minimal Trees Hypothesis. Unlike
the Minimal Trees Hypothesis, the Valueless Features Hypothesis claims that L1
lexical and functional categories are indeed transferred to the L2 and they are
present in the earliest mental representations of learners. However, despite the
fact that the functional categories are available, the strength of the features
associated with L1 functional categories does not transfer. In short, the functional
phrases are hanging around in the syntax with little to do because they are, in
essence, devoid of anything that would cause syntactic operations involving
movement (see above with the wh-example and the strong Q feature).
Curiously, Eubank claims that a consequence of Valueless Features is that
learners may sometimes show evidence of L2 word order and sometimes evidence
of L1 word order. Given that nothing is forcing compliance with L2 word order
and nothing is forcing compliance with L1 order in the functional phrases, then
learners may produce whatever they wish, in a sense. So as not to rely on
production data alone, Eubank tested the hypotheses derived from Valueless
Features in several studies involving sentence matching and truth-value. In
sentence matching, participants read a pair of sentences and have to determine if
they match or not. Response times are tracked and recorded. Critical are the
response times comparing pairings of grammatical sentences and pairings of
ungrammatical sentences. It should take longer to match the ungrammatical
sentences if an underlying representation prohibits such sentences. Here are
sample pairings from this kind of study:

Paired grammatical: The woman often loses the books.

The woman often loses the books.

Paired ungrammatical: The woman finds sometimes the

pencils.

167
 The woman finds sometimes the
pencils.

In truth-value tasks, learners read a scenario and then determine whether the
sentence that follows logically flows from what they just read. Here is one
example from a Eubank study.

Tom likes to draw pictures of monkeys in the zoo. He likes his pictures to be
perfect, so he always draws them slowly and carefully. All the monkeys
jump up and down really fast.
True or false? Tom draws slowly jumping monkeys.

The target sentence can be read one of two ways: that the monkeys are
jumping slowly (which is the typical English reading) or that Tom draws slowly
and what he draws are jumping monkeys (i.e., slowly modifies what Tom does
and not how the monkeys jump). The latter might be a reading if your L1 is
Spanish or French, for example, and you transferred verb movement from your L1
(e.g., Spanish and French both allow sentences like “Tom drinks slowly his
coffee” whereas English allows “Tom slowly drinks his coffee”).
What Eubank found out in a series of experimental studies was that the
predictions of the Valueless Features Hypothesis did not hold up. At the very
least, the data are not conclusive but, as some scholars have suggested, they tend
to point to disproving the hypothesis rather than supporting it. For example, in the
sentence-matching task, while the native speakers of English performed as
expected, the L2 learners did not and their response patterns did not coincide with
what the hypothesis would predict.
Not much work is done any longer with the Valueless Features Hypothesis
for a variety of reasons. It is safe to say that as of the writing of this book, it is a
proposal that has not received strong empirical support and the methodology
associated with it has been heavily critiqued. We no longer see many references to
this hypothesis but we have included it here for historical reasons.

168
 Quick Recap
Before continuing, we should recap where the generative field is now in terms of
the three positions presented. Both the UG-only and the Partial Transfer positions
have tended to be overwhelmed by the bulk of evidence suggesting there is L1
transfer all along the way. In other words, the dominant generative position today
is that the initial state is the L1. In a 2016 publication, for example, Roumyana
Slabakova says that there is almost universal consensus that although UG is
operative in L2 acquisition as it is in L1 acquisition, in the L2 context, learners
begin acquisition with an L1. One would be hard pressed to find a publication at
this date that takes a non-L1 transfer perspective within the generative tradition.
This does not mean that there are not some who hold on to the other two
positions, but, as we have suggested, they now form a very small minority of
scholars. Within the generative tradition, most scholars simply assume L1 transfer
from the outset and we no longer see research testing competing proposals such as
the three different perspectives just presented.

169
Consider this …
Although the evidence strongly tips in favor of L1 as the starting point, to
what extent might this be an artifact of trying to get learners to show
something early on? Imagine you are asked to perform a task after two
months of being in an L2 environment. That task could be a
grammaticality judgment, a truth-value test, an oral production task. How
do you do the task if what it asks you to do is beyond your current
underlying mental representation? Could it be that we use our L1 to
perform tasks when our L2 is not up to snuff?

170
 Other Perspectives
In this chapter we have taken a generative approach to the question of the initial
state largely because this field originally motivated the question and has tended to
dominate the discussion. However, there are several other approaches that we will
review briefly so that the reader understands that an initial state may refer to
something other than underlying mental representation.
In usage-based models of acquisition in which there is no UG and no special
mechanism(s) for language, it is widely assumed that there is L1 influence from
the beginning. Because usage-based models tend not to talk about mental
representation but underlying connections that are related to performance (e.g.,
the presence of the is strongly connected to something “nouny” as in the dog and
not strongly connected to something “prepositional” as in *the with), they differ
from a generative perspective in terms of what might get transferred. Currently,
the predominant position in usage-based models is that the L1 connections are
present from the beginning of L2 acquisition and may actually block new
connections from forming (or might help, depending on the L1/L2 pair). Nick
Ellis and Stefanie Wulff, for example, discuss this in a 2015 publication in which
they say that the initial state for L2 acquisition is not a “plastic system” as it is for
L1 acquisition. Instead it is a system that is “tuned and committed” to whatever
the L1 is. So even though usage-based models are predicated on non-linguistic
learning mechanisms, the consensus among the researchers is that the initial state
is the L1. In this framework, the initial state is the L1 plus general learning
mechanisms that are involved in everything from language acquisition to learning
what a dog is and how it isn’t a cat.
Another perspective on the initial state is Input Processing as articulated by
Bill VanPatten in a series of publications from the 1990s until the present day.
Input Processing is less concerned with underlying representations at the outset
and instead is concerned with how learners process input sentences during

171
comprehension. Theoretically, there are three choices (much like above): (1) the
learner uses a set of universal processing strategies and procedures that are not
derived from the L1; (2) the learner uses well-worn L1-based processing
strategies to try to comprehend the L2 from the outset; (3) the learner has some
universal and some L1-based processing strategies that work during
comprehension. The third position is not widely accepted and in fact it is almost
impossible to find any research or theory on it so we won’t dwell on it here.
VanPatten has taken the position that all learners, regardless of L1, possess a set
of universal processing strategies they take to the task and that it is not an “L1
processor” that attempts to compute sentence structure during comprehension. To
be sure, these universal strategies have nothing to do with Universal Grammar.
For example, one such processing strategy is that learners come to the task
knowing that words exist and that there are ways to “label” meaning with
language. A consequence of this, according to Input Processing, is that at the
earliest stages, learners search for content lexical items to try to comprehend what
they hear and they may “skip” function words and other small things in the input
for the time being. At the same time, they rely on what is called the first-noun
strategy (or principle), which says that as long as there is no contextual or
semantic evidence to the contrary, learners assume the first noun or pronoun they
encounter is the subject/agent of the sentence. So let’s imagine the learner hears a
simple sentence such as “The dog is chasing the cat.” The learner knows that
someone is talking about a dog and cat and what they’re doing because it’s
happening in front of the learner or there is a picture or some other way the
learner connects the input string to some event. At the earliest stages of
acquisition, the learner may process only “dog,” “cat,” and “chasing.” Now let’s
imagine that learner knows these words but hears a new sentence: “The dog was
chased by the cat.” Here the learner once again processes “dog,” “chase,” and
“cat” and assumes the same meaning as the first sentence. This is because the
learner assumes that “dog” is both subject and agent. Words such as “was” and
“by,” for example, aren’t processed – meaning the learner doesn’t link them to

172
any meaning or function. They are just “noise.” This would happen even if the L1
and the L2 both have the same sentence structure for a passive sentence such as
“The dog was chased by the cat.”
What the model of Input Processing predicts, then, is that the input
processors filter out data from the input and deliver structural information to the
internal mechanisms that build representation (e.g., UG). The way the filtering
happens is universal. Some scholars disagree with this position, believing that the
L1 processor is active at the outset. VanPatten has recently suggested that the L1
processor can “kick in” at some points during acquisition but that at the outset, it
is the universal set of strategies that learners take to the task of comprehending
input at the very beginning.

Input Processing predicts that the input processors filter out data from the
input. The way the filtering happens is universal.

Yet another perspective on the initial state can be found in Processability

Theory. Processability Theory was formulated by Manfred Pienemann and is a
theory not about input processing or what is in the mental representation but
instead about universal constraints on how learners process output. In other
words, how learners put words together in real time to make sentences. The
theory claims that output processing starts with and is constrained by a universal
set of output processing procedures that are hierarchically ordered. What this
means is that learners must acquire one processing procedure before they can
acquire another processing procedure, which in turn they must acquire before
acquiring yet another processing procedure. The theory is much more complex
than this, but for now we will illustrate it with just a subset of these procedures.
Central to Processability Theory is the idea that when stringing sentences
together, there may need to be an exchange of grammatical information. A classic
example is subject–verb agreement for languages that have such agreement. If a
subject is, say, 2nd-person plural, that information has to be exchanged with the

173
verb so that the verb is also 2nd-person plural. Another example is noun–adjective
agreement for languages that have this agreement. If a noun is feminine and
singular then the adjective that goes with that noun must also be feminine singular
(i.e., the grammatical information needs to be exchanged between noun and
adjective). Also central to the theory is the idea of syntactic boundaries. The
exchange of grammatical information is easiest when there is no information to
exchange, while the most difficult is when grammatical information has to be
exchanged across sentence or clause boundaries, with various difficulties in
between depending on whether the grammatical information has to cross a
boundary or not. In English, producing “the rat” is not a problem because no
information is exchanged between the determiner and rat. However, “The rat
hates cheese” requires grammatical information to be exchanged between the
noun phrase the rat [3rd-person singular] and the verb in the verb phrase. This
involves crossing a syntactic boundary.
Now let’s look at an example from Spanish that helps us understand the
boundary and distance issue a bit better. In Spanish, nouns and adjectives have to
agree in number and gender. A single word such as casas ‘houses’ does not
exchange grammatical information with anything if it’s not part of a sentence.
Now let’s put it into a noun phrase with an adjective: casas blancas ‘white
houses.’ Here grammatical information has to be exchanged between a noun and
an adjective but the information remains in the same phrase, the NP. Now here’s
another sentence: casas son blancas ‘houses are white.’ Here the grammatical
information has to travel outside the NP over to the VP. In these three examples
we see three distinct grammatical information procedures:

plain word: no information exchanged

within a phrase: information is exchanged between two words in the same

phrase

across a phrase boundary: information is exchanged between two words in

two different syntactic phrases.

174
 Processability Theory claims that there is a universal and hierarchically
ordered set of output processing procedures that are acquired over time
and not influenced by the L1. Instead, they constrain the influence of the
L1.

Processability Theory says that these three procedures (part of a set of seven)
are ordered as we see them from easiest to more difficult. In turn, this means
learners will need to develop these procedures in the order in which we see them.
They are universal and not influenced by the L1. Whether you are an English
speaker learning Spanish (English does not have adjective agreement) or a
speaker of French learning Spanish (French does have such agreement), you will
have to progress through the development of the procedures in the same way.
However, and this is important, this does not mean that there is no transfer. What
it means is that L1 transfer is constrained in that things can transfer only when a
learner is developmentally ready for those things to transfer. A learner with
French as L1 acquiring Spanish as L2 can only transfer agreement between a
noun and an adjective across an NP into a VP after that learner has first been able
to do the same inside the NP. Pienemann has suggested that when such transfer
does happen, it may accelerate or benefit acquisition. But again, a learner cannot
simply transfer all the procedures of the L1 from the outset. In a sense, transfer
may be “triggered” at points when there is a match between what the learner
creates based on universal procedures and what the L1 looks like.
We have been brief in this section to demonstrate other non-generative and
non-representational approaches to the nature of the initial state. In the first –
usage-based – we see something that resembles full transfer. The other two, both
related to processing but one focused on input processing and the other focused
on output processing, do not posit the L1 as a starting point. Instead, L1
processing procedures or strategies may be triggered at particular times during
development.

175
Consider this …
We have seen perspectives on the initial state that are related to
underlying representation. That would be in the generative tradition. We
have also seen three perspectives outside of the generative tradition:
usage-based, input processing, and output processing. Is it possible that
some of these perspectives are compatible? For example, could it be that
learners’ internal representation begins with an assumption that the L2 is
like the L1 but learners process input using universal strategies? Consider
the various combinations.

176
 Recap
Here are some major ideas and topics covered in this chapter.

The generative tradition has led the “debate” on the initial state. There are
three positions that have emerged within this framework:

◦ the initial state is UG by itself

◦ the initial state is the L1 (with UG still constraining and guiding)

◦ the initial state includes only partial, non-functional aspects of the L1

(with UG still constraining and guiding).

Of the three positions, the one with the most empirical research behind it
is the second: The initial state is the L1 plus UG.

There are other frameworks that make claims about the initial state:

◦ usage-based approaches converge on the idea that L2 learners come to

the task with an L1 system as the starting point

◦ input processing posits that independent of underlying representations,

learners begin input processing at the outset with universal strategies for
comprehending sentences

◦ Processability Theory focuses on output processing and claims that L1

transfer is constrained by universal output processing procedures and
that the L1 thus is not the starting point for how learners develop the
ability to string words together to make sentences.

What about Social Factors?

As the reader might guess, the issue of the initial state for L2 acquisition is
viewed as an internal, cognitive/linguistic matter. That is, the focus is on
formal properties of language and how and when they transfer, if at all, as

177
well as the presence of something like UG at the outset. But there is a role
for social factors in how the learner approaches the entire process of L2
acquisition. Let’s consider the question of learner identity, for example,
and that learner’s perceived relationship to the world around him or her.
Elizabeth Miller and Ryuko Kubota, for example, discuss the issue of how
learners construct a “second language identity.” The question becomes,
how do they do this and what is the starting point for the construction of
that identity? Do learners begin the task of identity construction with an
“open mind,” a sort of socio-personality version of UG? Or do they begin
identity construction based on their own L1 community (or communities)
identity? Identity construction is as complex as acquisition of language
itself, and scholars like Miller and Kubota review scholarship on
everything from social, to cultural, to political aspects of identity. Other
scholars speak of something called socialization – defined largely as
integration into a particular society or social group. How do learners begin
this process? These are good questions but most research in this area
focuses on factors and contexts that affect identity and socialization. So is
it possible to think of an “initial state” for socialization and identity
construction?

178
 References and Readings

Dube, B. (2000). Where are the minimal trees? Evidence from early Zulu second
language subordination. Second Language Research, 16, 233–265.

Ellis, N. & Wulff, S. (2015). Usage-based approaches to SLA. In B. VanPatten &

J. Williams (Eds.), Theories in second language acquisition (pp. 74–93). New
York: Routledge.

Eubank, L. (1993). Sentence matching and processing in second language

development. Second Language Research, 9, 253–280.

Eubank, L. & Grace, T. (1998). V-to-I and inflection in non-native grammars. In

M.-L. Beck (Ed.), Morphology and its interfaces in second language knowledge
(pp. 69–88). Amsterdam: John Benjamins.

Hawkins, R. (2001). Second language syntax: A generative introduction. London:

Blackwell.

Haznedar, B. (1997). Second language acquisition by a Turkish-speaking child:

Evidence for L1 influence. In E. Hughes, M. Hughes, & A. Greenhill (Eds.),
Proceedings of the 21st annual Boston University conference on language
development (pp. 245–256). Somerville, MA: Cascadilla Press.

Miller, E. R. & Kubota, R. (2013). Second language identity construction. In J.

Herschensohn & M. Young-Scholten (Eds.), The Cambridge handbook of second
language acquisition (pp. 230–250). Cambridge: Cambridge University Press.

Pienemann, M. & Kessler, J.-U. (Eds.) (2011). Studying processability theory.

Amsterdam: John Benjamins.

Platzack, C. (1996). The initial hypothesis of syntax: A minimalist perspective on

language acquisition and attrition. In H. Clahsen (Ed.), Generative perspectives

179
on language acquisition: Empirical findings, theoretical considerations,
crosslinguistic comparisons (pp. 369–414). Amsterdam: John Benjamins.

Radford, A. (2004). Minimalist syntax: Exploring the structure of English.

Cambridge: Cambridge University Press.

Schwartz, B., & Sprouse, R. (1996). Second language cognitive states and the full
transfer / full access model. Second Language Research, 12, 40–72.

Slabakova, R. (2016). Second language acquisition. Oxford: Oxford University

Press.

Vainikka, A. & Young-Scholten, M. (1994). Direct access to X´-theory: Evidence

from Korean and Turkish adults learning German. In T. Hoekstra & B. Schwartz
(Eds.), Language acquisition studies in generative grammar: Papers in honor of
Kenneth Wexler from the 1991 GLOW Workshops (pp. 265–316). Amsterdam:
John Benjamins.

Vainikka, A. & Young-Scholten, M. (1996). Gradual development of second

language phrase structure. Second Language Research, 12, 7–39.

VanPatten, B. (2015). Input processing in adult SLA. In B. VanPatten & J.

Williams (Eds.), Theories in second language acquisition (pp. 113–134). New
York: Routledge.

White, L. (2003). Second language acquisition and universal grammar.

Cambridge: Cambridge University Press.

Yuan, B. (1998). Interpretation of binding and orientation of the Chinese reflexive

ziji by English and Japanese speakers. Second Language Research, 14, 324–340.

180
 Following Up
1. Explain in your own words what the three positions are on the initial state
for L2 acquisition.

2. Before the contemporary era of linguistics and acquisition research,

children were thought to come into the world with a “blank slate.” Do some
internet and bibliographic research on this term and the debate surrounding
it. You might wish to consider looking at Steven Pinker’s book published in
2003 titled The Blank Slate: The Modern Denial of Human Nature.

3. Most non-researchers (including teachers) would argue that “Of course,

the L1 influences L2 acquisition and probably does so from the outset.” (1)
What do you think such people mean and what kind of examples would they
offer? (2) Based on what you’ve read, what might you tell them?

4. In the 1980s, Bill VanPatten established the first stage of copular (linking
verbs) in Spanish to be an absence of copular verbs. (Spanish has two verbs
to mean ‘be’: ser and estar.) In the earliest stages of acquisition, VanPatten
reported utterances such as Juan alto ‘John tall’ and Juan no aquí ‘John no
here’ for Juan es alto and Juan no está aquí, respectively. These data came
from classroom learners of Spanish with English as L1. Knowing that
English has the word be and that it is required (e.g., John is tall/*John tall),
do you think these data speak against L1 transfer at the outset? Why do you
think the learners in this early stage don’t have a copular verb in Spanish L2
when they have one in English L2? Note: Since VanPatten’s early research,
the phenomenon has been found to be widespread and also with speakers of
Chinese as L1 who are learning Spanish. Chinese has no copular verb
equivalent to ‘be.’

5. Under the Full Transfer approach, what does it really mean to have the L1
as the initial state? Under this proposal, learners have to “overwrite” what’s

181
in the L1 to build an L2 representation. What does this actually mean?
Clearly, the L1 does not disappear, so the learner doesn’t really rewrite the
L1. Does the L2 learner “create a copy” of the L1 and write that over? Can
you think of some other way to talk about this idea?

6. If we assume that the L1 is the starting point for learners, should this
assumption inform teaching materials in any way? Why or why not? Note:
You may want to come back and revisit this question after Chapter 6.

182
 5
Can L2 Learners Become Nativelike?
◈

In this chapter, we consider whether adult L2 learners can achieve the same
knowledge and facility with an L2 that native speakers of that language have. At
first blush, the answer to this question seems obvious: No. After all, conventional
wisdom and experience both tell us that people who learn an L2 as adults still
speak with accents and make grammatical “mistakes.” This is true even of people
who function at a professional level and live their adult lives in the country where
the language is spoken. We also assume that the younger people start to learn an
L2, the better their odds of becoming nativelike. While this experiential evidence
is important, it’s not the whole story. Consider the language classroom experience
of one of the authors of this book for a minute. In high school, Megan studied
both French and Japanese. Her French teacher was a native French speaker. Her
Japanese teacher was a native English speaker who had worked in Japan for
several years, married a Japanese woman, and, at the time, was finishing a Ph.D.
in Japanese literature. Both of these teachers were qualified to teach the language
classes, but one was a native speaker, and one was not. That we hire non-native
speakers as language teachers suggests that at least some adults can acquire an L2.
Similarly, you may have had professors who, though not being native English
speakers, live, work, and generally manage their professional lives in English.
That means that they have acquired an English proficiency that allows them to
handle all of the tasks that native English speakers do. Anecdotally, then, we have
two answers to the question of whether L2 learners can become nativelike. On the

183
one hand, we have some evidence that the answer is no, and that the age at which
someone starts acquiring a language has a relationship with their final proficiency
level. On the other hand, we also have evidence that some people manage to
acquire an L2 well enough to live and work in that language. We’ll come back to
these observations at the end of the chapter, but before we do, we’ll first define
“nativelikeness.” We’ll also discuss three possible answers to the question of
whether L2 learners can become nativelike: (1) L2 learners cannot become
nativelike, (2) L2 learners can become nativelike in some domains but not others,
and (3) nativelikeness is possible.

184
 What Is Nativelikeness?
In Chapter 1, we suggested that the question behind all of the questions in the
field of SLA is whether L1 acquisition and L2 acquisition are the same thing.
Ultimate attainment – basically, how far in language acquisition learners get – is
no different. If we had definitive evidence that learners can’t become nativelike,
that would suggest a major difference between L1 and L2 acquisition. So we need
to take a step back and think about what nativelikeness is. Nativelikeness
typically refers to the implicit representation of a language that someone who has
grown up speaking that language has. This includes pronunciation, vocabulary,
and what most of us would call “grammar.” In short, adult native speakers know
what’s possible in the target language. They also come to know what is
impossible in the target language. To see this, let’s consider the sentences in (1).
Sentences (1a) and (1c) are declarative sentences, and sentences (1b) and (1d) are
questions based on those declarative sentences.

(1)

a. John said Bill left.

b. Who did John say left?

c. John wondered whether Bill left.

d. Who did John wonder whether left?

Does (1b) sound better – more like English – to you than (1d)? If you’re a
native English speaker (and even if you’re not a native English speaker), you
should have had a little bit of a “what?!” moment when you read (1d). This is
because your underlying English competence tells you that (1d) is not a possible
English sentence. Given that (1d) is formed in exactly the same way as (1b), the
impossibility of the sentence is surprising. The precise reasons why a sentence
like (1d) is impossible are beyond the scope of this book, but the general point is
that part of an adult’s steady-state mental representation of language (i.e., a

185
representation that is no longer in development) includes this kind of knowledge.
That is, not only do you know what the grammar of your language allows, you
also know what is not permitted in the grammar – and it may not be clear where
this knowledge comes from.
This mental representation for language is what people usually have in mind
when they ask whether someone knows a language. Researchers and language
teachers are concerned with the formal properties of language: for example,
pronunciation, grammar, and vocabulary. The mental system for language is
really important, but it’s not the only thing that is necessary for communication to
be successful. Part of growing up in a culture means learning how to use your
language in ways that communicate in that culture. This includes, but is not
limited to, knowing how to start and participate in a conversation, knowing how
to make requests, and knowing how to shift between informal and formal modes
of speech. It also includes learning how to read and write. These skills can
develop unevenly, so that people are capable in one domain but not another
domain. That adult native speakers of a language may have different types of
linguistic skills is important for how we conceive of nativelikeness. Does it mean
being able to do everything with language that an adult native speaker who works
in a professional job can do? Or does it mean something else? If we consider
educated adults our benchmark for nativelikeness, then we conflate underlying
linguistic knowledge with linguistic skill (communication). It’s important that we
keep them separate – communicative skills depend on having the corresponding
linguistic knowledge, and these skills develop in the communicative contexts in
which they are used. This isn’t to say that communicative skill isn’t important for
L2 learners – being able to write a professional email is very important in work
contexts, for instance – but it means that skill development is secondary to our
concern in this chapter. We are primarily concerned with the development of a
mental representation for another language. So we will define nativelikeness as
having a mental representation for another language that more or less matches that
of adult native speakers of that language. This means that we will focus on the

186
underlying and implicit (unconscious) knowledge that non-native speakers
acquire and the processes by which they acquire that knowledge.

Nativelikeness means having a mental representation for the target

language that more or less matches that of adult native speakers of that
language.

187
Consider this …
It is very common for children acquiring English as their native language
to go through a period of time in which they overgeneralize regular verb
endings to irregular verbs, so that went becomes goed/wented and brought
becomes bringed in the child’s speech. Similarly, one of the authors of
this book has a friend whose 3-year-old daughter couldn’t say the sp-
cluster and turned words like spoon and Sparty (the name of the author’s
dog) into words like [sun] and [sarti]. This is not an adult system for
English. Does this mean that the child is not a native English speaker?
Why or why not? Given that these non-adult-like utterances are normal in
child language development, what does it mean if we see similar patterns
in adult L2 development? Is there a point in adult development when non-
nativelike forms would be more problematic? What point would that be?

188
 Second Language Acquisition
Now that we’ve defined nativelike as having a mental representation for the target
language that more or less matches that of native speakers, we can consider three
possible answers to the question of whether L2 learners can become nativelike.
The first possibility is that L2 learners cannot become nativelike. The second
possibility is that nativelikeness is possible in all domains of the L2 system, and
the third possibility is that nativelikeness is possible in some domains, but not in
others. We will explore each of these possibilities below. Although learners need
to be able to put their linguistic competence to use in real-world situations, again
we will limit our discussion in this chapter to the development of L2 mental
representation. There are two reasons for this: First, most of the major hypotheses
and questions in this area deal with underlying knowledge or mental
representation only, and second, communicative skill depends, in part, on
linguistic knowledge.

189
 L2 Learners Cannot Become Nativelike
This position is fairly self-explanatory. The main claim is that the processes – and
therefore the outcomes – of L2 acquisition are different from those of L1
acquisition, and therefore L2 learners cannot acquire a mental representation for
the target language that matches that of the native speakers of that language. In
this section, we discuss two major proposals: the Fundamental Difference
Hypothesis and the Critical Period Hypothesis.

The Fundamental Difference Hypothesis

The Fundamental Difference Hypothesis (FDH) was proposed by Robert Bley-
Vroman in two articles, one published in 1989 and the second published in 1990.
Bley-Vroman argued that, because of observed differences between child L1
acquisition and adult L2 acquisition, the two processes – and therefore the two
outcomes – are distinct from each other. Bley-Vroman took the standard
generative position that child language acquisition is the result of Universal
Grammar and domain-specific learning mechanisms. These mechanisms are
specific to language, and children use them to construct a mental representation
for their L1 using input (see Chapter 3) as their data. Both Universal Grammar
and these learning mechanisms push the child to construct an L1 mental
representation that more or less matches that of adults in their community. Given
this account of L1 acquisition, Bley-Vroman builds his argument for a
fundamental difference between L1 and L2 acquisition on four key observations
about adult L2 acquisition.1
The first of these observations is that adult L2 acquisition is much more
variable than child L1 acquisition. L2 acquisition is variable in several ways.
First, most adults do not achieve high proficiency levels in their L2 and they fall
on a broad range of levels (i.e., some stop learning and are intermediate-like while
some go on and are advanced-like). Second, even relatively accomplished adult
L2 learners do not seem to have the same kind of mental representation that native

190
speakers do. L2 pronunciation seems to be particularly prone to divergence from
native-speaker norms. Third, adults, unlike children, adopt different strategies and
goals in language learning, and seem to be content with that. Some people take
several weeks of language classes in preparation for a trip overseas; other people
learn diplomatic and political language but would have a hard time ordering a
beer or reading a children’s book. None of these observations is characteristic of
child L1 acquisition.
The second observation is that adult L2 acquisition often seems to fossilize.
Fossilization happens when a specific form or structure in the L2 is only partially
acquired and does not seem to become more target-like over time. For example, a
learner could fossilize with respect to English past-tense marking in that they do
not consistently supply a past-tense verb in required contexts. This is, of course,
in contrast to native speakers who, if they fail to supply the appropriate form of
the verb, are assumed to have had performance problems but are not assumed to
have a deficient mental representation (i.e., they made a momentary mistake but
their underlying system is intact). Fossilization is not a construct in child L1
acquisition.
The third observation is that adult L2 mental representations are often
marked by indeterminate intuitions. Indeterminate intuitions refers to the
relative strength of the unconscious knowledge people have about the language
that they speak. Let’s start with native speakers’ intuitions. Native speakers tend
to have pretty strong intuitions about what is grammatical and what is
ungrammatical in their language. To see this, take a minute and indicate whether
the following sentences are possible, slightly strange, or impossible. When you do
this, think about what you might say or have heard other people say – don’t worry
about whether the sentence conforms to rules from an English class.

(2)

a. David likes to talk about politics.

b. What does David like to talk about?

191
 c. About what does David like to talk?

d. Sally turned the television on.

e. What did Sally turn on?

f. On what did Sally turn?

Native English speakers usually rate the sentences in (2) as follows: (2a),
(2b), (2d), and (2e) are totally possible and sound like good English sentences.
Judgments might differ for (2c); at the very least, it should sound less natural than
(2b) but it is generally accepted as possible. The example in (2f), however, should
sound very strange – and, any interpretation you give it is not related to (2d).
Regardless of what your judgments were, the knowledge that you relied on to
make them is part of your mental representation for English. Bley-Vroman’s
argument is that non-native speakers’ intuitions about their L2 aren’t as strong as
native speakers’ intuitions. Non-native speakers may not have a strong feeling one
way or another. Or, they might judge something grammatical in one instance and
ungrammatical in another instance. This is why these intuitions are called
indeterminate. Bley-Vroman argues that indeterminate intuitions reflect clear
differences in the type of mental representations that native and non-native
speakers have. Bley-Vroman claimed that because native speakers typically have
clear-cut intuitions about grammaticality, indeterminate intuitions indicate that
non-native speakers have a different kind of mental representation than native
speakers.
The fourth observation is that L2 acquisition often involves explicit
instruction and often includes negative evidence. Negative evidence is
information about what is not possible in a given language. To see this, consider
the example in (3), which presents a contrast between English and French related
to the way wh-questions are formed with prepositional phrases.

(3)

a. John talked to Mary.

192
 b. Who did John talk to?

c. To whom did John talk?

d. Jean a parlé à Marie.

e. *Qui a Jean parlé à?

f. A qui a Jean parlé?

In English, as in French, the wh-word replaces the object, Mary in this case.
English has two options for question formation: Either only the wh-word is moved
to the front of the sentence, as in (3b), or both the preposition and the wh-word are
moved, as in (3c). The example in (3c) sounds more formal than that in (3b), but
both sentences are possible. In French, however, only the equivalent of (3f) is
possible. To form a question, French moves both the preposition and the wh-
word; it is not possible to move only the wh-word. If L1 English learners of
French transfer the English options to French, they will have an L2 grammar for
French that has more options than the native French grammar has. They will have
to learn that only one of the options – and it’s the less common option in English
– is the only possible option in French. If L1 English speakers are told that
sentences like (3e) are not possible in French or if they are corrected when they
produce something like (3e), they are provided with negative evidence for that
structure. In other words, negative evidence is information about what is
impossible in the target language, usually in the form of error correction. So if an
L1 English speaker produces a sentence like (3e) and is corrected, that learner is
receiving negative evidence. Child language acquisition is assumed to take place
without negative evidence because there is a significant amount of research that
shows that parents usually don’t correct grammatical errors in their children’s
speech, and that if they do, the child typically ignores the correction (see Chapter
1 for some examples and discussion). Because negative evidence doesn’t seem to
play a role in L1 acquisition, a significant role for negative evidence in L2
acquisition would point to a major difference between L1 and L2 acquisition.

193
 Consider this …
It is true that L2 acquisition often involves explicit instruction and
negative evidence for many adults and teenagers. It is also true that
language classes are often set up to prioritize explicit instruction and
negative evidence. Do those two facts necessarily mean that L2
acquisition has to involve explicit instruction and negative evidence?
Does this way of structuring L2 instruction suggest that a larger
assumption about the nature of L2 acquisition is at work in how we think
about language curricula? If so, what is that assumption? Note: We will
explore the role of instruction in Chapter 6.

Based on these four observations, Bley-Vroman argued that L2 acquisition

looks more like general skill acquisition than it does like child L1 acquisition. In
other words, Bley-Vroman suggests that the domain-specific language-learning
mechanisms available to the child are no longer available to the adult. Instead,
adults rely on general learning mechanisms, which are the only mechanisms
available to them. General learning mechanisms are the cognitive mechanisms
that allow us to acquire new skills and new knowledge. They are not specific to
any one domain, skill, or knowledge area. Because these are the only mechanisms
available to adult learners, L2 linguistic systems will necessarily be different from
L1 linguistic systems. Concretely, Bley-Vroman proposes that L1 acquisition is
the result of input, domain-specific learning mechanisms, and Universal
Grammar. In contrast, L2 acquisition is the result of input, general learning
mechanisms, and the native language. We’ll evaluate the FDH in a minute, but
before we do, let’s think about what it predicts, and what we might expect to see
in the data. The FDH makes two predictions: (1) adult L2 acquisition does not
depend on the same mechanisms as L1 acquisition, and (2) it results in a
qualitatively different system. These predictions can be tested. If, for instance, it
can be shown that adults rely on the same mechanisms and progress through the
same stages of development as children, then the first prediction would be

194
falsified. Similarly, if it can be shown that adults acquire mental representations
for an L2 that look like native speakers’ mental representations, then the FDH
would be falsified on the second prediction.

The Fundamental Difference Hypothesis predicts that L2 acquisition

depends on different mechanisms than L1 acquisition and that it results in
a different kind of linguistic system than L1 acquisition does.

The FDH is problematic on both logical and empirical grounds. On logical

grounds, the FDH rests on observations about L2 acquisition that sound plausible
but actually have very little to do with the nature of the developing linguistic
system itself. For example, the observation that adult L2 acquisition usually
involves explicit instruction and negative evidence is an accurate one. At the very
least, it describes most language classrooms we are familiar with. That doesn’t
mean that L2 acquisition has to happen with explicit instruction and negative
evidence. People have been acquiring L2s without having access to classroom
instruction for millennia, and this still happens today. For example, some
anthropologists, linguists, and missionaries move to remote parts of the world to
work with groups of people whose language has no writing system and for which
there are no textbooks and there is no formal instruction. It takes time, but these
adults do manage to acquire the target language. Similarly, another observation
that the FDH rests on is that L2 learners differ from children in terms of
outcomes: Children are guaranteed to converge on a system for their native
language that matches the adults in their community, while many L2 learners fail
to acquire even the ability to order a cup of coffee in the target language. This
observation does, indeed, match our experience. It does not necessarily mean,
however, that adults and children rely on different systems for language
acquisition. It could be due to other factors, such as differences in the quantity and
quality of input that children and adults receive, or it could be due to non-

195
linguistic factors, such as motivation. (See Chapter 8 for a longer discussion of
motivation.)
In addition, the FDH doesn’t provide good coverage of the empirical data.
One of the predictions that it makes is that adults should not acquire implicit
knowledge of what is impossible in the L2, especially if the L1 and the L2 work
differently. We have seen examples of these constraints already: The
ungrammaticalities in sentences (1) and (2) above are based on subtle constraints
on English. The FDH predicts that non-native English speakers will not have the
same judgments for these sentences as native English speakers do. In fact, there is
evidence that these kinds of subtle constraints are acquirable in an L2. This
evidence comes primarily from research that investigates poverty of the stimulus
situations. Poverty of the stimulus (POS) situations are those in which a particular
syntactic structure is impossible in a language, but there is nothing about the input
that would tell the learner that the structure is impossible. Nor is there any
information that the learner could transfer from the L1. There are particular L1
and L2 pairs providing clear test cases for knowledge related to POS situations.
For example, English and Japanese differ in terms of whether every sentence
requires an overt subject. In English, every sentence requires a subject, and this is
true even in cases where the subject really has no semantic content (i.e., when we
can’t link the subject to a person or thing in the real world), as in (4c). The word
pro is used in this example and in (5) to indicate an unpronounced pronoun. When
we talk about these differences, we refer to overt subjects (subjects and subject
pronouns you can hear or see) and null subjects (subjects and subject pronouns
you can’t hear or see but are hidden in the syntax).

(4)

a. Megan drinks too much coffee.

b. *pro drinks too much coffee.

c. It’s ten o’clock.

d. *pro is ten o’clock.

196
 Unlike English, Japanese permits null subjects, and there are some cases in
which an overt subject is actually ungrammatical, as in (5d). (Note to the reader:
The second line in these examples is called a ‘gloss.’ Glosses are how linguists
provide syntactic and morphological information to the reader in a “literal”
rendering of the sentence in English. The third line in each example is how we
would normally render the sentence in English. For the reader unfamiliar with
glosses, the abbreviations are these: TOP = topic; NOM = nominative; ACC =
accusative; pro = null-subject pronoun; C O P = copular verb like ‘be’.)

(5)

a.

Hanako-wa koohii-o nomi.sugi.ru.

Hanako-T O P coffee-A C C drink.too much.N O N - P A S T

‘Hanako drinks too much coffee.’

b.

pro koohii-o nomi.sugi.ru.

pro coffee-A C C drink.too much.N O N - P A S T

‘I/you/he/she/we/they drink too much coffee.’

c.

pro jyuji desu.

pro ten o’clock COP

‘It’s ten o’clock’

d.

197
 *Kare-wa jyuji desu.

He/it-TOP ten o’clock COP

‘It’s ten o’clock.’

Essentially, the difference between Japanese and English is as follows:

English always requires that something (a word, a phrase) occupy the subject
position. In some cases, there is no semantic subject. In these cases, English
requires it or there to occupy subject position as in It’s raining or There’s fresh
coffee in the kitchen. In these sentences, there is no it or there to point to; they
don’t refer to anything (as opposed to him in I saw him yesterday, which
presumably refers to some guy the speaker and hearer both know). In Japanese,
overt subjects are optional in many cases. In addition, they are prohibited in cases
where there is no person or thing in the real world for the subject to refer to. (You
may recall that we discussed this in relation to Spanish in Chapter 2. Note how
UG puts restrictions on how null- and overt-subject pronouns work even in
languages as different as Spanish and Japanese.) To see this, compare the
sentences in (4c) and (4d) with those in (5c) and (5d). The English pattern and the
Japanese pattern are the reverse of each other. For the most part, this difference
between English and Japanese can be acquired on the basis of the input to which
learners are exposed. L1 English speakers learning Japanese as an L2 will hear
sentences like those in (5a–c) on a regular basis. Those sentences provide
evidence that Japanese permits null subjects.
Languages like Japanese are called null-subject languages, and languages
like English are called non-null-subject languages. Things get more interesting
when we look at how non-null-subject and null-subject languages handle
pronouns in embedded clauses. Let’s look at some data from English first.
Consider the sentence in (6) and think about what the pronoun he can refer to.

(6) Johni said that hei/j would be late.

198
In English, he can refer to either John or to someone else. So, sentence (6) can
mean that John would be late, or it can mean that someone else would be late. By
itself, the first interpretation, where he refers back to John, is probably more
natural, but both interpretations are possible. This ambiguity is shown by the little
letters i and j attached to the pronoun he. Note that John also has a subscript letter
i attached to it. The subscript i on he matches the one on John, so he refers to
John. The word he has a second subscript, the letter j, that doesn’t have a match in
the sentence. This indicates that the pronoun can refer to someone not mentioned
in the sentence. Now let’s see how the same sentences work in Japanese. Recall
that Japanese allows both a null pronoun (pro) and an overt pronoun. The
example in (7) shows two sentences; (7a) has a null pronoun and (7b) has an overt
pronoun.

(7)

a.

Taroi-ga proi/j mou tabe.ta to it.ta.

Taroi-N O M proi/j already eat.P A S T COMP say.P A S T .

‘Taroi said that proi/j already ate.’

b.

Taroi-wa kare-gai/j mou tabe.ta to it.ta.

Taroi- T O P hei/j - N O M already eat.P A S T COMP say.P A S T .

‘Taroi said that hei/j already ate.’

In both of these sentences, the pronouns work like they do in English. In (7a)
pro has a subscript i, indicating that it can refer back to Taro, so that the sentence
means that Taro ate. It also has a subscript j, which doesn’t have a match in the

199
sentence, and indicates that the pronoun can also refer to someone outside the
sentence so that someone other than Taro ate. The same pattern is there in (7b):
The pronoun kare can refer to either Taro or to someone other than Taro. So far, it
looks like null and overt pronouns work the same way in both English and in
Japanese.
Now let’s look at what happens when the subject of the main clause is a
word like everyone, who, or no one. Again, we’ll look at English first.

(8) Everyonei said that hei/j was hungry.

The pronoun he in (8) works the same way it does in (6): It can refer to everyone
or it can refer to someone not mentioned in the sentence. In formal English, he is
used as a neuter pronoun while in spoken English they is more popular. Even with
they the sentence would work the same: The pronoun could refer back to everyone
or it could refer to people outside of the sentence. Now let’s look at the Japanese
equivalent of (8), given in (9). Again, because Japanese has both a null and an
overt pronoun, we will look at two sentences. The sentence in (9a) has a null
pronoun, and the sentence in (9b) has an overt pronoun.

(9)

a.

Minnai-ga proi/j mou tabe.ta to it.ta.

Everyonei-N O M proi/j already eat.P A S T COMP say.P A S T

‘Everyonei said that proi/j already ate.’

b.

Minnai-wa kare-ga*i/j mou tabe.ta to it.ta.

Everyonei-T O P he-N O M *i/j already eat.P A S T COMP say.P A S T .

200
 ‘Everyonei said that he*i/j already ate.’

The sentence in (9a) works the same way as the examples we’ve already
seen. When the pronoun is null, the sentence is ambiguous and can mean either
that each person has already eaten or that someone not mentioned in the sentence
has already eaten. This pattern is broken in (9b), however. In (9b), when the
pronoun is overt, the sentence can only mean that some other person, not included
in everyone, has eaten. In other words, the only possible interpretation of (9b) in
Japanese is that everyone who was asked said that another person (Taro, or Kenji,
or someone else) already ate. It cannot mean that each person who was asked said
that they themselves ate. This difference in the interpretation of null and overt
pronouns is a property of null-subject languages more generally – the same thing
happens in languages as different from each other as Spanish, Turkish, Korean,
Italian, Mandarin, and Portuguese – and it’s called the overt pronoun constraint
(OPC). The important thing about the OPC is that there is no way to acquire it
from the input alone. Instead, once learners of null-subject languages
unconsciously determine that the language they are learning is a null-subject
language, the OPC kicks in without anyone knowing. This is a classic poverty of
the stimulus case: The OPC cannot be learned based solely on the input, and yet
it’s something native speakers have very clear intuitions about.
Now let’s consider what this means for language acquisition. Remember that
the FDH predicts that L2 learners will not have very strong intuitions about
syntactic constraints in the L2, and that this indicates that the L2 system is
qualitatively different from the L1 system. Let’s think about what this means for a
native English speaker learning Japanese as an L2. English speakers come to
Japanese with a non-null-subject mental representation, which means that they
must acquire a different representation for Japanese. They will get evidence from
the input that Japanese does not require subjects; some of it comes from time
expressions like those in (5). More will come from exposure to sentences without
subjects, which are very common in Japanese. In addition, classroom learners will
probably be told early on that subjects are optional in Japanese. So the fact that

201
Japanese permits null subjects is not a particularly difficult thing to acquire; the
evidence is there in the input. The question is whether these Japanese learners also
show evidence of the interpretive constraints on null and overt pronouns
exemplified in the OPC. Because the ambiguous interpretations in (7) and (9a) are
the same in Japanese as they are in English, English learners of Japanese might
transfer their knowledge that pronouns are ambiguous from English to Japanese.
If they do this for sentences like (9b), however, they will also interpret the
pronouns in these sentences as ambiguous. Thus, the crucial question is whether
L2 learners know that sentences like (9b) have to be interpreted following the
OPC, and are therefore not ambiguous. There is nothing about English or the
Japanese input that provides evidence for the interpretation in (9b). It’s also not
something that is discussed in language classes. The research has shown that
native English speakers are sensitive to the OPC, and that this sensitivity kicks in
relatively early in the acquisition process. Kazue Kanno tested a group of fourth-
semester L2 Japanese learners with English as an L1 and found that the L2
learners gave the same interpretation for sentences like (9b) as native Japanese
speakers. Again, to underscore the point, these learners could not have deduced
the OPC from the input. They could not transfer it from the L1. But the OPC is
part of UG. How did the learners show evidence of the OPC if they don’t have
access to UG as the FDH predicts?
In addition to Kanno’s research with learners of Japanese, other researchers
have investigated whether non-native speakers of Italian, Spanish, Turkish, and
Portuguese are sensitive to the OPC in those languages. Overall, study after study
has found what Kanno did: L2 learners are sensitive to the interpretive constraints
captured in the OPC. In addition, this sensitivity starts to show up very early; it’s
usually apparent in classroom learners during their second year of college-level
study of the language. Studies such as these, which describe clear POS situations
in L2 acquisition, run contrary to the predictions of the FDH. The FDH predicts
general failure on subtle constraints like the OPC, but the empirical evidence
actually suggests that these subtle constraints show up in L2 learners’

202
unconscious mental representations. To be sure, there is research on other
constraints and other non-OPC phenomena that point to a greater picture of UG
working in adult L2 acquisition. We have focused here on the OPC to offer a
more detailed description of this kind of research.

The Critical Period Hypothesis

One of the authors of this book worked in Japan for several years. She had
American co-workers who moved to Japan when their children were three and
five years old. Her co-worker said that he expected his children to pick up
Japanese easily, whereas he and his wife would have to work hard to learn it. This
person’s beliefs about the ease with which his children would acquire Japanese
reflect something known as the Critical Period Hypothesis (CPH). In this
section, we examine the CPH and the evidence for and against it.
The CPH proposes that there is a critical period for language acquisition.
What this means is that for a linguistic system to be fully acquired, it must be
acquired by a certain age. The term critical period comes out of biology and what
we know about how certain aspects of biological development take place. For
instance, there is a critical period for imprinting in ducklings. When ducklings are
born, there is a certain window of time in which they will bond with the mother
duck, and whatever animal they see during that critical period is the animal that
they imprint on. Similarly, there is a critical period for vision in kittens. If kittens
are not exposed to certain visual stimuli within a certain period after they are
born, they will never develop the ability to see as accurately as they would
otherwise. In both cases, it’s important to note that these functions are input
dependent (they depend on access to a specific kind of stimuli in the real world),
and that a lack of access to the critical stimuli does not entail no ability
whatsoever. Instead, it means that ducklings might look to a member of another
species to be their caregivers, and that kittens won’t develop the full range of
visual acuity they would otherwise have done.

203
 The Critical Period Hypothesis proposes that a linguistic system can only
be fully acquired if people are exposed to it before a certain age.

The CPH for language acquisition comes from Eric Lenneberg’s 1967 book,
The Biological Foundations of Language. On the basis of the neurological
correlates for human language, evidence from children who do not succeed in
learning an L1, and because child language acquisition takes place during a period
of time that is marked by a lot of neurological activity, Lenneberg argued that L1
acquisition is subject to critical period effects. Specifically, Lenneberg argued that
if children are not exposed to an L1 by the age of about five, they will not acquire
language. Although most proponents of a CPH for adult L2 acquisition cite
Lenneberg, he limited his proposal to child L1 acquisition. He mentions L2
acquisition almost in passing, stating that if someone is of average intelligence,
that person can learn another language after the beginning of what he calls “their
second decade.” He adds that the incidence of “language-learning blocks”
increases after puberty. He ends by saying that because natural languages tend to
resemble each other in so many ways, the ability to acquire another language is
present in humans.
Although Lenneberg says nothing directly about a critical period for L2
acquisition, there is some circumstantial evidence for a critical period for child
language acquisition. One piece of evidence comes from the case of Genie. Genie
was locked in a closet by her step-father at a young age and was deprived of any
linguistic input. When she was discovered by child protective services, she was
13. Despite receiving intensive interventions to teach her English, she never
developed the ability to speak fluently. Although she acquired some vocabulary,
her language production lacked syntactic structure and also lacked systematic
morphological marking. We should avoid drawing too many conclusions from the
case of Genie, though. The fact is, Genie’s childhood was marked by severe and
lengthy emotional and physical abuse, part of which included linguistic
deprivation. In short, her language deficits could be related to the cognitive and

204
emotional problems stemming from the abuse she endured. Other pieces of
evidence for a critical period for L1 acquisition come from feral children. Feral
children are children who are abandoned and spend their early years living away
from human communities. When they are discovered, they face a long road to be
reintegrated into society. These children often do not acquire the language of their
communities completely, but, again, these outcomes cannot be separated from the
neglect they endured. This is one of the major problems with directly testing the
CPH: Most of the cases we have of people who lack exposure to language do so
because they have been deprived of the social environment necessary for all kinds
of healthy human development, including the development of human language. In
fact, it is impossible to directly test the CPH because to do so would essentially
involve depriving infants and children of human contact. Thus, although the
evidence for critical period effects in L1 is suggestive, it is impossible to test the
CPH directly.

Consider this …
Assume for a minute that we had definitive evidence for a critical period
for child L1 acquisition. Does this necessarily mean that there would be a
critical period for adult L2 acquisition? Why or why not? What are some
of the other ways in which the L1 and the L2 could interact? What are
some other differences between L1 and L2 acquisition that might
influence how far learners get?

Many L2 researchers adopted Lenneberg’s hypothesis as a possible

explanation for the observed lack of success in L2 acquisition. The first papers to
investigate age effects in L2 acquisition were done by Jacqueline Johnson and
Elissa Newport. In one of these papers, Johnson and Newport investigated
whether native Chinese and Korean speakers living in the United States had
nativelike knowledge of a range of English syntactic structures. Johnson and

205
Newport gave their participants a task that asked them to listen to English
sentences and indicate whether the sentences were grammatical or ungrammatical.
Participants had different ages of arrival; some had moved to the United States to
do their graduate work, and others had moved in childhood. Johnson and Newport
found that participants who had moved to the United States between the ages of 3
and 7 performed like native English speakers on the judgment task. The older
participants were when they arrived in the United States, the more their judgments
differed from those of native English speakers. Johnson and Newport argued that
these results support the hypothesis that L2 acquisition is subject to critical period
effects.
As other researchers have pointed out, there are a number of confounding
factors that may have influenced the results that Johnson and Newport reported.
One of these factors is that age of arrival correlates strongly with other variables
that we know influence L2 acquisition, such as quantity and quality of input,
motivation to integrate, and so on. Researchers have also pointed out that
evidence that many L2 learners do not perform like native speakers on tests like
those given by Johnson and Newport is not necessarily evidence that L2 learners
cannot acquire a nativelike mental representation of language. This means that if
there is evidence that some people do attain an L2 competence that is virtually
indistinguishable from that of native speakers, this is counter-evidence to the
CPH. One study that presents counter-evidence is a case study that was conducted
by Georgette Ioup and her colleagues. This study presents evidence from two
native English speakers who acquired Egyptian Arabic as adults. It is presented in
more detail in the Exemplary Study.
One participant, Julie, learned Egyptian Arabic by interacting with native
speakers in Egypt. She did not receive any explicit instruction. Despite having
lived in Egypt for 26 years at the time of the study, she could not read or write
Arabic. The second participant, Laura, had had some formal instruction in Arabic
before moving to Egypt. She was literate in Arabic, and taught Arabic at the
University of Cairo. Both women married Egyptian men and spoke Arabic at

206
home. Ioup and her colleagues collected speech samples and syntactic judgments
from their participants. Native Arabic speakers rated the speech samples for
nativelikeness. Both women were rated as native speakers, and performance on
the syntactic judgment task was consistent with native-speaker judgments in all
but a handful of cases. Those cases were cases in which native speakers also
demonstrated some variability in their judgments. Research results such as these
are difficult to explain under the CPH.
Work on the CPH has continued apace since the early 1990s, but the basic
challenges in conducting this research are still present. Researchers have added
measures, such as length of residence (as distinct from age of arrival) and aptitude
scores, as well as created criteria for inclusion in these studies that require
participants to be perceived as native speakers. For instance, Niclas Abrahamsson
and Kenneth Hyltenstam tested L1 Spanish speakers on their L2 Swedish
competence. To be included in the participant pool, participants had to be rated as
native speakers by L1 Swedish speakers. Participants then completed tasks that
tapped lexical and morphosyntactic knowledge at the upper range of native-
speaker proficiency levels. The L1 Spanish speakers did not perform like L1
Swedish speakers, so the authors concluded that critical period effects are active
in adult L2 acquisition. Studies like those conducted by Abrahamsson and
Hyltenstam are important because they show that nativelikeness is quite rare.
Even these studies, though, have a hard time isolating age itself from other factors
that we know influence L2 attainment, such as access to sufficient input. What is
clear from critical period studies is that most L2 learners stop acquiring their L2
somewhere short of nativelike knowledge. These studies have not, however,
definitively shown that this is due to the age at which one begins acquiring an L2.
In addition, there is evidence that, in some cases, L2 learners do develop
nativelike knowledge of their L2.

Exemplary Study

207
Ioup, G., Boustagui, E., El Tigi, M., and Moselle, M. (1994).
Reexamining the Critical Period Hypothesis: A case study of successful
adult SLA in a naturalistic environment. Studies in Second Language
Acquisition, 16, 73–98.

Research Questions
How does the linguistic competence of a naturalistic L2 learner
compare to that of native speakers?

Are there differences between tutored and untutored near-native

speakers in terms of grammatical competence?

Participants
Two L1 English speakers with Egyptian Arabic as an L2:

One participant, Julie, moved from the UK to Egypt when she

married an Egyptian. She acquired Arabic entirely through
naturalistic exposure to the language. At the time of the study,
she was working as an ESL teacher/trainer at the university
level and had lived in Egypt for 26 years. She could speak
Arabic but could not read or write it.

The other participant, Laura, was an American who studied

standard Arabic in college and earned an M.A. and a Ph.D. in
standard Arabic. She moved to Egypt while working on her
Ph.D., married an Egyptian, and stayed in Egypt. At the time of
the study, she was teaching standard Arabic at the university
level and had lived in Egypt for ten years. She was fully literate
in Arabic.

Linguistic Domains Investigated

speech production

accent identification

208
grammatical intuitions

Materials, Method, and Procedure

Speech production

◦ elicited spontaneous speech samples from both L2 speakers

◦ intermixed with speech samples from native and non-native

speakers

◦ a separate group of native speakers categorized the speech

samples as native or non-native

Accent identification

◦ participants identified regional Arabic dialects and regional

Egyptian dialects

◦ participants classified speech samples from Egyptian and non-

Egyptian Arabic speakers

◦ participants classified speech samples from speakers from

different parts of Egypt

Grammatical intuitions

◦ translation task from English into Arabic containing relative

clauses, yes/no questions, wh-questions, and noun phrases joined
by words like and

◦ grammaticality judgment task that included Arabic-specific

constructions and UG-constrained constructions

◦ anaphora interpretation task that tested participants’ pronoun

interpretation preferences in different syntactic contexts

Results
Both participants were rated as native speakers by most of the judges in
the speech production task.

209
Both participants accurately discriminated between Egyptian and non-
Egyptian accents.

Julie performed more closely to native speakers in her ability to

distinguish between Egyptian accents; Laura could not distinguish
between them.

Both participants scored perfectly on the grammatical contrasts tested in

the translation task, including with respect to morphology.

Both participants performed close to native speakers on the structures

tested in the grammaticality judgment task, including with respect to
relativized questions, headless relative clauses, resumptive pronouns,
and multiple wh-questions.

Both participants performed like native speakers with respect to

anaphora resolutions in most contexts. The exception was Julie, who
did not perform like native speakers on items testing for interpretation
of pronouns in texts.

Conclusion
With the exception of one kind of anaphora, little distinguishes tutored
and naturalistic L2 learners at the end state of language acquisition.

These results provide counter-evidence to the Critical Period

Hypothesis: Both participants began acquiring Arabic as adults, and
both perform like native speakers on multiple measures.

210
 Learners Can Become Nativelike in Some Domains, But Not Others
The proposals that fall under this category differ in terms of scope and where they
place the locus of divergence between L1 and L2 grammars. We deal with two
major proposals here: representational deficit hypotheses, which deal with the
underlying mental representation, and the Shallow Structure Hypothesis, which
deals with language processing.

Representational Deficit Hypotheses

This section is called “Representational Deficit Hypotheses” because there are
several proposals that all fall under a general category. Rather than treat each one
separately, we’ll sketch out the broad outlines of the proposal. Throughout this
book, we have been talking about what it means for learners to have a mental
representation for language. Part of that representation includes morphology, the
units of sound and meaning that encode grammatical information. This
grammatical information can include agreement information, such as agreement
between subjects and verbs or between nouns and adjectives. If a language has
overt agreement, such as when a verb has a different ending depending on
whether the subject is singular or plural, both the subject and the verb have to
share the same features. In this case, the subject would have a [+plural] feature
and so would the verb. If the features on a subject and a verb don’t match, they
can’t be used together. To see how this works, let’s look at the example in (10),
from French.

(10)

a.

Je joue au tennis

I play.1S G DET tennis

211
 ‘I play tennis.’

b.

Nous jouons au tennis.

We play.1P L DET tennis.

‘We play tennis.’

c.

*Je jouons au tennis.

I play.1 P L DET tennis.

*‘I play tennis.’

The basic idea in syntactic theory is that if something is marked overtly,

either through a word or a bound morpheme (i.e., an ending or prefix that is
attached to a word; literally the morpheme is bound to another morpheme and
can’t appear by itself, such as -s for plural), there has to be something in the
syntax that reflects that difference. Overt morphology is the result of something
called functional features. These indicate syntactic information such as person
marking, tense marking, or gender agreement. Agreement happens because the
noun and the verb, in the case of subject–verb agreement, have to match. So in
(10b), for example, both the pronoun and the verb ending have a feature for first
person plural. The verb also has a feature for present-tense marking.

Proponents of representational deficit hypotheses argue that if a learner’s

L1 lacks certain functional features, learners will not be able to represent
those features in an L2.

212
 Representational deficit hypotheses rest on a fairly robust observation of a
major difference between L1 and L2 acquisition. Children generally do not have a
hard time acquiring the morphology of their native language. Adults, in contrast,
do seem to have a hard time with L2 morphology, and this is despite the fact that
language classes tend to emphasize accurate morphology, such as subject–verb or
gender agreement. Even though classroom language learners receive instruction
and practice with morphology, they often do not produce it accurately and do not
rely on it for comprehension until they are very proficient. And, in many cases,
L2 speakers never demonstrate nativelike mastery of morphology in their second
language. This difference between native and non-native speakers can’t be due to
a lack of explicit instruction. Instead, some researchers have proposed that it is
due to a systematic deficit in the underlying competence of L2 speakers. In
particular, the claim is that L2 learners do not or cannot represent new functional
features in their mental representation. Their mental representations can only
contain abstract features that already exist in their L1. So, for instance, native
English speakers who learn Spanish or French do not acquire gender marking in
these languages because these features are new and not present in their L1.
Because Spanish and French both mark gender, these hypotheses further predict
that native Spanish speakers who learn French should be able to represent the
feature for gender in French. These learners still have to acquire French
vocabulary, which takes time. Crucially, however, native Spanish speakers who
learn French will be more accurate with gender marking than native English
speakers.
Representational deficit hypotheses share an important commonality with the
FDH. Both of them predict that L2 mental representations will lack the syntax or
morphology that governs native speakers’ representations. Representational
deficit accounts are more limited than the FDH because representational deficit
accounts limit non-nativelikeness to aspects of the grammar that depend on
abstract functional features. In order to evaluate representational deficit accounts,
we need to know whether it is true that L2 speakers do not represent features in

213
their mental representation for the target language. A lot of the research that has
investigated this question has looked at whether L2 speakers accurately produce
morphology and whether L2 speakers are sensitive to violations of agreement
relationships (e.g., whether they know that sentences like *John drink coffee are
ungrammatical in English). The limitation of this research is that both
subject–verb agreement and gender agreement depend on words being well
represented in learners’ mental dictionaries for the target language. That is, for
subjects and verbs to agree, verbs must be strongly situated in the mental
dictionary. For nouns and adjectives to agree, both nouns and adjectives must be
strongly situated in the mental dictionary. So it’s not always clear whether non-
nativelike performance on these tasks is due to a representational problem or
whether it’s due to a vocabulary problem (e.g., slowness in retrieving lexical
items from the mental dictionary). Morphology is not the only aspect of language
governed by features, however. Features also govern movement. Movement is a
process by which a word or a phrase is generated in one position of the sentence
and is moved to another position as a result of a syntactic operation, such as
question formation (we touched on this in Chapter 4 with the example Alessandro
lee a menudo la ciencia ficción ‘Alessandro often reads science fiction.’). Moving
a word from one place to another leaves a trace in the word’s original position and
creates a link between that position and its final position.
For example, the standard analysis of English wh-questions involves
movement. Let’s look at the sentences in (11). The first one is a declarative
sentence, and the second is a wh-question that targets the object in (11a).

(11)

a. Bill drinks wine while he makes dinner.

b. Whati does Bill drink ei while he makes dinner?

Intuitively, there is a semantic relationship between the sentences in (11a)

and (11b). To see this, make a list of possible answers to the question in (11b).
They should all be able to replace the word wine in (11a). In other words, the

214
sentence in (11b) starts out as Bill drinks what for dinner, and then the word what
moves to the front of the sentence (actually, up higher into the sentence, as we
saw in Chapter 4) to get the final word order. This is called wh-movement, and
it’s the primary way that English forms wh-questions. In syntax, words can’t
move around in a sentence without good reason. Words can move due to the
presence of strong features. In English, there is a position at the beginning of the
sentence that determines whether the sentence is a statement or a question. When
it is a question, it is associated with [+Q] features that indicate that the sentence is
a question. In addition, the word what is tagged with the same [+Q] features. The
word what then moves to the front of the sentence so that its [+Q] features will
match the features in the initial position in the sentence. Although some languages
move the question word to the front of the sentence to form a question, other
languages, such as Japanese and Mandarin, form questions without moving the
wh-word. In these languages, the wh-word simply replaces the part of the sentence
the question is about. In other words, wh-questions in these languages look like
Bill drank what? These languages lack the strong [+Q] features that allow the
movement to take place. Because representational deficit hypotheses predict that
L2 learners cannot acquire new features in the L2, they predict that native
speakers of languages like Japanese and Chinese will not acquire nativelike
knowledge of question formation in L2 English.
Most of the work that has investigated representational deficit hypotheses
has focused on the acquisition of verbal morphology (e.g., verb endings) and
gender marking. There is significant evidence that both of these are difficult for
L2 learners. Less work has investigated the acquisition of wh-movement, but the
work that has been done tends to find that certain properties (if not all) of wh-
movement are acquirable. Wh-movement is similar to null subjects in that there is
robust positive evidence available to learners in the input (e.g., learners hear
sentences with wh-movement in them every time they encounter questions like
What does Bill drink? and Where does Megan live?). There are also subtle

215
properties of language that are related to wh-movement but are not obvious from
the input. For example, consider the sentences in (12).

(12)

a. John said Sally left.

b. Who did John say left?

c. John said that Sally left.

d. ?/*Who did John say that left?

Sentences (12a) and (12c) show two possible options for embedded clauses
in English; the complementizer that introduces the clause is optional
(complementizers are words like which, that, and who that introduce a
subordinate clause). The sentence in (12b) indicates that it is possible to ask
questions about the subject of an embedded clause in English; that is, Sally is the
subject of the verb left in the embedded clause and who refers to Sally. Given that,
it’s not clear why the sentence in (12d) should be ungrammatical or at least sound
weird – the only difference between this sentence and the sentence in (12b) is the
presence of the word that. It’s not clear why keeping the complementizer that
should make the wh-movement from the subject of the embedded clause
questionable in (12d). This is not a syntax textbook, so we won’t dwell on why
(12d) is ungrammatical here. For our purposes, what’s important is that contrasts
like those in (12) can be used to test intuitions about wh-movement in English.
Research that has investigated this has found that native speakers of languages
without wh-movement are sensitive to these contrasts (and to many others that
involve wh-movement) in learning an L2 like English, but that these intuitions can
take a while to develop. Based on what we have seen so far, it is not clear that the
hypotheses about representational deficits hold up. Learners can develop mental
representations with new features and new operations that don’t exist in their L1.

The Shallow Structure Hypothesis

216
Representational deficit hypotheses deal specifically with the linguistic system.
The Shallow Structure Hypothesis (SSH) deals with the parser. The parser is the
mechanism in the mind/brain that mediates between the input (language that we
hear or read) and our mental representation of language. During sentence
comprehension, the parser constructs a moment-by-moment (actually, millisecond
by millisecond) computation of sentence structure. As such, the parser is integral
to language comprehension, and because language acquisition is input dependent,
some kind of parsing is implicated in language acquisition. L1 parsers for fully
competent speakers rapidly integrate information from a variety of modules –
syntax, semantics, pragmatics, and real-world knowledge – as the input stream is
encountered. Native speakers often construct rich, detailed syntactic
representations for the sentences that they read or hear. Consider the example in
(11b) again, repeated here as (13).

(13) Whati does Bill drink ei while he makes dinner?

As we discussed earlier, this sentence is formed by replacing the object with a wh-
word and moving that word to the front of the sentence. Movement creates a link
between the space occupied by the original position of the wh-word and the
position where it winds up. For now, let’s call the original position represented by
ei “a gap” as in “there is a gap here because something is missing.” Research has
shown that native English speakers who read sentences like those in (13) slow
down on the word immediately following the gap; in this case, they would slow
down when reading the word while compared to environments where while is not
precede by a gap (e.g., John played the piano while Mary sang). This slowdown
in reading times indicates that the parser is integrating the function of the wh-
word with its original position in the sentence before the next word is
encountered; in other words, the parser is processing the gap itself even though
we can’t see it or hear it.
Although the L1 parser relies on a variety of sources – semantics, syntax,
and pragmatics, among others – in order to compute the L1 input, it depends
heavily on syntactic information. In fact, there is some evidence that syntactic

217
information is primary, which leads to garden-path effects. Garden-path effects
happen when the parser originally assigns a syntactic representation that turns out
not to match the intended meaning of the sentence, as in (14).

(14) The cotton clothing is made of grows in Mississippi.

In this sentence, the parser initially interprets cotton as an adjective that modifies
clothing and interprets the string the cotton clothing as the subject of the sentence.
This parse works until the parser encounters the word of, at which point it can no
longer make the structure work. At this point, it has to go back and reanalyze the
sentence so that clothing is made of is in a separate clause that modifies the
cotton. This is illustrated in (15) where R C stands for relative clause.

(15) The cotton [RC clothing is made of] grows in Mississippi.

In short, the parser heads down a “garden path” only to find out later it is on the
wrong path. The presence of garden-path effects in native speakers indicates that
the L1 parser relies heavily on syntactic information to parse sentences.
Harald Clahsen and Claudia Felser have proposed that even highly proficient
L2 speakers do not rely on the same type of syntactic information to parse
sentences in the L2 that native speakers of that language do. Clahsen and Felser
base this proposal, in part, on evidence of differences in how native and non-
native English speakers process sentences containing wh-movement in English.
The SSH also predicts that the inability to parse sentences like native speakers is
true even when the L1 and the L2 share the same syntactic structures. Thus, native
speakers of Spanish, which is a wh-movement language, and native speakers of
Korean, which is not a wh-movement language, should perform more similarly to
each other on wh-movement structures in English than either group will to native
English speakers. In other words, at least when it comes to parsing wh-movement
sentences, L1 Spanish speakers will not have an advantage over L1 Korean
speakers. Specifically, both may fail to slow down and process gaps the way
native English speakers do.

218
 The SSH has generated a significant body of research in recent years, and
results are mixed. Some researchers have found that non-native speakers
consistently rely on parsing strategies that differ from those of native speakers.
This finding is particularly robust when it comes to the processing of certain kinds
of wh-gaps. For example, work by Theodore Marinis and his colleagues
investigated whether four groups of non-native English speakers – L1 German
speakers, L1 Greek speakers, L1 Japanese speakers, and L1 Chinese speakers –
processed sentences like those in (16) in the same way as native speakers. German
and Greek are wh-movement languages, and Japanese and Chinese lack wh-
movement. If the L1 makes a difference, the Greek and German participants
should perform more like L1 English speakers than the Japanese and Chinese
speakers.

(16) The captain whoi the officer decided ei that the young soldier had
displeased ei will write a formal report next week.

There are two possible origins for who in the sentence in (16). The first one is the
object of decide, and the second is the object of displease. For the sake of
efficiency, the parser looks to the first possible position (the object of decide) to
resolve where the wh-element originated. If it can’t do it there, it keeps looking
until it reaches displease. Native speakers demonstrate longer reading times at
both that and will in sentences like (16). Marinis and his colleagues found that all
four of their L2 groups had longer reading times at will, which is where the wh-
word actually originates, but that there was no evidence of longer reading times at
that. (Again, keep in mind the researchers compare reading times of words like
that and will to sentences where gaps aren’t involved.) The researchers concluded
that this was evidence that the L2 parser doesn’t process this type of wh-
movement in the same way as the L1 parser.
At the same time, however, there is some evidence that quantity and quality
of exposure to the target language make a difference for L2 parsing. For instance,
in 2013 Christos Pliatsikas and Theodoros Marinis tested two groups of L1 Greek
speakers on sentences like those in (16). The groups were matched for

219
proficiency, but one group had not spent a significant time in an English-speaking
country, and the other group was living in the United Kingdom at the time of
testing. The group that was living in the United Kingdom was indistinguishable
from L1 English speakers, whereas the group that had not lived outside of Greece
did not process the intermediate gap. The results of this study suggest that L2
speakers can rely on the same syntactic information to process sentences in the L2
as native speakers, but that it takes a significant amount of input and exposure to
the L2 to do so.

Consider this …
Representational deficit hypotheses focus on the nature of the mental
representations L2 speakers have. The Shallow Structure Hypothesis
focuses on language processing, which involves whether speakers can use
linguistic knowledge efficiently as their parsers compute sentence
structure during comprehension. Is it possible that L2 speakers could have
knowledge of a particular structure but not be able to use that structure
while processing language? If not, can you imagine a reason why?

220
 Nativelikeness Is Possible
The last position we consider in this section is that nativelikeness is possible. As
we do this, it’s important to keep in mind that even researchers who argue that
nativelikeness is possible do not argue that nativelikeness is guaranteed for all
learners. Rather, they suggest that when non-native speakers do not perform like
native speakers at advanced proficiency levels, this is due to factors other than
systematic differences between the processes of L1 and L2 acquisition. In this
section, we consider two main proposals. The first proposal is Full Transfer/Full
Access, which deals primarily with the process of L2 acquisition. Under the broad
category of Full Transfer/Full Access, we briefly introduce non-representational
deficit accounts. The second proposal is the Interface Hypothesis, which is about
what speakers know about the L2 when they have stopped acquiring it.

Full Transfer/Full Access

We introduced the Full Transfer/Full Access (FT/FA) hypothesis in Chapter 4.
But it’s not a bad idea to review it in some detail here. Proposed by Bonnie
Schwartz and Rex Sprouse, the FT/FA attempts to account for the following
observations: (1) there is evidence for L1 influence on L2 mental representation,
particularly at early stages; (2) L2 learners can and do move away from L1
influence; and (3) L2 representations show poverty of the stimulus effects similar
to those found in L1 acquisition. FT/FA also tries to answer two questions: what
L2 learners start with and how the L2 representation changes over time. In answer
to the first question, Schwartz and Sprouse propose that the initial state (see
Chapter 4) of L2 acquisition is the L1. In other words, L2 learners will initially
use their mental representation for their L1 to process L2 input and develop a
linguistic system. This means, for instance, that native English speakers acquiring
Spanish as an L2 will start with a representation for Spanish that does not allow
null subjects, contrary to the actual properties of Spanish. Even though learners
start with the abstract features of their L1, they restructure their mental

221
representations for the L2 when the existing representations are not sufficient to
process the target language input. We can illustrate with null subjects. As English
speakers begin to learn Spanish as an L2, they process Spanish-language input. In
this input, they are forced to process sentences that don’t have overt subjects such
as Juan no está aquí. ¿Está enfermo? ‘John isn’t here. pro is sick?’ and No es
cierto ‘pro is not true.’ As the result of multiple encounters with these kinds of
sentences, their mental representations for Spanish posit that the language must be
null subject (i.e., that there is such a thing as pro, even though this doesn’t exist in
English). So the full transfer part of FT/FA refers to where learners begin. In the
scenario just described above, L1 English speakers fully transfer the properties
and features of a non-null subject to Spanish at the outset and soon these Spanish
L2 learners accept and produce sentences that don’t have overt subjects.

Full Transfer/Full Access proposes that L2 learners assume the grammar

of their L1 when they begin acquiring an L2. Because they still have
access to Universal Grammar, it is possible for them to acquire a mental
representation for the L2 that matches that of native speakers.

The full access part of this scenario is a little bit more difficult to see. As we
discussed above in the section on the Fundamental Difference Hypothesis, null-
subject languages allow both null and overt pronouns. They also place subtle
restrictions on what overt pronouns can refer to. Null pronouns can refer to nouns
and to words like everyone, no one, and who. Overt pronouns can refer to nouns
but not to words like everyone, no one, and who. This is called the overt pronoun
constraint (OPC) and it is part of UG (we reviewed this in detail above in the
examples from Japanese). Because the properties that govern the OPC are abstract
and not available from the input alone, the only way that learners could be
sensitive to the OPC is if they still have access to UG. And there is a significant
amount of evidence that L2 learners are sensitive to the OPC (again, see the
discussion above). This is the full access part of FT/FA: just as in L1 acquisition,

222
UG constrains and guides the mental representation of L2 acquisition. It has not
gone away.
Proponents of FT/FA argue that, in principle, nativelikeness is possible; they
do not argue that it is guaranteed. To acquire nativelike representation is
dependent on access to sufficient quantities of the relevant input (among other
things). At the same time, the L1 may influence the way the mental representation
for the L2 develops, and in some parts of the language this may be persistent over
time. Simply put, FT/FA claims that the mental representation for the L2 will be
consistent with UG combined with the input learners receive. This consistency is
true at all stages of acquisition. In short, learners don’t develop mental
representation willy-nilly. Whatever exists in their heads must demonstrate the
properties of UG.
Finally, FT/FA applies only to the aspects of language that are constrained
by UG. So, for example, the OPC in null-subject languages is part of UG, which
is why we consistently see it emerge in learners of null-subject languages so early.
But the difference between something like regular and irregular past-tense verbs
is not. Even though most learners might eventually learn regular and irregular
past-tense verbs, this is not because UG is involved. And just to be sure, we add
here that learners might indeed develop nativelike representations for various
aspects of the L2, but performance factors and processing demands may mask this
underlying knowledge. Indeed, Donna Lardiere has made exactly this argument
based on longitudinal data from an L1 Chinese speaker. In her research, she
demonstrated that her subject, Patti, had all of the syntactic representation for
tense and finiteness in English, but did not consistently inflect verbs for past
tense, for example.

The Interface Hypothesis

The Interface Hypothesis was formulated by Antonella Sorace and is based
primarily on non-native speakers’ knowledge and use of null and overt pronouns
in null-subject languages. As we have seen, native speakers of null-subject

223
languages, such as Italian, Spanish, Japanese, and Chinese, have a mental
representation that allows for both null- and overt-subject pronouns. They also
have an unconscious knowledge of how to interpret null and overt pronouns.
People develop interpretive preferences based on their experience with the
language. These preferences are not governed by UG (except for those related to
the OPC) and are generally assumed to be language specific. The Interface
Hypothesis proposes that properties of language governed by UG are ultimately
acquirable, but it also proposes that interfaces between different parts of language
may cause long-term non-nativeness. What’s an interface? An interface refers to
when two different parts of language must work together in sentence formation or
comprehension. In the case of null- and overt-subject pronouns, there are
syntactic properties of these pronouns and then there are discourse properties of
the pronouns used in comprehension and production. We have seen the syntactic
properties of these pronouns earlier. Discourse refers to that part of language that
connects sentences to each other to ensure cohesion. For subject pronouns,
discourse is involved in that a pronoun can take an antecedent outside its own
sentence or clause. So, on the one hand, a subject pronoun has syntactic “stuff” it
needs to obey, and then there are discourse options for that pronoun when it
comes to linking it to an antecedent (except for the OPC). Proponents of the
Interface Hypothesis argue that the syntactic properties of null and overt subjects
in a language like Spanish or Italian should be no problem for learners but the
discourse properties may be. Why? Because the discourse properties have to be
learned from the input and speak to speaker preferences and not to hard-and-fast
“rules.” Very often, these preferences require lots and lots of input. Let’s illustrate
with an example. Consider the sentences in (17), for English, and (18), for Italian.

(17) Billi called Walterj when hei/j returned from California.

(18)

a. Bill ha telefonato Walter quando pro è arrivato da California.

b. Bill ha telefonato Walter quando lui è arrivato da California.

224
 The embedded clause pronoun in (17) is technically ambiguous in that either
Bill or Walter could have been in California. In practice, native English speakers
prefer to interpret the pronoun he as referring to Bill, and not Walter. In other
words, the preference in English is for the pronoun in the embedded clause to
refer to the subject of the main clause. There is nothing in the mental
representation that dictates that it has to be this way; this is just a preference, and
there is wiggle room in that native speakers show variability in their
interpretations. Similarly, context and real-world knowledge can override these
preferences. If, for instance, the speaker and the hearer both know that Walter was
in California, then the pronoun would clearly refer to Walter. Interpretative
preferences are evident when the context is neutral or disambiguating knowledge
isn’t available. In Italian, the interpretive preferences change depending on
whether the pronoun is null or overt.
In (18a), the embedded clause pronoun is null (pro), and it behaves the way
it does in English. It can technically refer to either Bill or to Walter, but native
speakers prefer to interpret it as referring to the subject of the sentence (research
suggests upwards of 80 percent of the time). In contrast, in (18b), Italian speakers
prefer to interpret the overt pronoun lui as referring to Walter, or the non-subject
of the main clause. The idea here is that null subjects are part of “topic continuity”
(i.e., the topic of the firstsentence is Bill and it is expected to continue being the
topic of the next sentence or clause). Overt subjects are part of “topic shift” (i.e.,
the topic of the first sentence is Bill but the use of lui suggests the speaker is
changing topics in this new sentence or clause). Topic continuity is a feature of
discourse, not syntax.
A number of studies have investigated L1 English speakers’ ambiguity
resolution preferences in L2 Spanish and Italian. This research finds that, at
advanced proficiency levels, L1 English speakers know that Spanish and Italian
are null-subject languages – they are indistinguishable from native speakers on
tests of the grammatical properties of null subjects, such as the OPC and when
null subjects are required or overt subjects are forbidden. However, L1 English

225
speakers tend to diverge from native speakers when it comes to interpretive
preferences for ambiguity resolution. L2 Italian and Spanish speakers tend to
resolve ambiguities like those in (18) in favor of the main clause subject; that is,
regardless of pronoun type, there is a tendency among these learners to assume no
topic shift from one clause to the other. This lines up with native-speaker
preferences in (18a) but not in (18b). In short, work that has investigated the
Interface Hypothesis has found that L2 learners show nativelike representation for
syntactic aspects of pronouns but that they do not necessarily develop nativelike
abilities when it comes to interpreting and using pronouns. It is worth pointing out
that some L2 learners do, however, resemble native speakers on both syntactic
and discourse aspects of subject pronouns. This suggests that nativelikeness is
possible. But these learners tend to be in the minority. The question is why this is
so. The answer probably lies in the fact that discourse preferences involving such
things as topic continuity/shift with pronouns requires lots and lots of exposure.
There is some research on Spanish L1 acquisition, for example, that suggests this
is the case. Children learning Spanish as an L1 do not show adult-like preferences
for subject pronoun interpretation until around puberty. It is not surprising, then,
that L2 learners would take a long time as well, given that they often do not have
the massive exposure to input (in terms of daily contact hours) that L1 learners
have over the course of 12–13 years, which would include, of course, literacy and
massive amounts of written input.

226
 Overall Summary and Discussion
In this chapter, we’ve seen the observations we started with emerge again and
again. On the one hand, many L2 learners do not end up with an L2 representation
that matches that of native speakers in their community. This differs from L1
acquisition, in which convergence is taken for granted. On the other hand, we
have also seen evidence for fairly high achievement on measures of L2
performance and competence. Researchers have proposed various explanations
for apparent divergence from native-speaker competence. The FDH proposed that
L2 learners lack access to UG and domain-specific acquisition mechanisms. The
CPH proposed that L2 learners’ age is the problem – after a certain age, L2
acquisition isn’t possible. Representational deficit accounts propose that the
problem is in the linguistic system, but that it’s narrowly constrained. The SSH
proposes that the problem is in the L2 parsing mechanisms. The Interface
Hypothesis proposes that the problems lie not with the system or with the
acquisition mechanisms, but rather with a general lack of ability to integrate
information rapidly and efficiently across subsystems such as syntax and
discourse. As we’ve considered each of these proposals, we’ve also seen that, in
some cases, L2 learners do acquire aspects of the target language that look
nativelike.
What do we make of all of this? First, it seems likely that Position 1 (L2
learners cannot become nativelike) is too strong. Evidence for poverty of the
stimulus effects in L2 acquisition suggests that the L2 system is constrained in the
same way that the L1 system is. Similarly, evidence from stages of development
and orders of acquisition research (see Chapter 2) suggests that the process of L2
acquisition is similar to the process of L1 acquisition. In addition, it seems
unlikely that the CPH is an accurate account of the data. This is because CPH
work has a hard time separating age from other variables. It’s also because there
is some evidence for nativelikeness in adult L2 acquisition more generally. It’s a
bit harder to decide between Positions 2 (nativelikeness is possible in some

227
domains but not others) and 3 (nativelikeness is possible), at least with the
evidence we’ve presented here. In both cases, we’ve seen evidence for
nativelikeness, and evidence against it. Essentially, researchers who take Position
2 have to explain why nativelikeness happens in some cases but not others.
Similarly, researchers who take Position 3 have to explain why non-nativelikeness
happens. Both sets of researchers agree that nativelikeness is not a given for all
learners. Overall, we think the following statements are the most consistent with
the evidence we have:

It is possible for L2 learners to develop advanced proficiency in the L2.

L2 learners acquire target-like mental representations for aspects of the L2

that are governed by UG, assuming they have been exposed to the relevant
input.

The aspects of the L2 that are not governed by UG (such as vocabulary,

morphology, and discourse structure) will be acquired late, and may not be
acquired at all.

There are some documented cases of successful L2 acquisition, to the

point that native speakers rate L2 speakers as nativelike. These cases are
the exception, not the norm, but non-nativelikeness is probably not
directly due to age of acquisition or lack of access to UG. Instead, the
most likely source of non-nativelikeness is L2 learners’ access to quality
input in sufficient quantities.

Some L2 learners become nativelike; most don’t.

In sum, 40 years of research has provided the following answer to the question
“Can L2 learners become nativelike?” Some do; most don’t. Given this, we think
that it would be productive to start asking a different question: Given that some
L2 learners achieve nativelike proficiency in their L2, why don’t more L2 learners
do so? We suspect that the answer to this will come down to the interaction of

228
several factors, but one that has been relatively overlooked in the literature is the
issue of input. Some researchers, such as Maria Pavesi, have compared groups of
learners who have access to classroom input or naturalistic input (see Chapter 6).
These researchers find group differences based on quantity or quality of input.
Similarly, input has been lurking in the bushes in several of the proposals we’ve
discussed in this chapter. For instance, it could be that the difference between
early arrivals and late arrivals in much of the critical period work is due to the fact
that children who arrive at the age of, say, 10, receive a significant portion of their
schooling in English, have access to English-speaking peers, and overall, have
access to more comprehensible English input than adults who arrive in their mid
20s. If this is the case, age and access to input are highly correlated. Given that
input is a necessary condition for language acquisition, it seems that the type of
input learners get will emerge as a major factor in L2 attainment. And as we saw
in the case of L1 learners of null-subject languages, it may be literacy and
massive exposure to written input that pushes some aspects of acquisition along
(e.g., those involved in the Interface Hypothesis). Written input is much more
complex and involves many more dependencies across clauses and sentences than
everyday conversational input. This may be a much-understudied part of the
picture of how learners develop nativelike representations and abilities.

What about Social Factors?

We know that one of the major factors that influences L2 acquisition is

quantity and quality of input. Input doesn’t exist in a vacuum, however,
and the social environment someone is in may determine how much input
they receive and how rich the input is. For example, Tokyo is home to a
considerable number of expatriates from English-speaking countries.
Many of them work in English-speaking offices, speak English at home
with their families, and send their kids to English-medium instruction
schools. These expatriates are in Japan temporarily, and the expectation is
that they and their children will move back to their home country. For this

229
to happen, their kids need to have the English skills and the academic
background to fit back in to their home country. Kids and adults in this
environment might not learn much Japanese because most of their social
interactions – home, school, other activities – take place in English. On
the other hand, other expatriates in Japan, such as English teachers or
missionaries, are exposed to different networks. In many cases, these
expatriates work to integrate as much as possible into Japanese
communities. They study Japanese and develop networks with Japanese
people. Some marry Japanese citizens, and many send their children to
Japanese schools. This gives them access to different social networks, all
of which increase their access to rich input in Japanese. This still doesn’t
guarantee convergence on native-speaker norms, especially with respect to
the writing system. The Japanese writing system is particularly
complicated, and many expatriates do not need to read and write fluently
to go about their daily lives. Some do learn to read and write Japanese
fluently, though. What this example points to is that, to the extent that
social factors influence the amount of input learners have access to,
ultimate attainment will also be influenced by these factors.

230
 Recap
Here are some of the main ideas we discussed in this chapter.

Native speakers of a particular language differ from each other in terms of

their steady-state language skills. Because of this, nativelikeness is a
difficult construct to define in L2 acquisition.

The Fundamental Difference Hypothesis proposes that L2 acquisition

differs from L1 acquisition in terms of both the process of acquisition and
the nature of the system being acquired.

The Critical Period Hypothesis was originally a proposal about L1

acquisition and proposed that if children are not exposed to L1 input
within a critical period, they will not acquire their L1. Researchers have
applied the CPH to L2 acquisition, but because age, access to input,
instruction, and exposure are all confounded for adult learners, it is
difficult to state definitively that the Critical Period Hypothesis applies to
adult L2 acquisition.

There is a substantial amount of evidence that suggests that L2 learners

develop the same kinds of syntactic intuitions as native speakers of that
language. This suggests that nativelikeness, at least in the domain of
syntax, is possible.

The Interface Hypothesis proposes that to the extent that optionality

persists in L2 grammars, it is found at the interfaces between different
linguistic modules, and at the interface between the linguistic system and
other systems, such as discourse and pragmatic knowledge.

231
 References and Readings

Abrahamsson, N. & Hyltenstam, K. (2009). Age of onset and nativelikeness in an

L2: Listener perception versus linguistic scrutiny. Language Learning, 59,
249–306.

Birdsong, D. & Molis, M. (2001). On the evidence for maturational effects in

second language acquisition. Journal of Memory and Language, 44, 235–249.

Bley-Vroman, R. (1989). What is the logical problem of foreign language

learning? In S. Gass & J. Schacter (Eds.), Linguistic perspectives on second
language acquisition (pp. 41–68). Cambridge: Cambridge University Press.

Bley-Vroman, R. (2009). The evolving context of the fundamental difference

hypothesis. Studies in Second Language Acquisition, 31, 175–198.

Clahsen, H. & Felser, C. (2006). Grammatical processing in language learners.

Applied Psycholinguistics, 27, 3–42.

Ioup, G., Boustagui, M., El Tigi, M., & Mosel, M. (1994). Reexamining the
Critical Period Hypothesis. Studies in Second Language Acquisition, 16, 73–98.

Johnson, J. & Newport, E. (1989). Critical period effects in second language

learning: The influence of maturational state on the acquisition of English as an
L2. Cognitive Psychology, 21, 60–99.

Johnson, J. & Newport, E. (1991). Critical period effects on universal properties

of language: The status of subjacency in the acquisition of an L2. Cognition, 39,
215–258.

Kanno, K. (1997). The acquisition of null/overt pronominals in Japanese. Second

Language Research, 13, 265–287.

232
Lardiere, D. (1998). Dissociating syntax from morphology in a divergent second
language end-state grammar. Second Language Research, 14, 359–375.

Lenneberg, E. (1967). The biological foundations of language. New York: John

Wiley & Sons.

Marinis, T., Roberts, L., Felser, C., & Clahsen, H. (2005). Gaps in second
language processing. Studies in Second Language Acquisition, 27, 53–78.

Pavesi, M. (1986). Markedness, discoursal modes and relative clause formation in

a formal and informal context. Studies in Second Language Acquisition, 8, 38–53.

Pliatiskas, C. & Marinis, T. (2013). Processing empty categories in an L2: When

naturalistic input fills the (intermediate) gap. Bilingualism: Language and
Cognition, 16, 167–182.

Schwartz, B. (1998). The second language instinct. Lingua, 108, 133–160.

Schwartz, B. & Sprouse, R. (1996). Second language cognitive states and the Full
Transfer/Full Access model. Second Language Research, 12, 40–72.

Sorace, A. (2011). Pinning down the concept of “interface” in bilingualism.

Linguistic Approaches to Bilingualism, 1, 1–33.

Sorace, A. & Filiaci, F. (2006). Anaphora resolution in near-native speakers of

Italian. Second Language Research, 22, 339–369.

White, L. (2003). Second language acquisition and universal grammar.

Cambridge: Cambridge University Press.

233
 Following Up
1. In some circles of L2 research, the idea of whether someone can become
nativelike or not is questioned because of the definition of native. That is,
some researchers question whether there is such a thing as a native speaker
and, if there is, whether they all have the same competence in language.
Clearly, native speakers diverge on some aspects of language but not on
others. Which of the following do you think are the places where native
speakers are more or less likely to diverge? (Consider what it means for
natives to be from the “same community.”)

underlying syntax (e.g., constraints on wh-movement such as those

reviewed in this chapter)

words and their meanings (e.g., whether the word politics means the
same thing to all native speakers)

social appropriateness (e.g., what to say when someone says that a

parent has just died)

forms of verbs (e.g., infrequent verbs in the past such as strive)

inflections on nouns (e.g., adding plural markers such as the ‘z’ sound
on dog when saying “dogs” as opposed to the ‘s’ sound added to cat
when saying “cats”)

2. Review your responses from question 1 and now ask “So when we talk
about L2 learners becoming nativelike, how should we phrase the question to
make it more precise?”

3. In what ways are the Fundamental Difference Hypothesis and the Critical
Period Hypothesis related? How do they overlap? Do you think that scholars
who believe in one have objections to the points raised in the other or do you
think they would more or less agree?

234
 4. Try to articulate on your own three of the basic ideas behind the
arguments that nativelikeness is possible in some domains but not others.
Can you give examples?

5. Review the Exemplary Study in this chapter and then read the original
article in its entirety. How would you revise the summary in this chapter to
add additional information? What information would you add, and why that
particular information? Hint: Consider creating a table comparing results
from the two participants to demonstrate their knowledge and ability.

6. Find ten non-native speakers of English in your community who speak

English fairly well in your opinion. Ask them to judge the following
sentences. Is each sentence a possible English sentence (sounds OK) or not a
possible sentence (doesn’t sound OK)? Note: The contraction of want to to
wanna is typical in English and perfectly fine in spoken language, so don’t
let people get confused that wanna is “bad English.” The question is does the
person accept wanna in each sentence as used?

a. Who do you wanna ask to the party?

b. What do you wanna make for the picnic?

c. Who do you wanna call Mary on the phone?

d. What do you wanna call Mary if she asks for a nickname?

e. Who do you wanna fire the employee?

f. What do you wanna fire from this list?

Do your participants’ reactions jive with your intuitions about good and bad
uses of wanna? How about other native speakers? If there is general agreement,
would you say these non-natives are nativelike in what they “know” about how
wanna works in English?

1 Bley-Vroman (1989) lists nine observations; we have condensed them here

for ease of reading.

235
 6
Does Instruction Make a Difference?
◈

Although SLA as a field of research was inspired by early research on child L1

acquisition (see Chapter 1), almost from the beginning research on L2 acquisition
began to concern itself with instruction. In other words, researchers asked whether
explicit instruction on how language works coupled with practice impacts
acquisition in some way. For many readers, the answer may seem intuitively
obvious: Of course instruction impacts acquisition! After all, haven’t most of us
learned a language through some kind of instruction plus practice? As we will see
in this chapter, the answer isn’t so obvious. In fact, for most people what they did
consciously as learners and what actually happened to them unconsciously during
acquisition may be unrelated. To understand what instruction does or doesn’t do
in acquisition, we will begin by looking at instruction and ordered development.
But before we do, we need to touch on the issue of explicit knowledge versus
implicit knowledge of language.
Explicit knowledge of language is conscious knowledge. It is often
verbalizable, such as “to talk about someone else, add an -s sound to the end of
the verb, for example I/you/we see versus he/she sees.” Learners of English might
have this explicit knowledge. In a language like Spanish, learners might have a
conscious, explicit rule such as “adjectives must agree with nouns so that un libro
es blanco but una casa es blanca (‘a book is white-M A S C ’ but ‘a house is white-
F E M ’)”. Learners might pick up such rules from textbooks, teacher explanations,
the Internet, or other places.

236
 Implicit knowledge, on the other hand, is unconscious knowledge. Generally,
it is not verbalizable and it bears little resemblance to explicit knowledge. For
example, in the case of Spanish noun and adjective agreement, the underlying
implicit knowledge might be described by linguists as something like this:
“learners’ systems know that nouns must ‘move up’ in the phrase in which they
are found in order to get their abstract feature related to gender checked.” This
probably sounds esoteric and may not make much sense to you (i.e., What’s a
phrase? What does “move up” mean? What is feature checking?), but see
Chapters 4 and 5. The point is that implicit knowledge is qualitatively different
from explicit knowledge, which is why most people – native speakers and
language learners – cannot verbalize it. As an example of your own implicit
knowledge, try to explain the following: You know that re- can be added to verbs
to mean to do something again (e.g., paint/repaint, record/rerecord). Why are the
following impossible in English? bake → *rebake, pet the dog → *repet the dog,
stare at him → *restare at him? In short, what’s “the rule” for adding re- to verbs
to mean ‘do again’?
Within research on the effects of instruction, an important question is this:
To what extent does instruction affect the underlying implicit knowledge that
learners have regarding language? We know that instruction can affect explicit
knowledge. We see this every time an instructor teaches something, learners
practice it, and then the teacher gives a paper-and-pencil test. But acquisition is
not about the development of explicit knowledge; acquisition involves the
development of implicit knowledge – what we have referred to off and on in this
book as underlying mental representation. We also know that implicit knowledge
develops as a by-product of input processing in which there is a focus on
meaning. We also know that implicit knowledge is qualitatively different from
explicit knowledge; that is, it is not explicit knowledge that somehow is now
“automatic” and “unconscious.” So a fundamental question in the sections that
follow is this: Does instruction influence the development of underlying implicit

237
knowledge? (In Chapter 7, we will look at whether there is any relationship
between explicit and implicit knowledge.)

238
 Instruction and Ordered Development
You may recall from Chapter 2 that ordered development is well documented in
L2 acquisition, just as it is in L1 acquisition. Learners acquire morphemes (e.g.,
verb inflections, noun inflections, articles, linking verbs, auxiliaries) in a certain
order over time. In addition, learners have demonstrated clear staged development
in the acquisition of particular aspects of language, such as negation, question
formation, and other parts of sentence structure.
Our first foray into the question of whether or not instruction makes a
difference is to note that the research on ordered development has happened in
and out of classrooms. It seems that whether language is learned in classrooms or
not makes no difference in ordered development: Order still shows up. As just one
example, you may want to review the Exemplary Study by Rod Ellis published in
1989 that appears in Chapter 2. In that study, Ellis sought to determine whether
the staged development in the acquisition of German word order that had been
documented for non-classroom immigrant learners in Germany would also show
up in exclusively classroom learners of German in the United Kingdom. What he
found was that it did: The stages of acquisition of German word order were the
same in and out of the classroom.
Other such studies have all revealed the same thing. For example, also in the
latter part of the 1980s, Maria Pavesi published a study in which she compared
what she called formal and informal learners of English as an L2. Their L1 was
Italian. The formal learners were all English as foreign language students in Italy.
The informal learners were immigrants living in the United Kingdom who had
little to no formal training in English and largely “picked up” English from work
and friends. The vast majority of these learners were waiters. As Pavesi notes, the
major distinction between the formal and informal contexts for her was not just
whether learners received explicit instruction but also the type of input learners
are exposed to. In formal contexts, such as classrooms, learners are often exposed

239
to more complicated and planned discourse. Planned discourse is speech that is
not spontaneous and conversational, but rather prepared before it is produced,
such as the discourse found in a formal speech (that is read out loud), readings,
literature, and so on. Such discourse may contain many more instances of
linguistic features not found in everyday spoken language. The informal learners
were not exposed to such discourse, at least not on a regular basis like the formal
learners.
Pavesi tested her groups on relative clause formation. (You may wish to
review the more detailed discussion in Chapter 2 on markedness and relative
clauses.) What Pavesi found was that both formal and informal learners obeyed
the acquisition orders found for relative clauses. For example, they were able to
produce relative clauses on subjects (Bill is the man who wrote this chapter)
before relative clauses on objects (Bill is the man who Megan met in 2010). In
turn, they produced subject relative clauses before indirect relative clauses (Bill is
the man Alessandro sent an email to in 1996), and so on. So learning context
(formal versus informal) did not alter the sequence in which these Italian learners
of English acquired relative clause formation. (We will return to this study in a
later section of this chapter on ultimate attainment.)
Pavesi did find one difference between her formal and informal learners
unrelated to ordered development. Her formal learners, when producing non-
nativelike relative clauses, tended to use resumptive pronouns in a clause such as
Bill is the man who Alessandro sent an email to him in 1996 more than nouns, as
in Bill is the man who Alessandro sent an email to the man in 1996. In these two
examples you will see that the difference is whether to him or to the man was
inserted into the relative clause. The informal learners showed the reverse,
preferring full nouns when producing non-nativelike relative clauses. It is
important to note, however, that this difference in “strategy” when producing
relative clauses did not alter the sequence; again, subject clauses before direct
object clauses before indirect object clauses, and so on. More germane to the
present discussion is this: Why would the instructed learners insert anything at

240
all? Clearly, inserting a pronoun copy or a noun copy into the relative clause is
not English and certainly was not something they were taught. In a sense, then,
the classroom learners were doing the same thing as the non-classroom learners:
They just preferred pronouns instead of nouns. While interesting, this does not
demonstrate any significant effect of instruction.

If we reviewed all of the studies on ordered development and instruction,

we would not find one that actually shows instruction altering ordered
development in any real way.

If we reviewed all of the studies on ordered development and instruction, we

would not find one that actually shows instruction (for formal contexts) altering
ordered development in any real way. Interestingly, there are some studies that
show negative effects of instruction on ordered development. What we mean by
negative effects will become clear in a moment. One study was reported by Patsy
Lightbown and appears in this chapter as the Exemplary Study, so you can review
that study there. The other study is by Manfred Pienemann and was published in
the 1980s.
Manfred Pienemann looked at children acquiring German word order (again,
see Chapter 2 for details on the acquisition of German word order in the
Exemplary Study in that chapter). Pienemann had children who were at different
stages of the acquisition of word order. He found two things. One was that no
matter what he tried to teach the children, they could not skip stages in the
developmental sequence. Instruction did not alter the ordered development. The
second finding in his study had to do with readiness. He found that his
participants benefitted from instruction only if the instruction was on something
related to the next stage of development. If the instruction was “too advanced,” it
either had no effect or, in the case of one participant, had a detrimental effect.
Pienemann found that for that particular participant, she reverted to an earlier

241
stage of development. In short, the advanced instruction actually caused her to
regress. Hence, a negative effect for instruction.
Based on his research, Pienemann coined the term “teachability” and
launched what he called the teachability hypothesis, namely, that instruction was
only beneficial if it targeted the next stage in a developmental sequence. Anything
other than teaching the next thing the learner was developmentally ready for
would result in either no effect or possibly negative effects. Research since the
1980s on this hypothesis has yielded mixed results. In their 2017 book, Bronwen
Patricia Dyson and Gisela Håkansson summarize the research on the teachability
hypothesis. They point out that the evidence for beneficial or detrimental effects
of instruction are mixed. However, the research is consistent in that stages cannot
be skipped or changed. Once again, instruction does not seem to alter ordered
development.

242
 Consider this …
The research has shown that instruction does not alter ordered
development. The data used in most of the research is spontaneous or
semi-spontaneous oral data, not paper-and-pencil tests. Does the fact that
ordered development is unaffected by instruction mean that the underlying
implicit knowledge is not affected by instruction? Most researchers would
argue that yes, it does mean that implicit knowledge is not affected. How
would you make the argument?

To conclude this section, we will point out several things. First, research on
instruction and ordered development has been conducted from different
frameworks, using various kinds of assessment tasks to see what learners know
and can do, and has been conducted in various contexts (e.g., with children, with
adolescents, with adults, with different languages in different countries). Ordered
development always seems to assert itself, so it is probably not the case that we
haven’t tried the right instructional technique or we have found that instruction
has no impact on this kind of learner but may have an impact on a different kind
of learner. Second, although it is now accepted that instruction does not seem to
alter ordered development in any significant way, if at all, it does not necessarily
follow that instruction has absolutely no effect. The next sections of this chapter
discuss the effects of instruction unrelated to ordered development.

Exemplary Study

Lightbown, P. M. (1983). Exploring relationships between developmental

and instructional sequences in second language acquisition. In H. W.
Seliger & M. H. Long (Eds.), Classroom oriented research (pp. 217–245).
Rowley, MA: Newbury House.

243
 Participants
36 students in grade 6

L1 = French; L2 = English

location: Quebec

students rarely used English outside of class including with friends,

business, TV and so on

244
 Major Target Item
various morphemes in English, but we will focus here on -ing (He’s
eating) and 3rd person -s (He eats)

recall that -ing is acquired relatively early and 3rd person–s is acquired
relatively late

245
 Materials and Procedure
Participants were recorded three times during the sixth grade, spaced at
roughly equal intervals.

Each participant was engaged in “picture games” consisting of the

participant describing a picture and the interviewer trying to determine
which picture (from an array) the participant was describing.

The pictures were designed to create “obligatory contexts” for the use
of -ing and -s on verbs.

246
 Results
At the first recording, learners were producing correct -ing verbs about
56 percent of the time; by the third recording, their correct use of -ing
dropped to about 30 percent.

Use of 3rd person -s started at about 5 percent and increased to about 20

percent.

Interestingly, bare verbs (uninflected verbs as in He take the cake for

“He’s taking the cake”) were used about 42 percent of the time at the
first recording, increased to about 60 percent, then dropped to about 52
percent.

In short, learners got worse overall at using verbs in English to talk

about actions in progress and iterative (repeated) activities over the
course of the year.

When the researchers looked at textbook materials and what teachers

emphasized in class (i.e., explicitly worked on), they could see no
relationship between that and what the learners were doing with verbs.

247
 Conclusion
Although there was a strong focus on grammatical accuracy in the
learners’ ESL classrooms, this did not affect their acquisition of English
verbal morphemes.

What is more, the increase in bare verbs shows natural acquisition

processes asserting themselves, given that this is where learners
normally start when producing verbs in English; in short, these
classroom learners’ internal systems were “overriding” what was
happening in class. That is, they were overriding the emphasis on -ing
and -s to “start at the beginning.”

248
 Instruction and Rate of Development
Another area of focus for research on the effects of instruction has been rate of
development. Rate of development refers to the relative speed with which
learners progress through ordered development. One claim has been that
classroom learners progress through ordered development more quickly than non-
classroom learners; that is, their rate of development is faster. This idea was first
launched by Michael Long in a well-known essay he published in 1983, “Does
second language instruction make a difference?” He reviewed almost a dozen
studies comparing classroom and non-classroom learners. None of the studies
showed that ordered development could be altered, but he did find that six of the
studies concluded that acquisition seemed to proceed faster under instruction. The
rest of the studies showed no effect or at best a marginal effect.
In a 1988 essay, Bill VanPatten argued that the findings of the research
reviewed by Long must be taken with caution for several reasons. One major
reason is that it is problematic to conclude that it is actual instruction on formal
properties of language that distinguish classroom and non-classroom
environments. As Maria Pavesi pointed out in her research (see the previous
section on instruction and ordered development), there is a qualitative difference
in input and discourse when we compare classrooms and non-classrooms. The
spoken everyday language of non-classrooms that immigrant learners tend to
encounter is not as rich, elaborate, and complicated as classroom language. This is
especially true when we consider the push to make classroom learners read
written texts as soon as possible. Written texts contain longer sentences compared
to everyday spoken language and they also contain many more subordinate
clauses of varying types. This kind of language pushes learners to process formal
features of language and hold them in working memory as they make their way
through sentences. In a sense, the language of written texts is “more challenging”
and may push learners in a classroom to do more internal mental work than

249
learners in non-classrooms who aren’t working at being literate. Classroom input
and written texts also contain more examples of marked aspects of language and a
broader range of vocabulary. All of these differences most likely work to give
classroom learners richer input compared to non-classroom learners, especially
when the latter are not working toward any degree of literacy.
VanPatten also argued that it might not be appropriate to compare classroom
and non-classroom learners, especially if the latter consist of economically and
politically motivated immigrants. Their reasons for learning a language may be
quite different from those of classroom learners and their support for acquisition
may be less. What is more, they may not have as much access to input and
speakers of the language as researchers think they do. Consider for a moment the
Italian immigrants in Pavesi’s study. Why were they in the UK? And what were
their occupations? Clearly, they were there for economic reasons and not to learn
English. They worked largely in service, with most of them being waiters. How
much English does a waiter need to learn to be a waiter in the UK? And who are
the friends and networks of people with whom these waiters interact? Many must
have families and live in neighborhoods with other immigrants. Who do they
identify with? Thus, their interactions with native speakers of British English may
be limited. Compare this scenario to one of the authors of this book. He emigrated
to the UK from Italy to pursue graduate work, eventually earning a Ph.D. in
Applied Linguistics. He married an English-speaking Irish woman and raised his
children in the London area. His network and support group consists not of other
Italian speakers but instead of native speakers of English, including his immediate
family. This is another instance in which social factors influence acquisition, a
point we will return to later in this chapter.

No strong case can be made for instruction speeding up acquisition.

Classroom learners may be faster than non-classroom learners for a
variety of reasons unrelated to instruction.

250
 The point of these comparisons is that motivation and the consequences of
that motivation for learning a language may profoundly affect how fast and how
far one goes in the acquisition process. The research comparing classroom and
non-classroom learners has tended not to take such factors into consideration. For
this reason, we must take with a grain of salt that it is the focus on formal
elements of the classroom that causes the differences that Long observed in his
review of studies back in 1983. With that said, we can’t rule out completely that
instruction isn’t an important variable in such studies either. Instead, we are
simply saying no strong case can be made for instruction speeding up acquisition.
Classroom learners may be faster than non-classroom learners for a variety of
reasons: the quality of input, motivation, access to interaction with speakers of the
language, among others, as well as instruction itself (if at all).
Since the 1980s, there has been little direct work comparing classroom and
non-classroom learners. As we will see in a later section, most research on the
effects of instruction has focused on short-term comparative studies of classroom
learners only. So for now, it seems that the jury is out on whether instruction on
formal features of language is the causative variable to explain why, in some
cases, classroom learners seem to acquire language faster than non-classroom
learners. What is more, if there is an effect on rate, it is not clear if the rate is on
explicit knowledge or implicit knowledge. Classrooms may help learners build up
explicit knowledge quickly; but they may or may not accelerate the development
of implicit knowledge. Much depends on what is measured in the research (i.e.,
what kind of “tests” are given to learners). We will discuss this issue in a later
section in this chapter.

251
 Instruction and Ultimate Attainment
Somewhat related to rate of acquisition is ultimate attainment. Ultimate
attainment refers to how far learners progress and to what degree they approach
anything that looks nativelike. You may recall from Chapter 5, which asked
whether learners can become nativelike, that although nativelikeness seems to be
possible, it is not the norm for most L2 learners. In this section, we will discuss
the extent to which instruction affects how far learners get in acquisition.
Very little research has been conducted in this area and little research can be
used to address the question in an indirect way. For the present discussion, we
will focus on two studies that are suggestive. The first is the Maria Pavesi study
reviewed in the first section of this chapter. Pavesi compared formal (classroom)
and informal (non-classroom) learners of English with Italian as an L1. She
looked at their acquisition of relative clauses. You may recall that she found that
instruction did not alter ordered development: Both groups acquired relative
clauses in the same order. She did find that there was a strategy difference when
producing non-native relative clauses. The classroom group tended to use
resumptive pronouns (e.g., Bill is the man who Megan met him seven years ago)
while non-classroom learners tended to use nouns (e.g., Bill is the man who
Megan met the man seven years ago).
Pavesi also found one other difference between the two groups. While the
order of acquisition was the same for both groups of learners, significantly more
classroom learners acquired more types of relative clauses than non-classroom
learners. You may remember that the order of acquisition is this: subject clause →
direct object clause → indirect object clause → oblique clause → genitive clause
→ object of comparison clause. Pavesi found that many more classroom learners
showed knowledge of and ability with object of comparison clauses compared
with non-classroom learners, for example. This did not mean that non-classroom

252
learners did not acquire object of comparison clauses. Some did. The point is that
significantly more classroom learners than non-classroom learners did.
One interpretation of this particular finding is that somehow classrooms help
learners get further down the acquisition path. That is, classroom learners may
reach higher levels of ultimate attainment for some things in language when they
are compared to non-classroom learners. Pavesi herself did not conclude that
these results were necessarily a reflection of instruction on relative clauses.
Instead, her conclusion was that classroom learners, because of the focus on high
degrees of literacy, tend to get more elaborate input (see the discussion in the
previous section on rate of acquisition). More elaborate input contains more
relative clauses and more types of relative clauses. Non-classroom learners, if
exposed largely to spoken speech, would encounter many fewer relative clauses
and fewer types of relative clause. For Pavesi, then, the differences she found in
ultimate attainment for her two groups of learners comes down to quality (and
possibly quantity) of input.
To be sure, as we said earlier, we cannot completely rule out an effect for
instruction itself for higher levels of ability with relative clauses in Pavesi’s study.
Another study that compares advanced classroom and advanced non-classroom
learners might shed some light on this issue. Jason Rothman published a study in
2008 in which he looked at classroom and non-classroom learners of Spanish as
L2 with English as L1. He was interested in their acquisition and use of the
preterit and imperfect verb forms in Spanish. The preterit and imperfect are past-
tense verb forms used to encode information about how a speaker views a
particular event relative to a point in time in the past. The distinction between the
two and how they are used involves complex and abstract information that we
cannot describe in this chapter. What is critical here are two things: (1) much of
the difference in use between the two verb forms is difficult to teach and is not
readily clear from the input that learners hear or see; (2) English has no such
contrasting verb forms so learners cannot transfer any knowledge or ability with
the past tense into learning Spanish as an L2.

253
 Rothman tested his learners of Spanish using a cloze-selection task based on
Goldilocks and the Three Bears as well as a sentence-level completion task. The
cloze-selection task required the participants to read the story and select from the
verb choices available. The sentence-level completion task was basically a fill-in-
the-blank test in which participants had to write in a verb form based on what they
understood from the sentence. He also tested a group of native speakers of
Spanish for comparative purposes.
Rothman’s results are somewhat surprising. On the Goldilocks cloze-
selection test, there was no significant difference in how the native speakers and
the non-classroom learners performed. However, there was a significant
difference in how the native speakers performed and how the classroom learners
performed, as well as a significant difference between how the non-classroom and
the classroom learners performed. In all comparisons, the natives and the non-
classroom learners performed at higher levels compared to the classroom group.
On the sentence-level completion task, Rothman obtained the same results: The
native and non-classroom learners performed similarly, with both groups better
than the classroom learners.
What Rothman’s results suggest is that the classroom learners did not fare as
well as the non-classroom learners; that is, they weren’t getting as far in
acquisition. So perhaps classroom instruction (and the classroom learners did
receive regular instruction on the past-tense verb forms throughout their studies of
Spanish) was actually detrimental. Before accepting this conclusion, we need to
examine one other possibility: namely, that the non-classroom group was already
at a higher level of proficiency compared to the classroom group. If this were the
case, then we would expect them to perform better. Rothman did not include a
general test of proficiency so there could have been a real difference between the
groups from the outset. However, an interesting aspect of his study might lead us
to lean more toward the detrimental effect of instruction in this particular case.
When Rothman analyzed the actual non-nativelike responses, he found that the
classroom learners were non-nativelike precisely on those items on which they

254
could apply pedagogical rules they had learned. These rules are simplified and
somewhat erroneous in what they describe. For example, one such rule is that
stative verbs such as ‘be’ and ‘have’ typically occur in the imperfect. This led the
classroom learners in Rothman’s study to overuse the imperfect with these verbs
when the preterit was required. The non-classroom learners had no access to such
a rule. Another example is with the verb querer, for which learners are taught
lexical translations; in the imperfect the verb means ‘want’ but in the preterit it
means ‘tried to.’ This is not necessarily the case (certainly they don’t mean two
different things to native speakers), and in the tests that Rothman used he noticed
the classroom learners not using the preterit with querer when it indeed meant
‘want.’ Again, the non-classroom learners did not exhibit this kind of error, as
they had no recourse to such pedagogical rules.

If classroom learners are filtering the input through pedagogical rules by

attaching the wrong function or meaning to items they encounter in the
input, this may slow down acquisition and affect ultimate attainment

In short, Rothman’s research is suggestive: If classroom learners are learning

and applying pedagogical rules and/or if they filter the input through these
pedagogical rules by attaching the wrong function or meaning to items they
encounter in the language surrounding them, this may slow down acquisition
and/or negatively affect ultimate attainment.
Missing from the research on ultimate attainment are long-term studies.
That is, there is no real research that looks at how instruction at one time affects
learners many years down the road. Such research would be extremely difficult to
conduct for all kinds of reasons. But the main reason is that such a study cannot
control for the most important intervening variable: input. An intervening
variable is a factor that influences the outcomes of a study that a researcher cannot
account for or control. In the case of ultimate attainment, a researcher cannot
control the variation of quantity and quality of input, and interaction with that

255
input, that learners receive over the years. Most recent research on the effects of
instruction (over the last several decades) has focused on short-term studies of the
effects of instruction. We will turn our attention to this kind of research in the
next section.

What about Social Factors?

As mentioned in Chapter 5, identity is a powerful factor that shapes and

guides how we feel about language acquisition as well as our behaviors in
acquiring another language. Scholars working in the area of social context
and identity offer clear examples of how a learner’s feelings toward the
environment or how that learner feels about being part of another culture
may work against acquisition. Bonnie Norton, for example, wrote about
Mai in 1997. Mai was a Vietnamese immigrant who’d made tremendous
effort and sacrifice to attend an evening ESL course to improve her
English. However, the teacher’s method and the focus on students
bothered Mai so much she eventually stopped attending. This clearly
affected how the classroom might have helped Mai in any way. Socially,
she just couldn’t connect with that particular classroom community. And
way back in the late 1970s John Schumann described what he called the
pidginization process. In a longitudinal case study, Schumann examined
Alberto’s development of English. Alberto was an economically
motivated immigrant from Central America who had a low-paying job in
the Boston area. Schumann found that Alberto’s English hardly developed
during the time Schumann studied him and attributed this in part to
Alberto’s non-identification with the new culture he was in. Alberto
seemed to be fine with “good enough to get the job done.” He had no real
intention of integrating into a new culture.
In short, both context and identity can affect learner motivation,
which in turn affects the efforts they make to engage with the second

256
language. This is true both in and out of the classroom. As such, these
factors affect rate and ultimate attainment.

257
 The Effects of Instruction and Short-Term Studies
It is fair to say that dozens of studies – well over a hundred – have been
conducted on the effects of instruction on L2 acquisition using what is called
short-term research design. By this, we mean that measurement of what learners
can do or what they know is conducted immediately after instruction or soon after
(usually within a month). Only a handful of studies have examined the effects of
instruction over a longer period (i.e., six months to a year), which we will discuss
a bit later. These short-term studies may take place in a classroom with whole
groups of students as participants or they may take place in laboratories where
participants undergo a treatment individually via computer-mediated instruction.
In most such research, the standard design is: (1) pre-test the learners to see what
they know or can do with linguistic feature X, (2) provide some kind of treatment
(instruction) on X, (3) post-test the learners and compare results to the pre-test. If
for (2) there are different kinds of treatments being compared, then in step (3) the
research also compares the groups’ performance relative to each other.
Analyses of this line of research (called meta-analyses), in which all eligible
studies are submitted to statistical analysis to determine the strength of any effects
of instruction, have concluded that instruction makes a difference. (Eligibility
includes such things as appropriate and full reporting of statistical information,
inclusion of such things as means, standard deviations, sample sizes, and so on,
for all groups in a study.) For example, John Norris and Lourdes Ortega published
one such meta-analysis in 2000. After combing through the many studies that had
been conducted prior to 2000, they found that 49 met the criteria to be included in
a meta-analysis. Their conclusion was that there did seem to be some benefit to
instruction. That is, overall, the studies showed that learners were better on post-
test measures than on pre-test measures. However, Norris and Ortega offered
several very important caveats regarding this finding – caveats that we will review
in a minute.

258
 In 2010, Nina Spada and Yasuyo Tomita published another meta-analysis.
They located 41 eligible studies investigating whether explicit instruction (that is,
explaining things to learners / giving them “rules” as part of the instructional
treatment) yielded different outcomes from implicit instruction (instructional
treatment in which explanation about what was being learned was not included).
What they determined was that, overall, there was an effect for instruction and the
effect was greater for learners engaged in explicit instruction. Again, “effect” was
determined by comparing post-test scores to pre-test scores. A similar study was
conducted by Jaemyung Goo and his colleagues, published in 2015 with similar
results.
At first blush, it seems that instruction does make a difference. These meta-
analyses, which summarize all the eligible research, point to positive effects for
instruction on grammatical properties of language. At the same time, a number of
questions arise that cause us to take such conclusions with caution. These
questions include the following:

What is it that the researcher is measuring? What does it mean to “know”

something? In particular, does the research on the effects of instruction
measure explicit knowledge or implicit knowledge (and what does the
researcher think is in the implicit knowledge)?

What kind of task can be used to measure what learners know and how do
we know it is a valid measurement?

If we are looking at what learners can do after treatment, the same

questions apply about what is measured and how it is measured.

What is the nature of the treatment? What is it trying to affect? Does the
treatment attempt to affect explicit knowledge or implicit knowledge (or
something else)?

How does the treatment reflect what we know about acquisition more
generally and what does the researcher believe he or she is “altering” in

259
 the learner? That is, what underlying “processes” does the instruction
attempt to affect?

According to a number of scholars, there is a bias toward the testing of

explicit knowledge in research on the effects of instruction.

Norris and Ortega, for example, provide a significant caveat in their

discussion that it is not clear that most of the studies they included in their meta-
analysis measured implicit knowledge. In fact, the opposite seemed to be true: 90
percent of the studies in the meta-analysis used tests that probably measured
explicit knowledge. They suggest that had different kinds of tests been used, then
the outcome for the effects of instruction they observed might have been different.
Catherine Doughty, in a 2003 critique of such research, argued that there is a bias
toward explicit knowledge and the testing of explicit knowledge in research on
the effects of instruction.
To see what all of this means, let’s imagine a researcher is interested in
whether instruction makes a difference in the acquisition of passive structures
such as The cow was kicked by the horse. Let’s suppose that the researcher uses
Chinese and Turkish learners of English as an L2, largely because these languages
form passives differently from each other and quite differently from how passives
are formed in English. The researcher gives a pre-test and finds the learners are
terrible with passives. Then some kind of treatment or instruction is provided to
learners on passives followed by a post-test similar to the one given as a pre-test.
The learners all perform significantly better. The researcher concludes that
instruction made a difference. In this fictitious but stereotypical study, we have to
ask ourselves how learners were tested on their knowledge of passives and how
that test avoided learners using the explicit knowledge they gained during the
treatment as opposed to any fundamental change in underlying implicit
knowledge. What if the researcher asked the learners to “judge” sentences by

260
indicating whether the sentence was grammatical or ungrammatical or “good” or
“bad,” as in these three examples (the first is good; the second two are not):

The cow was kicked by the horse.

The cat was bitten the dog.
The mouse trapped by the cat.

Asking learners to judge sentences about something they have just learned
about most likely invites them to use their new explicit knowledge. “Now let’s
see. I just learned about passives. So is this sentence trying to be a passive? Is it
correctly formed? No it’s not. Oh, there’s a by missing in this sentence …”
Alternatively, as in some published research, our participants might be given a
prompt and asked to write out a sentence. In our fictitious passive study,
participants might get a picture of a cat being bitten by a dog. On a piece of paper
or on a computer screen, the participant is told to complete this sentence based on
what is in the picture: “The cat _______________________.” Again, this type of
measurement invites the use of explicit knowledge. “Let’s see now. We just
learned about passives so I’m assuming they want me to use a passive here. How
was that formed again? Oh yeah, I need a form of ‘be’ and also I have to use ‘by.’
I think it’s ‘the cat is bited by the dog.” Note that the learner doesn’t have to get
the past participle correct in order to engage explicit knowledge. The learner is
consciously applying what he or she remembers about the overall rule.
As we said above, research on the effects of instruction cannot mimic what
happens in classrooms; we know that instruction “makes a difference” for test-
taking in classrooms. Research has to have measurements that probe underlying
implicit knowledge that is qualitatively different from explicit knowledge. This
has proven to be elusive for researchers. Let’s look at a different example to see
the opposite: how instruction is questionable if we attempt to measure implicit
knowledge.
In 2012, Bill VanPatten and his colleagues, Gregory Keating and Michael
Leeser, published a paper in which they used what is called self-paced reading to

261
determine learners’ sensitivity to violations of various structures in Spanish (the
learners’ L1 was English). In self-paced reading, a participant sits in front of a
computer screen and reads a sentence in bits and pieces. After reading the
sentences, the participant answers a comprehension question about the content of
what was just read, not about its well-formedness as a sentence (i.e., they are not
asked to judge whether a sentence was good or bad). So what a participant
initially sees on the screen is a series of blanks, such as this:

_____ ______ ______ ______ ______ ______

As the participant presses a button, a blank disappears to reveal part of the

sentence, as in:

Now _____ ______ ______ ______ ______

When the participant presses the button again, the part that was just revealed
disappears and the next part appears, as in:

_____ I lay ______ ______ ______ ______

And this continues with each button press, as in

_____ ______ me ______ ______ ______ [press button]

_____ ______ ______ down ______ ______ [press button]
_____ ______ ______ ______ to ______ [press button]
_____ ______ ______ ______ ______ sleep [press button]

In short, the participant reads at his or her own pace, with each piece of the
sentence appearing one at time, hence the name “self-paced reading.” As buttons
are pressed, this procedure pushes the reader to keep in mind what he or she just
read to “build the sentence” and its meaning. Researchers manipulate parts of
sentences to include wrong or odd things and the software records the reading
time for each segment (we will see examples in a minute). The researcher then
compares these reading times with the segments from other sentences in which

262
nothing was wrong or odd. After the participant reads a sentence, a button press
reveals a question about the content of the sentence, thus encouraging the reader
to focus on meaning and not on how the sentence is formed. What this kind of
research demonstrates is that even if learners are focused on meaning, there are
generally slightly longer reading times on sentence segments that contain wrong
or odd things compared to those that don’t.
Here are two examples from VanPatten, Keating, and Leeser’s study. Both
sentences mean ‘Right now I’m drinking a soda in the cafeteria.’ However, the
second sentence is ungrammatical in that verb form means ‘he’s drinking’ not
‘I’m drinking.’ That is, the verb and subject do not match. (Note: The slash ‘/’
indicates the segments that appeared one at a time during self-paced reading.)

Ahora /yo /tomo /un refresco/ en/ la cafetería. [match]

Ahora/ yo /toma/ un refresco/ en/ la cafetería. [do not match]

The sentence was followed by a comprehension question “In the sentence

you just read, the person is drinking something cold. True or False?” VanPatten,
Keating, and Leeser measured the reading times on the verb forms and the phrase
immediately after the verb (called the “spillover region” – sometimes there is a
lag in reading and the longer reading time happens in the spillover region).
They also looked at sentences such as the following. Again, the first sentence
is grammatical and the second is ungrammatical and both mean ‘Robert no longer
travels to France because he has no money.’ (The ungrammaticality has to do with
word order in that Spanish puts the verb between the ‘no’ and the ‘longer’
whereas English puts the verb after ‘longer.’)

Roberto/ no viaja más/ a Francia/ porque/ no tiene/ dinero. [word order like
Spanish]
Roberto/ no más viaja/ a Francia/ porque/ no tiene/ dinero. [word order
like English]

263
 The sentence was followed by a comprehension question “In the sentence
you just read, the person has extra money for travel. True or false?” What the
researchers compared were the reading times on no viaja más and no más viaja as
well as the spillover region a Francia. Their participants were native speakers of
Spanish and intermediate-level learners of Spanish who were taking their first
courses in Spanish literature and culture. Note that this was not a study on effects
of instruction: that is, there was no pre-test, treatment, and post-test. The
researchers were simply testing learners on their underlying knowledge of
particular features of Spanish to answer a question unrelated to instruction. But
the study and results are actually relevant to the discussion at hand.
What VanPatten, Keating, and Leeser found was that on the sentences where
the verb did not match the subject, native speakers slowed down on the expected
areas of the sentence when there was a mismatch. The L2 learners did not.
However, on the sentences where word order was an issue, both native speakers
and the L2 learners slowed down. What is significant about these findings is that
learners are taught subject–verb agreement and practice it almost from the first
day of learning. It is also something they receive feedback on and are tested on.
However, the word order with no VERB más is not taught and is a very infrequent
structure in the language these learners were exposed to. In short, on the thing for
which they did not receive instruction, these intermediate learners performed in a
manner similar to native speakers. On the thing for which they did receive
instruction and feedback over the years, they showed no sensitivity to
grammatical violations. The question is this: What happened to all that
instruction, practice, feedback, and testing on verb forms? And how could these
learners perform like native speakers on something else for which they did not
receive instruction, practice, and so on? For the second question, we know this to
be a poverty of the stimulus situation (see Chapters 2 and 5). Learners have come
to know more about the language than what they were exposed to. However, for
the first question, it is not clear what happened to all the instruction, practice, and
so on regarding subject–verb agreement in Spanish. All we can suggest here is

264
that, as in the case of the Lightbown study and others like it, instruction does not
seem to override how such things as verb forms get into learners’ mental
representation. Something internal to the learner is in charge of that.
Before continuing, it might be beneficial to make an observation about
research methodology. In this regard, two interesting things emerge from the
VanPatten, Keating, and Leeser study. First, the study shows how something like
self-paced reading can be used to measure underlying implicit knowledge – or
lack of it. Because participants’ attention is on reading for meaning and reading
times are measures of unconsciously “detecting something funny,” this kind of
measurement may be more appropriate for use in research on the effects of
instruction. A second interesting thing is that the researchers included completely
unrelated structures, which increases the internal validity of the study. Thus, the
results of the no VERB más reading times have to be an indication of implicit
knowledge because such structures are not taught or practiced. Thus, the results of
subject–verb reading times – in which no sensitivity was demonstrated – suggest
that in the implicit knowledge of the learners, subject–verb agreement just wasn’t
there yet. We can only make this conclusion because the researchers had other
structures for comparison – structures on which the learners had no instruction or
practice. (For the record, the researchers also had a third structure – also not
taught or practiced in Spanish language classrooms and for which the learners
showed sensitivity to ungrammaticality just like native speakers – but we want to
keep the presentation simple here.)

265
 Consider this …
We said early on in this chapter that there is a qualitative difference
between explicit knowledge and implicit knowledge. One way to look at
this is that explicit knowledge tests “rules”: the kinds of rules we find in
textbooks, teacher explanations, online sites, and so on. Implicit
knowledge, on the other hand, does not look like rules as teachers and
students think of rules. As Bill VanPatten and Jason Rothman have argued
in a 2014 publication, mental representation cannot be rules in the classic
sense. Language is too abstract and complex to be captured by textbook
rules. Even a rule as seemingly simple as “subjects have nominative case
and objects have accusative case” is violated in English whenever and is
used as in “Me and Bob are going” as opposed to “Bob and I are going.”
This is because what actually underlies how subject pronouns and object
pronouns are used is much more abstract than the textbook rule. (By the
way, try saying this to yourself: “I and Bob are going.” Doesn’t it sound
awful?) So we may not make headway in research on the effects of
instruction until researchers grapple with the nature of implicit knowledge
as opposed to explicit “rules.”

The point we are making, as we did in earlier sections of this chapter, is not
that instruction makes no difference. Instead, the point is that we cannot rely on
the research to tell us instruction does make a difference – at least not on the
underlying implicit knowledge we are interested in. As stated earlier, there is a
clear and overwhelming bias toward testing explicit knowledge in the research on
the effects of instruction.
Absent so far in this discussion are issues related to several of the questions
we listed earlier. They are repeated here:

What is the nature of the treatment? What is it trying to affect?

266
 How does the treatment reflect what we know about acquisition more
generally and what does the researcher believe he or she is “altering” in
the learner?

To date, no systematic analysis of treatments has been conducted on the

research regarding the effects of instruction. Two of the meta-analyses referred to
earlier (Spada & Tomita, Goo et al.) did separate what they call explicit
instruction from implicit instruction. The basic difference between explicit and
implicit instruction in the meta-analyses centers on whether teachers (or
computers) teach participants rules and whether participants know what they are
learning and/or “practicing.” Under explicit instruction, learners get rules and
know what they are learning; under implicit instruction, learners don’t get rules
and presumably don’t know what they’re learning. Unfortunately, such a division
of treatments is too broad to be meaningful because within each group there is
such variation in what is actually taught and how it is taught as to render the two
categories meaningless. Some of these matters will be addressed in the next
chapter on the roles of explicit and implicit learning, but for now we can examine
one example to see the problem. In these meta-analyses, explicit instruction may
contain a treatment in which learners read a rule, write out sentences and practice,
and perform other kinds of activities. Implicit instruction may contain a treatment
in which learners hear sentences, one after the other and then select from several
pictures to indicate what they understood the sentence to mean. Clearly, we are
comparing apples and oranges when we lump studies under one category or
another. That is, we cannot tease out if the results are due to explicit versus
implicit instruction or the actual treatment used in particular studies.
What is more, this broad definition of explicit versus implicit learning
ignores the fundamental problem we have identified above: the inherent bias in
the research on testing explicit knowledge and not understanding that implicit
knowledge is qualitatively different.

267
 In none of the longer-term studies (in which researchers have tested
learners up to a year later) did the effects of instruction show up. Short-
term effects seem to disappear over time.

Before concluding, we should touch on the issue of longer-term effects of

instruction. As mentioned in the section on ultimate attainment earlier, the
problems of doing actual long-term research (i.e., over years) to see if the effects
of instruction exist are insurmountable. Still, there have been a handful of studies
that have looked at the long(er)-term effects of instruction, meaning after several
months to a year. In 2004, Bill VanPatten and Claudia Fernández published a
study in which they reviewed the four studies that could be considered longer-
term. In none of those studies did the effects of instruction show up at a later date.
That is, when the researchers went back to test learners’ performance on the
things they were taught, the short-term effects had disappeared. In VanPatten and
Fernández’s own study using processing instruction, they tested learners eight
months after instruction. They found that although learners performed
significantly better on the delayed post-test compared to the pre-test, their scores
had still declined significantly after the short-term post-test. One could speculate
that if VanPatten and Fernández had returned several months later and tested
again, they might find the effects of their treatment to have disappeared altogether
– in line with the studies they reviewed as part of the background to their own
research.
We conclude this section with the same caveat we have stated in previous
sections. Even though the research to date does not suggest that instruction has
short-term effects on implicit knowledge, this does not necessarily mean that it
couldn’t. As we pointed out, we need more sophisticated research designs that use
measurements (tests) that attempt to test implicit knowledge and not explicit
knowledge. For now, the best conclusion about research on the effects of
instruction is this: Instruction clearly helps explicit knowledge, at least in the

268
short run. It is not clear what it does, if anything, for the implicit mental
representation we call “language.”

269
 Recap
Here are some major ideas and topics covered in this chapter.

There is a qualitative difference between explicit and implicit knowledge

of language. Researchers are interested in the development of implicit
knowledge.

Instruction on formal features of language does not affect ordered

development such as morpheme orders and staged development.
Instruction does not change the order in which things are acquired.

It is not clear that instruction on formal features of language speeds up

acquisition. There are important intervening variables (e.g., quality of
input) that need to be considered when comparing those who get
instruction and those who do not.

It is not clear that instruction on formal features of language affects

ultimate attainment. It is almost impossible to do this kind of research.

It is not clear that instruction on formal features of language affects short-

term gains in acquisition. Too much of the research is biased toward the
testing of explicit knowledge.

For the research on the effects of instruction to make advances, it needs to

consider ways in which to assess implicit knowledge along with a
consideration of what implicit knowledge actually is.

270
 References and Readings

Doughty, C. J. (2003). Instructed SLA: Constraints, compensation, and

enhancement. In C. J. Doughty & M. H. Long (Eds.), The handbook of second
language acquisition (pp. 206–257). New York: Blackwell.

Dyson, B. P. & Håkansson, G. (2017). Understanding second language

processing: A focus on processability theory. Amsterdam: John Benjamins.

Ellis, R. (1989). Are classroom and naturalistic acquisition the same? A study of
the classroom acquisition of the German word order rules. Studies in Second
Language Acquisition, 11, 303–328.

Goo, J., Granena, G., Yilmaz, Y., & Novella, M. (2015). Implicit and explicit
instruction in second language learning: Norris and Ortega (2000) revisited and
updated. In P. Rebuschat (Ed.), Implicit and explicit learning of languages (pp.
443–482). Amsterdam: John Benjamins.

Long, M. H. (1983). Does second language instruction make a difference? A

review of research. TESOL Quarterly, 17, 359–382.

Norris, J. & Ortega, L. (2000). Effectiveness of second language instruction: A

research synthesis and quantitative meta-analysis. Language Learning, 50,
417–528.

Norton, B. (1997). Language, identity, and the ownership of English. TESOL

Quarterly, 31, 409–429.

Pavesi, M. (1986). Markedness, discoursal modes, and relative clause formation

in a formal and an informal context. Studies in Second Language Acquisition, 8,
38–55.

Pienemann, M. (1984). Psychological constraints on the teachability of languages.

271
Studies in Second Language Acquisition, 6, 186–214.

Rothman, J. (2008). Aspect selection in adult second language Spanish and the
Competing Systems hypothesis: When pedagogical and linguistic rules conflict.
Languages in Contrast, 8, 74–106.

Schumann, J. (1978). The pidginization hypothesis. Rowley, MA: Newbury

House.

Spada, N. & Tomita, Y. (2010). Interactions between type of instruction and type
of language feature: A meta-analysis. Language Learning, 60, 263–308.

VanPatten, B. (1988). How juries get hung: Problems with the evidence for the
effects of a focus on form. Language Learning, 38, 243–260.

VanPatten, B. (2016). Why explicit knowledge cannot turn into implicit

knowledge. Foreign Language Annals, 49, 650–657.

VanPatten, B. & Fernández, C. (2004). The long-term effects of processing

instruction. In B. VanPatten (Ed.), Processing instruction: Theory, research, and
commentary (pp. 273–289). Mahwah, NJ: Erlbaum.

VanPatten, B. & Rothman, J. (2014). Against “rules.” In A. Benati, C. Laval, &

M. J. Arche (Eds.), The grammar dimension in instructed second language
acquisition: Theory, research, and practice (pp. 15–35). London: Bloomsbury.

VanPatten, B., Keating, G. D., & Leeser, M. J. (2012). Missing verbal inflections
as a representational issue: Evidence from on-line methodology. Linguistic
Approaches to Bilingualism, 2, 109–140.

272
 Following Up
1. Explicit and implicit knowledge do not just differ in terms of whether the
knowledge is verbalizable or not. The two knowledge types also differ in
terms of substance (what that knowledge actually is). Read either VanPatten
and Rothman, “Against ‘rules’” or VanPatten, “Why explicit information
cannot become implicit knowledge” (see References and Readings) and
prepare a 100-word summary (or less) on implicit knowledge. Then conduct
your own research to see if you can find different definitions of the nature of
(implicit) linguistic knowledge.

2. Explain why you agree or do not agree with the following statement:
Because instruction in grammar does not affect ordered development, there
is no need to worry about grammar in language teaching.

3. A recurring theme in L2 research concerns the types of assessments we

use to measure L2 knowledge and ability. Conduct a search on the following
types of tasks in order to define each and offer examples, explaining what the
task attempts to measure.

grammaticality judgments

truth-value tests

sentence-combination tasks

information-gap tasks

self-paced reading.

4. One of the findings of long(er)-term research is that the effects of

instruction seem to fade after time. In the handful of studies we have that
actually measure development or knowledge many months later or up to a
year later, we see marked declines in performance on tests. In most cases,
learners are back where they were before instruction. Make an argument for

273
 why this would be the case. (Hint: What are the data for language
acquisition? How does language get in your head?)

5. Pavesi’s research suggests there is a difference in the type of language

learners are exposed to in and out of the classroom, especially if classrooms
are dealing with literacy. Select a reading from a second-year or
intermediate-level text in the language you teach or will teach. At the same
time, record an hour of conversation between two people in the same
language (or use dialogue from TV or a movie if you don’t have access to
speakers; reality TV would be best). For both data sets, analyze them for the
following.

length of sentence (i.e., mean number of words per sentence)

ratio of sentences with embedded clauses to single-clause sentences

percentage of person–number marking on verbs (i.e., 1st singular, 2nd

singular, 3rd singular, and so on)

something else you would like to look at.

Present your findings and decide whether indeed the language in the two
environments is different.

6. Imagine a teacher or someone else saying, “Of course instruction makes a

difference. I see it all the time in my students.” What would you say to this
person?

7. Conduct some bibliographic research on the teaching of pronunciation or

the teaching of vocabulary. Try to find at least five studies. How do they
measure the effects of instruction? Are the issues raised in this chapter valid
for those areas of instructed SLA as well?

274
 7
What Role Does Explicit Learning
Play in L2 Acquisition?
◈

In Chapter 6, we concluded that instruction on the formal properties of language

does little to alter acquisition. In this chapter, we look at a somewhat related issue,
namely, the role of explicit learning. As we will see, the issues are different in that
in Chapter 6 we looked at what effect, if any, outside interventions by teachers
and textbooks have on acquisition. In this chapter, we focus on what the learner
does during comprehension.
But before we get into the topic, let’s begin with something that is generally
accepted by researchers regardless of their theoretical orientations: that the result
of L2 acquisition is an implicit system. By implicit we mean that learners cannot
articulate what is the content of the mental representation we call language. In this
book we tend toward a generative perspective on language but what we claim is
equally true if you take, say, a usage-based perspective: Whatever language is, it
exists as an implicit system in the learner’s mind/brain. You may recall from
Chapter 6 that explicit knowledge is what a person can articulate about language
such as “To make a verb in the past in English add an -ed.” Generally, such
knowledge comes from textbooks, teachers, internet sites, and perhaps the
learner’s own conscious observations about how he or she thinks the language
works (e.g., “Oh. I have to put the adverb here and not there!” “Oh. Looks like I
have to insert a do to make a question.”). Implicit knowledge, on the other hand,

275
is knowledge that cannot be articulated, such as why re- can combine with paint
and write to make the verbs repaint and rewrite but can’t combine with sleep and
pet to make *resleep and *repet. Another example would be knowledge that you
can make a progressive out of verbs like eat and die (e.g., “The man is eating.”
“The man is dying.”) but not seem and know (e.g., “*The man is seeming sad.”
“*The man is knowing me.”)
What is more, if a learner can articulate a rule in some way, this does not
mean that what the learner says is how language is represented in the mind/brain.
Typically, rules found in textbooks and given by teachers are not psychologically
real – meaning that what the rule says is not what winds up in people’s heads. So
when we talk about what gets learned in L2 acquisition, we are not talking about
rules in the classic sense but something quite different. Here’s an example given
by Bill VanPatten in several of his publications. In Spanish, objects of verbs are
sometimes marked by the tiny word a. Textbooks and teachers refer to this as
“personal a” because it generally marks some kind of person as an object of a
verb. Here is a typical rule that learners may read.

In Spanish, when the direct object is a person, it is preceded by the

preposition a. The personal a is not used when the direct object is not a
person or is an animal for which no personal feelings are felt.

Now let’s look at some samples of language.

(1)

a. María conoce a Juan.

‘Maria knows John.’

b. María conoce la materia.

‘Maria knows the subject.’

This rule seems to work in (1a) and (1b): Juan is a person and the name is marked
with an a, but subject is not a person so there is no a. But what happens with the
following examples?

276
 (2)

a. ¿Conoces un buen médico?

‘Do you know a good doctor?’

b. Tengo una hermana.

‘I have a sister.’

(3)

a. El camion sigue al carro.

‘The truck is following the car.’

b. El chico asustó al coyote.

‘The boy scared the coyote.’

In (2a) and (2b), both doctor and sister are people but no a is used to mark
them as objects of the verb. And in (3a) and (3b) neither a car nor a coyote is a
person and yet the a is used. In short, the pedagogical textbook rule that students
learn does not reflect what winds up in the mental representation of both L1 and
L2 learners.
It is worth reminding ourselves of additional facts about acquisition that
contribute to the understanding that learners are building an implicit system over
time. As we have seen in this book, acquisition is ordered and piecemeal. There
are stages of development for syntactic and morphological properties of language.
Ordered development cannot be traced to teaching and instructional
environments. In addition, learners come to know more about language than what
they are exposed to, what we have called poverty of the stimulus (POS) situations
in previous chapters. With POS situations, learners show evidence of knowing
things that are possible about language, as well as things that are not possible, that
they haven’t been exposed to. And as we saw in Chapter 6, the effects of trying to
teach the formal properties of language are severely limited in that they do not
affect ordered development in any substantial way. In short, learners are doing
things to and with language that they are not aware of. So if learners wind up with

277
an implicit mental representation of language, and that representation bears no
resemblance to textbook rules or rules the learner gets from external sources, how
does that system get in the learner’s mind/brain? Does explicit learning have
anything to do with the development of that system? And just what is explicit
learning? These are the questions that underlie the discussion in this chapter.

278
 A Brief Discussion of “Learning”
It may seem odd to the reader of this book, but learning is an ill-defined concept
in L2 research. That is, researchers often assume a definition of learning but don’t
lay out their idea of just what learning is. In other words, researchers know that
learning happens, but they often aren’t specific about what processes the learner is
undergoing to learn something. The issue of explicit learning is also muddled
because we find that researchers use the terms explicit teaching and explicit
learning interchangeably. Explicit teaching involves instructors or online delivery
systems overtly teaching learners such things as “Verbs must agree with their
subjects. This is how it happens in Spanish” and then providing learners with
practice to “internalize” what has been explicitly taught. Explicit learning is
presumed to happen during the phase of practice in which learners’ overt attention
is on learning the rule in question. In a 2015 essay, for example, Nick Ellis,
defines explicit learning as form-focused instruction. In that same essay, he
describes implicit learning as something that happens to the learner internally by
way of mechanisms involved in general learning (e.g., frequency tabulations of
occurrences in the input, blocking). So, for Ellis, implicit learning is internal to
the learner and controlled by that learner, but explicit learning is external to the
learner and controlled by something else. However, relating explicit learning to
explicit teaching environments means that researchers often leave what is actually
learned and how it is learned undefined. Although we can define explicit teaching
– and researchers regularly do when they describe an experiment – explicit
learning is often simply assumed. In other words, researchers assume a clear
correspondence between what is taught and what is learned. The underlying
processes and mechanisms for explicit learning are generally left unspecified.
If we turn to psychology, learning involves the development of new
cognitive structures. For example, a child learns over time what a “bird” is,
developing a cognitive structure that has information in it such as “has feathers,”

279
“has wings,” “can fly,” and so on. Later, that cognitive structure is altered when
the child encounters penguins, emus, and ostriches, for example. But exactly how
does the cognitive structure develop in the first place? Clearly, from processing
stimuli in the environment. The child hears that the robin on the tree branch is a
bird, that the canary inside the cage at Aunt Tracy’s house is a bird, and that the
bald eagle on a sign is a bird. From these encounters, learning mechanisms
(unspecified for now) begin to tally relevant characteristics and from these tallies
the more abstract concept of “bird” emerges. Thus, when a child sees a blue jay
for the first time, that child says “Bird!” The child forms some kind of abstract
concept of “bird” based on various interactions with birds. So learning happens as
a result of processing data from the environment. Exactly what that processing is
we will leave for books on psychology.
Is language acquisition any different? Yes and no. To create a mental
representation of language, learners have to process data from the environment.
But in the case of language acquisition, those data aren’t rules but actual streams
of speech (either spoken or signed). The learner somehow processes the data, and
then internal mechanisms cull from those data to build an implicit system. We
will see a bit later what, exactly, we mean by processing, but for now the question
is this: To what extent do learners process linguistic data in the input explicitly?
Theoretically, there are three positions on the matter: Explicit processing of the
data is necessary, explicit processing is not necessary but beneficial, explicit
processing is neither necessary nor beneficial. (Note that from here on in we will
use explicit processing and explicit learning interchangeably.) The issue with
these positions is that it is not always clear what a researcher means by “explicit
learning.” Even though we have defined it here as the explicit processing of input
data (given the fundamental role of input) it’s not clear that all researchers do so.
As we said above, explicit learning is often inferred by researchers when (1)
explicit teaching is involved (see above) and/or (2) learners can state a rule or
articulate what they have learned (i.e., demonstrate explicit knowledge). What is
more, researchers often don’t make reference to how learners process input data

280
or to what they process in the data. This makes the task of discussing the role of
explicit learning a bit challenging.

To create a mental representation of language, learners have to process

data from the environment. But in the case of language acquisition, those
data aren’t rules but actual streams of speech (either spoken or signed).

281
 Consider this …
Think of abstract concepts that you have in your head; for example,
liberty, justice, comprehension, mathematics, and the universe. If pushed,
you could probably eke out some kind of definition as you reflect on what
these concepts are. But how did they get in your head? Did you memorize
a definition? Did you focus on the word liberty and then go about learning
it purposefully? If you are like most people, concepts such as these
“sneak” into your head through exposure. Try to think of concepts you
deliberately learned (i.e., concepts such as “hate” not activities such as
“baking a cake”). How much non-explicit processing do you think
happens on a daily basis?

To date, the field of SLA has focused on two broad questions related to
explicit learning. The first is whether explicit knowledge can somehow turn into
implicit knowledge. The second is whether explicit knowledge or explicit learning
can influence the development of implicit knowledge. As we get into these issues,
we will see how sometimes they get a bit confused.

282
 Can Explicit Knowledge Turn into Implicit
Knowledge?
We begin by noting that the first contemporary distinction between the roles of
explicit and implicit knowledge in L2 acquisition was articulated by Stephen
Krashen in the late 1970s and early 1980s. Krashen distinguished between
learning and acquisition. In Krashen’s terms, learning involves intentional focus
on the learning of rules through some kind of explanation and practice. The
learner is consciously aware of what he or she is doing and the purpose of the
effort is to learn the grammar (rules, sounds, and so on). The result is some kind
of explicit knowledge. At the time of Krashen’s writing, he was envisioning
learning as part of traditional language classes (and classes we still see today):
Teachers and textbooks present, learners practice, and performance is assessed via
tests to see if the learners have “internalized” what was presented and practiced.
Acquisition, on the other hand, happens as a by-product of input processing in
communicative contexts. To state this another way, acquisition happens through
sustained interaction with comprehensible input. The learner is not actively
focused on learning rules, sounds, and so on but instead is actively attempting to
understand what someone is saying (or signing). No conscious awareness of the
formal properties of language is involved. Krashen argued that there is no
relationship between explicit knowledge and implicit knowledge and, most
importantly, that explicit knowledge cannot become implicit knowledge. If we
return to our example from Spanish with the so-called “personal a,” Krashen
would argue that the rule that learners are given cannot turn into the implicit
knowledge that underlies the examples we saw in (1a) through (3b). Krashen’s
position is what’s called a no interface position. That is, he argued that the two
knowledge systems are distinct from each other and do not interact.

283
 In the 1970s and early 1980s, Stephen Krashen argued for a “no interface”
position; namely that explicit and implicit knowledge systems are distinct
from each other and do not interact.

Early on, Krashen’s distinction between learning and acquisition was met
with criticism, as was his no interface position. But since the 1970s, a good deal
of research supports the claim that explicit knowledge cannot turn into implicit
knowledge. Bonnie Schwartz, for example, provided a solid updating of the
distinction between explicit and implicit knowledge in her 1993 essay. Instead of
talking about learning and acquisition, she made the distinction between
competence on the one hand and learned linguistic knowledge on the other, as
well as between performance and learned linguistic behavior. She used
competence in the Chomskyan sense and in the way we use mental representation
in this book, whereas learned linguistic knowledge is the kind we typically find
that learners have based on textbook rules. Performance is what learners do with
their competence whereas learned linguistic behavior is what they do with their
learned linguistic knowledge. She argued, from a generative perspective, that
competence and learned linguistic knowledge are qualitatively different and one
cannot turn into the other. She also argued that performance and learned linguistic
behavior are different, and one does not lead to the other. Important here is that
competence (and the performance based on it) is not learned in any explicit way
but can only be learned through the processing of input coupled with language-
specific internal mechanisms such as Universal Grammar. For this reason, learned
linguistic knowledge cannot turn into competence because competence can only
develop as a result of processing input. (If you want to read about a concrete
example of what Schwartz meant, see Bill VanPatten’s 2016 essay cited in
Chapter 6. It also appears in the references for this chapter.)
Although we say that most scholars within and without the generative
tradition adopt the position that explicit and implicit knowledge do not interact,
there is one exception: the position adopted in skill acquisition theory – although

284
as we will see in a moment, there is some misunderstanding of what skill theory
actually claims.
Robert DeKeyser is the name most associated with skill acquisition theory
(2014). Under this theory, skills develop out of some kind of practice. To begin,
adults typically have skills in a wide range of areas, such as driving, cooking, or
playing a sport, and most of these are skills acquired in adolescence or adulthood.
DeKeyser and his colleagues argue that L2 acquisition should be seen as another
instance of skill learning. This means that the general stages of acquiring any skill
are also relevant for L2 learning. These stages involve three kinds of knowledge:
declarative knowledge, procedural knowledge, and automatized knowledge.
The most common way that people explain each of these types of knowledge
is with respect to driving. When someone first learns to drive, the person teaching
them shows them where the important parts of the car are and explains the basic
process of driving. A first-time driver might be told which pedal is the gas pedal,
which pedal is the brake, how the gearshift works, where the turn signals are, how
to execute smooth turns, how to keep alert to other drivers, how to stop correctly,
among many other things. All of this information helps learners develop some
kind of declarative knowledge. Declarative knowledge is not the most important
type of knowledge for executing skills, but this isn’t because the knowledge is
inherently complicated. For example, it’s easy enough to understand that if you
want to signal a left-hand turn, you press the lever on the left side of the steering
wheel down. What matters, though, is the ability to do this when you approach a
left turn. This type of knowledge is called procedural knowledge. Procedural
knowledge is knowledge how, and it encompasses the ability to execute a task in
some way. Procedural knowledge emerges from actually doing the task that you
are supposed to do. So while you might have declarative knowledge about the
necessity of putting your turn indicator on when making a left turn, procedural
knowledge develops only when you actually do this when making a left turn.
Procedural knowledge lies on a continuum from partly automatized to fully
automatized. Skills that are automatized can be performed quickly and accurately

285
with little to no attention, and with few errors. So as novice drivers improve,
reaching for the lever that activates the turn signal becomes second nature, and
instances of moving it in the opposite direction from the turn they want to signal
also decrease. Skills that are fully automatized aren’t necessarily always
performed perfectly – even skilled drivers sometimes forget to signal, or indicate
a right-hand turn when they intend a left-hand one. Skill acquisition theory has a
long history and a wide research base within the field of cognitive psychology,
and there is evidence that many skills – driving, swimming, cooking, and so on –
involve the interaction between declarative, procedural, and automatized
knowledge.
Now let’s consider how skill acquisition theory applies to L2 acquisition.
Skill acquisition theorists assume that linguistic skills are like any other kind of
skill. In other words, DeKeyser and his colleagues do not assume that linguistic
knowledge or skill is any different from any other kind of knowledge or skill.
Because of this assumption, it makes sense that learners would benefit from
explicit rules and practice for the development of L2 skills. Explicit rules are the
equivalent of declarative knowledge (even though they don’t have to be), and
practice is what is necessary for declarative knowledge to become procedural
knowledge. More practice over time leads to the automatization of knowledge.
The naïve reader might assume that under skill acquisition theory, explicit or
declarative knowledge turns into procedural and automatized knowledge – and
that the latter are equivalent to implicit knowledge. It is not clear that skill
theorists would accept this. While there is a relationship between automatized and
procedural knowledge, skill theorists don’t really claim that declarative
knowledge turns into procedural knowledge. The reader may notice that we used
the term “emerge” as in “procedural knowledge emerges.” It is not clear in skill
theory what exactly procedural knowledge is but it is qualitatively different from
declarative knowledge and is not “derived” from it. What is more, it would be
wrong to conclude that under skill theory, explicit knowledge turns into implicit
knowledge. These terms aren’t even used in skill theory. Instead, learners

286
somehow develop procedural knowledge that subsequently gets automatized. As
should be clear from the discussion above, skill acquisition theory is a theory that
explains how humans learn to perform behaviors. Because of that, the terms
declarative and procedural knowledge are a bit of a red herring – skill acquisition
theory isn’t really a theory of knowledge. It’s a theory of behavior.

It would be wrong to conclude that under skill theory, explicit knowledge

becomes implicit knowledge. These terms aren’t even used in skill theory.

DeKeyser and other researchers who subscribe to skill acquisition theory do

not claim that skill acquisition theory accounts for all of SLA, or that it is the only
process by which L2 acquisition takes place. Instead, DeKeyser limits skill
acquisition theory to adult instructed L2 learners who are relative beginners
learning relatively simple structures and who have a relatively high level of
linguistic aptitude. DeKeyser doesn’t think that all of these factors have to be in
place. Similarly, he suggests that low-aptitude adults will have a harder time
forming the kinds of declarative representations that would enable them to
automatize this knowledge. And he doesn’t think skill acquisition theory accounts
for language learning at levels beyond beginning-level learners. Advanced
learners are exposed to more input, which makes it more likely that they will be
able to engage the input and, thus, that implicit learning can (will) happen.
DeKeyser probably takes the strongest position in favor of declarative knowledge
of researchers working in SLA today, yet even he doesn’t argue that explicit
knowledge is the only – or the most important – factor in SLA. Rather, he argues
that in specific contexts, declarative knowledge is important because it can
facilitate the development of procedural and automatized knowledge.

287
 Consider this …
We have limited our focus on what is acquired in L2 acquisition to the
mental representations for language. We have not talked much at all about
linguistic skill. In your mind, what is skill? Is it the ability to
communicate? If so, what does this mean? And what is the relationship
between skill and mental representation as we have talked about it in this
book?

We conclude this section by saying that few scholars, if any, believe that
explicit knowledge turns into implicit knowledge. The question for most scholars
is whether explicit knowledge can somehow influence the development of
implicit knowledge. This would suggest that somehow explicit knowledge is used
during the processing of input and thus explicit learning is implicated in the
development of implicit knowledge. We turn our attention to this position in the
next section.

288
 Can Explicit Knowledge/Learning Influence the
Development of Implicit Knowledge?
Probably most researchers agree with the premise that the development of implicit
knowledge can be influenced by explicit knowledge or explicit learning. But
researchers are not in agreement about what this means. Some researchers, like
Rod Ellis, believe in what is called a weak interface between explicit and implicit
knowledge. According to these scholars, explicit knowledge can influence the
development of the implicit system in some way. However, the mechanisms for
how this happens are not clear, and the evidence is the same as the evidence we
reviewed in Chapter 6 on the effects of instruction. That evidence is problematic
at best. Because the research on instruction – and thus the research on explicit
knowledge and learning affecting the implicit system – is biased toward testing
explicit knowledge, we have no good evidence for the basic claim that explicit
learning somehow affects the development of implicit knowledge. The best
reasonable conclusion is that explicit learning affects the development of explicit
knowledge. Even if the research clearly suggested that explicit learning affects
implicit knowledge, there remains the problem of the learning mechanisms
themselves. If, as Rod Ellis and others claim, there is an interface between explicit
knowledge and implicit knowledge, just what is that interface? How does it work?
What learning mechanism “connects” the two knowledge systems? What does it
mean for one system to “influence” the other when we know they are qualitatively
different? Researchers who argue for such an interface really haven’t addressed
these questions; instead, they assume that because there are two knowledge
systems, they influence each other. In a sense, the assumption that this interface
exists may reflect the socio-cultural context in which much L2 instruction takes
place: classrooms. The assumptions work like this: We know learners are taught a
lot of explicit knowledge about the second language, so it must be useful to them
in some way. If it weren’t, we wouldn’t spend so much time focusing on it. This

289
reasoning is circular, and, as we’ve seen, there isn’t much evidence that
instruction influences the development of the underlying mental representation for
language.

Some researchers believe in a “weak interface” model; that somehow

explicit knowledge can influence the development of implicit knowledge.
However, the mechanisms for how this happens are not clear.

One of the most significant hypotheses to emerge regarding explicit learning

is the Noticing Hypothesis. The Noticing Hypothesis was formulated by Richard
Schmidt in a 1990 publication. The hypothesis states that learners have to notice
(i.e., be aware of) formal features in the input in order for these forms to become
intake and subsequently be acquired. The Noticing Hypothesis does not state that
this (conscious) awareness becomes intake. Rather, the claim is that the language
acquisition mechanisms are more likely to make use of data in the input if the
learner is somehow aware of those data. If we go back to our example of the
personal a in Spanish, the claim under the Noticing Hypothesis is that somehow
learners must consciously register the a when they encounter it. In other words, if
they are to acquire personal a, they can’t “skip over it” in the input; it has to
somehow enter their focal attention during the processing of input.
What is important to understand about the Noticing Hypothesis is that what
gets noticed can only be detectable surface features of languages such as verb
endings and what they mean or noun endings and what they mean or, in our
example, something like the personal a. In its original form, the Noticing
Hypothesis was not meant to apply to complex syntax or even some superficially
simple syntactic operations. Thus, researchers often talk about “noticing” past-
tense endings, case endings, and plural markers, for example, but never about “the
conjugated verb in German is always in second place” or “object pronouns in
Spanish never appear between an inflected verb and an infinitive.” What is more,
the Noticing Hypothesis is limited to what is observable in sentences and not what

290
is hidden or impossible in a language. In our example of the personal a, is it
possible to notice when a is not used? How do you notice things that aren’t there?
This observation makes the Noticing Hypothesis at best limited in its application
to L2 acquisition. Let’s go back to an example from the introduction to this
chapter to better illustrate the limited scope of the Noticing Hypothesis. Learners
can easily “notice” in the input that re- can combine with verbs to mean
something like “do again.” They hear repaint, redo, rewrite, redesign, and so on.
And yet, at some point they wind up with a linguistic system that tells them that
*resleep, *repet, *rebake, *resneeze, and *redrink are not possible. How does this
happen? How did the noticing of re- with some verbs lead to the
ungrammaticality of re- with other verbs? It can’t be because learners just haven’t
encountered them yet. After all, there are grammatical things we haven’t
encountered, yet we know they are grammatical. In the re- example, you may not
have encountered reinvite but when you hear it, you know it sounds OK (even
though it’s not in the dictionary!). Language is full of things that we know are
possible and impossible that we can’t have acquired because we “noticed” them in
the input.

In its original form, the Noticing Hypothesis was not meant to apply to
complex syntax or even to some superficially simple syntactic operations
– only to detectable surface features of language such as verb and noun
endings and what they mean.

What is more, noticing is at best a fuzzy concept. Over the years, Richard
Schmidt attempted to clarify just what noticing was, but in the end, the concept
continued to elude an easy definition, which in turn led to difficulty conducting
empirical research. Just how does the researcher operationalize a construct that is
not well defined? What has happened is that a number of researchers have used
the Noticing Hypothesis to motivate research on instructed L2 acquisition
(Chapter 6). In such research, they might compare one group who has prior

291
explicit knowledge of a formal feature of language with one that does not. The
learners are then exposed to the same input, and if the group with prior knowledge
performs better on a post-test on that formal feature, the researcher concludes that
noticing helped. That is, learners are able to somehow deploy explicit knowledge
during the processing of input, resulting in greater gains than the group that did
not get explicit knowledge. If the shortcomings of this kind of research are not
clear, we invite the reader to review the criticisms of instructed L2 research in
Chapter 6. But again, the major obstacles in this research are (1) demonstrating
that “noticing” happened (that is, that something explicit happened during input
processing) and (2) relying too much on tests on which learners can use explicit
knowledge in some way or another. (See also some of the problems in using
noticing to discuss interaction, Chapter 3.)
Some recent research is beginning to use online methods such as eye-
tracking to see where learner attention is directed during processing. However,
this research has not yet focused on the intersection of noticing and acquisition –
that is, how learners have to notice something in order to acquire it. Instead, it has
largely focused on whether learners can detect things they have already been
exposed to in the input or might have learned from a book, from a teacher, or
from the Internet. For example, in a study called “Coming Eye-to-Eye With
Noticing,” published in 2013, Patti Spinner, Susan Gass, and Jennifer Behney
examined second-semester university learners of Italian and whether their eye
movements indicated noticing cues on articles and nouns in order to determine
what gender a noun is. Participants read sentences such as Poi la carne è troppo
… on a computer screen and off to the side was a box with four adjective choices:
grassa, grasso, grasse, and grassi (i.e., ‘So the meat was too … fatty’). They
tracked their unconscious eye movements to see where they went in order to
determine which version of the adjective to select. They found that learners’ eye
movements went to both articles and nouns to determine gender, not just to the
nouns, for example. However, this is not a study of how learners acquire adjective
agreement or the gender of nouns to begin with. This use of eye-tracking is more

292
like a test of whether they have gender agreement and what cues they use to make
that agreement in a very conscious way – as they might do when taking a paper-
and-pencil test with multiple-choice answers. Thus, it is too soon to tell whether
this line of research will shed any light on the fundamental question of whether
learners have to notice (explicitly process) formal features in order to acquire
them. We will see later why there are additional problems with any strong version
of noticing when we review just what is involved in processing a sentence.

293
 Consider this …
We will see in a bit what some possible criticisms of the Noticing
Hypothesis are. But for now, ask yourself this question: Is it possible to
notice something that’s not in the input? In other words, if you can only
notice what you see or hear, how do you come to know what is impossible
in a given language?

There are some scholars who have been critical of the Noticing Hypothesis,
among them John Truscott and John Williams. In a 1998 essay, Truscott argues
that it makes little sense to say that learners have to notice formal features in the
input. His argument rests on two ideas. The first is along the lines we’ve laid out
here: The nature of the mental representations learners acquire is too complex and
abstract to be clearly related to the surface features of the input. The second is
along the lines we laid out in Chapter 6: There is no convincing evidence that
information about language influences the development of mental representations
for that language. We’ll return to Truscott’s first argument at the end of this
chapter with a detailed example. Williams, in a series of essays and empirical
studies, has argued that noticing involving some kind of awareness is not
necessary for acquisition. Coming from a psychological perspective and not a
linguistic perspective (such as Truscott’s), Williams suggests that we first need to
understand how implicit learning works and what exactly is learned implicitly.
Only after doing that can we determine to what extent any kind of explicit
learning contributes to acquisition.

We may first need to understand how implicit learning works and what
exactly is learned implicitly. Only after doing that can we determine to
what extent any kind of explicit learning contributes to acquisition.

294
 Although it was not intended to address the relationship between explicit
learning and processing, there is some research that provides some evidence that
helps us answer this question. In a series of studies, Bill VanPatten and his
colleagues have shown that explicit knowledge does not necessarily get deployed
during the processing of input and that the question of whether or not it does is
much more complicated than we might think. In a 2013 study, VanPatten and his
colleagues were interested in the intersection of the first-noun strategy and the
processing of different structures in Spanish, French, German, and Russian. The
first-noun strategy claims there is a universal tendency in learners to begin the
process of acquisition by assuming the first noun or pronoun they encounter when
comprehending a sentence is the subject/agent (see Chapter 4 for some
discussion). VanPatten and his colleagues identified four learning problems in the
languages they were interested in:

Spanish: object-verb-subject (OVS) sequences are problematic as in Lo ve

la chica (lit: him sees the girl, ‘The girl sees him’). The tendency would be
for learners to think the lo is the subject/agent and misinterpret the
sentence as ‘He sees the girl.’

German: OVS sequences are problematic as in Den Mann hört die Frau
(lit: the man-A C C hears the woman-N O M ‘The woman hears the man’).
The tendency would be for the learners to think Den Mann is the
subject/agent and misinterpret the sentence as ‘The man hears the
woman.’

Russian: OVS sequences are problematic as in Собаку слышит кошка

(lit: the dog-A C C hears the cat-N O M ‘The cat hears the dog’). The tendency
would be for the learners to think Собаку is the subject/agent and
misinterpret the sentence as ‘The dog hears the cat.’

French: causative structures with faire as in Le père fait lire la lettre à la

mère (lit : the man makes to read a letter to the mother ‘The man makes
the mother read a letter’). The tendency would be for the learners to think

295
 the father is reading the letter and to misinterpret the sentence as
something like ‘The father is reading a letter to the mother.’

In their experiment, half of the learners in each language got information on the
problem (e.g., lo is not a subject pronoun so don’t interpret it as one) and half did
not. During the treatment, learners were exposed to mixtures of sentences (e.g., in
German some were OVS and some were SVO) and were asked to match what
they heard to one of two pictures (e.g., a man listening to a woman and a woman
listening to a man). All conditions and languages were matched for vocabulary as
well as problematic versus non-problematic sentences to process. The researchers
measured how long it took learners to begin processing sentences correctly. They
found that for the Spanish and Russian groups, information about processing
strategies did nothing to influence the rate at which these participants began to
process sentences accurately, whereas the participants in the German and French
groups who got the information on the processing problem prior to hearing
sentences began to correctly process the problematic sentences sooner than those
who did not get such explanation. So it appears that explicit processing helped in
two of the four cases. The question of course is what was different about those
cases. The most interesting cases are the German and Russian groups. Both
German and Russian mark case on nouns but do it differently: German marks the
article whereas Russian (which doesn’t have articles) marks the noun. Why would
this make a difference? It’s not clear why it would. The point is that the question
of whether explicit processing (deploying explicit information during real-time
comprehension) makes a difference or not is much more complicated than some
claims would lead us to believe. What is also important about this particular kind
of research is that VanPatten and his colleagues do not rely on tests of knowledge
but on tests of processing (i.e., sentence interpretation). Although this is a step in
the right direction, we cannot claim here that such research is definitive. It is, at
this point, suggestive and warrants continued investigation.
Other researchers have begun to use online measures to explicitly test the
relationship between explicit information and input processing. In a 2015 study

296
(see the Exemplary Study summarized in this chapter), Sible Andringa and Maja
Curcic used eye-tracking to investigate whether participants who were provided
with explicit information interacted with input differently than participants who
did not receive explicit information (similar to the VanPatten et al. study just
reviewed). They investigated whether novice learners who were told that animate
nouns were marked with al in Esperanto were able to use this information to
attend to the relevant cues in the input. They hypothesized that these participants
might have been more likely to attend to the relevant cue (the presence or absence
of al) and use this knowledge to help them process sentences. Andringa and
Curcic provided their participants with input in a semi-artificial version of
Esperanto (semi-artificial because Esperanto doesn’t actually mark animate direct
objects with al). In the input their participants were exposed to, animate direct
objects were marked with the preposition al, and inanimate direct objects were
not marked with a preposition. The authors hypothesized that if learners were
sensitive to this distribution in the preposition al, they would use the presence of
the preposition to predict that the upcoming noun would be animate. They
measured this by having learners hear a sentence while pictures were displayed on
a screen. The authors hypothesized that if participants were aware of the
grammatical function of al, when they heard the preposition, their eyes would
unconsciously move toward the picture of an animate noun on the computer
screen in front of them. In short, if explicit information makes a difference in
whether participants relied on al to anticipate the direct object, the results should
show that participants who received explicit information had a higher proportion
of looks to the correct picture when they heard al. These results, however, were in
the opposite direction to what the authors predicted. Although the group who
received explicit information processed the input differently than the group that
did not receive explicit information, this difference didn’t make them more likely
to look at the right picture sooner than the implicit group. Instead, it just made
them more indecisive in their processing of the differential object marking
sentences. In other words, it doesn’t seem like explicit information helped these

297
participants use object marking as a disambiguating cue during real-time
processing of input.

Exemplary Study

Andringa, S. & Curcic, M. (2015). How explicit knowledge affects online

second language processing: Evidence from differential object marking
acquisition. Studies in Second Language Acquisition, 37, 237–268.

298
 Participants
51 L1 Dutch speakers with no prior knowledge of either Spanish or
Esperanto

2 groups: the explicit group (n = 26) got explicit information and an

input treatment task, and the implicit group (n = 25) got only the input
treatment task

299
 Materials
Input treatment based on a semi-artificial version of Esperanto:

◦ vocabulary: 10 animate nouns, 10 inanimate nouns, 19 transitive

verbs, 6 intransitive verbs, 5 adjectives

◦ syntax: SVO word order

◦ included differential object marking by marking animate direct

objects with the preposition al

◦ 104 black-and-white pictures that represented everyday objects

and activities

◦ 104 sentences that described the pictures and were presented

auditorily

◦ 52 sentences were contexts for differential object marking; 26

had animate direct objects and 26 had inanimate direct objects

◦ explicit group: a differential object marking rule explanation in

Dutch using two example sentences

Visual word eye-tracking task as a measure of implicit processing:

◦ participants picked the image that matched the last word in a

recorded sentence

◦ pictures were presented in pairs; one quarter of them had two

animate nouns; one quarter had two inanimate nouns; one half had
an animate and an inanimate noun

◦ 36 experimental sentences; 18 with animate nouns and 18 with

inanimate nouns

300
 ◦ sentences with animate nouns and a choice between an animate
and inanimate picture should have evidence of looks to the
animate picture when encountering the preposition al

◦ no anticipatory looks when the pictures match in animacy

Auditory grammaticality judgment task as a measure of syntactic

knowledge:

◦ participants heard a sentence describing a picture with a

transitive activity

◦ rated the sentence as grammatical or ungrammatical

◦ stated whether the decision was based on feel or rule

◦ 8 grammatical sentences (four with al and an animate object;

four with no preposition and an inanimate object)

◦ 8 ungrammatical sentences (al and an inanimate object; no

preposition and an animate object)

301
 Results
The group that got explicit information did significantly better on the
grammaticality judgment task than the group that got only input.

Both groups performed equally well in choosing the correct picture in

the eye-tracking task, indicating that both groups had understood the
treatment equally well.

Anticipatory looks

◦ Implicit learners showed a gradual increase in looks toward the

correct picture.

◦ Explicit learners showed a decrease in looks toward the correct

picture before they show an increase in looks toward the correct
picture.

◦ Neither group used the preposition al to predict the animate

noun.

302
 Putting Explicit Learning into Perspective
In this section we lay out what the underlying problem is in talking about the role
of explicit learning/processing (or even noticing) during acquisition. Let’s remind
ourselves of several things before continuing.

1. In the case of L2 acquisition we know that the result is the development of

an implicit, abstract, and complex system. The result is not the development
of an explicit system – even if learners get one along the way.

2. Learning is defined as processing external stimuli in some way that is

subsequently used by internal mechanisms to build a linguistic system. That
is, something “encodes” the data as the learner hears or reads it so that it can
be eligible as data for linguistic development. This is called input processing.

When we begin to scrutinize just what must be going on during acquisition,

we see just how complicated learning is. For example, let’s imagine a learner
hears a simple sentence such as ‘Megan is reading a book.’ At the same time, the
learner encounters this sentence as descriptive of an event in which Megan is,
indeed, reading a book. So the learner’s “job” is to map what he or she hears onto
the meaning conveyed by the event. This sounds easy enough, but let’s peel back
the layers on this example. Here’s what must happen for language to develop.

The learner tags ‘Megan’ as referring to the person in question. As a

result, the word ‘Megan’ is tagged with particular properties: it is a noun
([+N, −V]), it represents a proper name, and so on. At the same time, the
processing mechanism begins to project sentence structure. Given that
‘Megan’ is a noun, it is probably the subject so the learner’s processor
creates something like this upon hearing the word: [T P [S P E C Megan] [T
…]. This abstract information is how the system must encode that
something is a subject.

303
The learner then hears ‘is.’ The learner must determine what this little
word is. Is it part of ‘Megan’? Maybe a case marker to indicate nominative
case (e.g., Megan-is-N O M ). Is it a verb? If so, a main verb or an auxiliary?
Maybe the learner can’t make sense of it at all and it is simply dumped
from processing for the time being as the learner moves on to the word
‘reading.’ But for the sake of argument, let’s assume the learner does tag it
as a verb and successfully as an auxiliary (which is doubtful in the early
stages of acquisition, but again, we’re assuming here for the sake of
presentation). Let’s also assume the learner knows that the situation is
about what is going on now and not in the past or the future. The learner
encodes all of this information along with ‘is’: [−N, +V], [+aux], [+T],
[+pres, −past], [−plural]. Note: We’re not saying the learner does encode
all of this information nor all of it at once, we’re just indicating that if the
learner could encode all of the information, that is what the learner would
wind up with. Now the learner has projected something like this for the
sentence: [T P [S P E C Megan] [T is …].

The learner then hears ‘reading.’ The learner links this to the actual act of
reading, so the meaning of ‘reading’ is encoded. At the same time, the
learner knows this is the main verb, and given that ‘is’ has wound up in
the TP (tense phrase), this verb must be in the VP. Also given that ‘is’ is
marked with tense, the verb ‘reading’ must be non-finite. So in addition to
the meaning of ‘reading,’ the processor encodes the verb with this
information: [N, +V], [−T], [+progressive, −perfective]. At this point, the
processor has built this sentence structure in real time: [T P [S P E C Megan]
[T is [V P [S P E C –] [V reading …].

Then the learner encounters ‘a book.’ Here the processor must determine
whether there is one word ‘abook’ or two words ‘a book’ given the
prosody of how this phrase is pronounced. (Depending on how fast the
sentence is spoken, there are some conditions under which the learner
might think he or she is hearing ‘readinga’ and ‘book.’) If the learner

304
 successfully determines that these are two words, then the processor must
decide what ‘a’ is and what ‘book’ is. Given the strong stress on ‘book’
the processor probably decides that it is a content word and refers to the
book in the event under question. So the processor ascribes it the meaning
of ‘book’ as well as these features: [+N, −V], [+count], and possibly
others. If the learner’s processor figures out that ‘a’ is a determiner, the
processor has to figure out from the situation whether it is definite or
indefinite. For the sake of argument, let’s just say the processor correctly
marks ‘a’ as an indefinite determiner. This would mean it ascribes to this
function word the following properties: [+D], [−spec, −def]. At this point,
the processor builds the sentence structure this way: [T P [S P E C Megan] [T
is] [V P [S P E C –] [V reading] [D P [S P E C ] [D a] [N P book]]]].

Sentence computation is complete, the processor is happy, properties have

been ascribed in a way that makes sense, and meaning and form are
mapped onto each other.

The question now for the role of explicit and implicit learning is what parts
of this process happen explicitly and what parts happen implicitly? The reader
might now see the problem with attributing any strong or beneficial role to
explicit learning (as processing) once we delineate what must be learned. There is
no scenario we can think of where a learner engages conscious processes to
attribute properties such as [−N, +V], [−T], [+progressive, −perfective] to the
word ‘reading’ in our example. There is also no scenario where in real time a
learner projects syntactic constituents such as TP, VP, DP, and so on, as well as
the dependencies among them, consciously or with some kind of awareness. In
this example, we have left out other elements of the formal properties of language
that are present alongside the morphosyntactic properties we have highlighted
here, such as phonology (sound classes, syllable structure and syllable segments,
prosody or pitch, stress, and intonational contour of the sentence). So while
processing the syntactic units of the sentence to build structure, and tagging
meaning onto morphological units as well, the learner is also processing things

305
such as which parts of words and the sentence get strong stress and which get
weak stress, that English syllables can end in consonants, and so on. Clearly, the
assignment of properties to words and the building of sentence structure must
happen outside of awareness. If not, could the learner ever make it past the first
word? Conscious explicit processing of formal features would be exceedingly
slow and acquisition might not happen at all.

Clearly, the assignment of properties to words and the building of

sentence structure must happen outside of awareness. If this were not the
case, conscious explicit processing of formal elements would be
exceedingly slow and acquisition might not happen at all.

To be sure, our example is one in which the learner is successful with all
parts of the sentence – and our example is a relatively simple one. Imagine if the
sentence were more complicated (e.g., ‘I’m pretty sure Megan is reading a book
right now at a coffee shop somewhere.’). It is clear from research (see, for
example, Chapter 2) that learners get bits and pieces of language over time. They
are not successful at processing all parts of a sentence in the beginning and even
intermediate stages. This is because the learner is most likely struggling to get
basic meaning of content words from the input and mapping those onto the event
talked about. The learner might miss or not process function words and possibly
not map certain inflections onto their meaning and vice versa. Nonetheless, for
whatever the learner actually processes in the input at any given point in time, the
formal features of language and sentence structure must be processed outside of
awareness.
We have limited our example to a simple sentence in English. Let’s imagine
that the next sentence the learner hears is ‘She likes it very much.’ Now the
learner must not only process all the words and compute sentence structure as
before, but there is the added layer of antecedence-pronoun indexing. The learner
must somehow link in real time that ‘she’ is a pronoun and is indexed to ‘Megan’

306
while ‘it’ is a pronoun and is indexed to ‘book.’ What conscious or explicit
processing could be involved in this scenario? In the case of ‘she’ the learner has
to somehow encode the formal properties of the word (e.g., [+N,−V] [pro] [+fem]
and so on) as well as its index to Megan. The learner must do the same for ‘it’
(e.g., Megani is reading a bookj. Shei likes itj very much.’).
What all of this means, then, is that when we break down how learning
happens in a moment-by-moment analysis of comprehending a sentence – that is,
when we get into the weeds of processing the formal elements in language – we
get a clearer picture of the learning problem. Against a definition of learning that
includes processing and against the learning problem outlined here, we are better
able to judge whether or not language acquisition happens with explicit learning.
Our conclusion is that explicit learning is largely excluded from learning the
formal elements of language. Instead, its role might be to figure out “what
something means” and by “means” we actually mean what dictionary definition
can be given to that word or phrase. For example, “Oh. That word must mean
‘book’!” or “Oh. He must be referring to what happened yesterday!” and even
“Oh. ‘She’ refers to Megan!” So explicit processes and learning are used to
uncover meaning, but implicit processes and learning are used to map formal
properties onto that meaning. These are some of the issues that John Truscott
raised in his 1998 essay and we hope to have illustrated them here in an accessible
way.

307
 Consider this …
We have excluded from our discussion any role of the L1 in processing
(see Chapter 4). Is there L1 influence during the processing of input or
does the learner rely on universal properties of language processing?
Although an important question, does the presence of the L1 impact the
implicit processing of formal features of the L1 during real-time
comprehension?

With the above scenario laid out, we are also better able to sit back and
assess the various positions on explicit and implicit learning. The debate is, in a
sense, a result of three issues: (1) whether or not the researcher considers the
fundamental role of input as data for acquisition; (2) how the researcher views the
nature of language; (3) how the researcher defines learning and its relationship to
processing input data. It seems to us that the field has not reached consensus on
the role of explicit/implicit learning largely because researchers have not come to
consensus on these three basic issues. As long as the researcher believes that
language is something like what we see in textbooks, then explicit learning can
play a role in L2 acquisition. But if the researcher believes language to be
something much more abstract and complex, then that researcher might conclude
that explicit learning plays little to no role in acquisition.

What about Social Factors?

Throughout this book we have touched upon ways in which social factors
(context, relationships between speakers, and so on) impact acquisition
from an external perspective – largely the quantity and quality of input
and interaction a learner receives (or seeks out). We have also touched on
how social factors can relate to internal issues such as identity and its

308
impact on learner willingness to engage, for example. Regarding the role
of explicit learning as explicit processing, there is not much to say about
how social factors might bear upon the questions at hand. Processing as an
unconscious phenomenon that exists outside of awareness is largely
linguistic in nature and it is difficult to conceive of ways in which social
context or social factors would affect that processing. Once a learner is
engaged in actively trying to comprehend a sentence, the mechanisms for
processing kick in.

309
 Recap
In this chapter, we have touched on these major points.

What learning is in its general sense (the creation of new cognitive

structures).

Language acquisition must be linked to input in some way, such that a

major part of language learning is processing data in communicatively
embedded input.

Scholars have debated to what extent explicit learning is involved in

language acquisition, with two major positions taken: no interface and
weak interface.

The Noticing Hypothesis is one of the most widely discussed positions

that some kind of conscious registration of the formal properties of
language is necessary for acquisition.

A continued problem in the debate on the role of explicit learning is a lack

of what it means to learn language (i.e., what the processes are) and a
vagueness about what is acquired (i.e., what the formal properties of
language are).

How complicated and abstract the processing is of even a simple sentence

such as “Megan is reading a book,” suggesting a very limited role, if any,
for explicit learning during the acquisition of the formal properties of
language.

310
 References and Readings

Andringa, S. & Curcic, M. (2015). How explicit knowledge affects online second
language processing. Studies in Second Language Acquisition, 37, 237–268.

DeKeyser, R. (2014). Skill acquisition theory. In B. VanPatten & J. Williams

(Eds.), Theories in second language acquisition. New York: Routledge.

Ellis, N. (2015). Implicit AND explicit learning of languages. In P. Rebuschat

(Ed.), Implicit and explicit learning of languages (pp. 1–23). Amsterdam: John
Benjamins.

Ellis, R. (2005). Measuring implicit and explicit knowledge of an L2: A

psychometric study. Studies in Second Language Acquisition, 27, 141–172.

Hulstijn, J. (2005). Theoretical and empirical issues in the study of implicit and
explicit second-language learning. Studies in Second Language Acquisition, 27,
129–140.

Krashen, S. (1982). Principles and practice in second language acquisition.

Oxford: Pergamon Press.

Krashen, S. (1985). The input hypothesis: Issues and implications. London:

Pearson-Longman.

Rebuschat, P. (Ed.) (2015). Implicit and explicit learning of languages.

Amsterdam: John Benjamins.

Schmidt, R. (1990). The role of consciousness in second language learning.

Applied Linguistics 112, 17–46.

Schwartz, B. (1993). On explicit and negative evidence affecting and effecting

competence and linguistic behavior. Studies in Second Language Acquisition, 15,

311
147–163.

Spinner, P., Gass, S., & Behney, J. (2013). Coming eye-to-eye with noticing. In J.
M. Bergsleithner, S. N. Frota, & J. K. Yoshioka (Eds.), Noticing and second
language acquisition: Studies in honor of Richard Schmidt (pp. 227–246).
Honolulu, HI: University of Hawai’i National Foreign Language Resource
Center.

Truscott, J. (1998). Noticing in second language acquisition: A critical review.

Second Language Research, 14, 103–135.

VanPatten, B. (2016). Why explicit knowledge cannot turn into implicit

knowledge. Foreign Language Annals, 49, 650–657.

VanPatten, B., Borst, S., Collopy, E., Qualin, A., & Price, J. (2013). Explicit
information, grammatical sensitivity, and the first-noun principle: A cross-
linguistic study in processing instruction. The Modern Language Journal, 92,
506–527.

VanPatten, B. & Rothman, J. (2015). What does current generative theory have to
say about the explicit-implicit debate? In P. Rebuschat (Ed.), Implicit and explicit
learning of languages (pp. 89–116). Amsterdam: John Benjamins.

VanPatten, B. & Williams, J. (Eds.) (2015). Theories in second language

acquisition. 2nd edition. New York: Routledge.

Williams, J. (2013). Attention, awareness, and noticing in language processing

and learning. In J. M. Bergsleithner, S. N. Frota, & J. K. Yoshioka (Eds.),
Noticing and second language acquisition: Studies in honor of Richard Schmidt
(pp. 39–57). Honolulu, HI: University of Hawai’i National Foreign Language
Resource Center.

312
 Following Up
1. Conduct a review of L2 research where the terms explicit learning and
implicit learning are used. Try to find instances in which the researcher
actually defines the terms and also instances in which the researcher assumes
the reader knows what is meant by the terms. What does your research
demonstrate?

2. In B. VanPatten and J. Williams second and third editions of Theories in

second language acquisition, various theoretical approaches are presented by
experts in those approaches. One of their tasks is to review the
“explicit/implicit” debate. First, review what VanPatten and Williams say
about the debate in their first chapter and then review what the individual
contributors say about this topic. What do you find?

3. Review the section of the chapter called “Putting Explicit Learning into
Perspective.” Can you summarize in your own words what the principal
argument is in this section?

4. In language-teaching circles, people heatedly debate whether teaching and

practicing grammar is necessary or beneficial. After reading Chapter 6 and
the present chapter, why do you think this debate exists among teachers?

5. In both Chapter 6 and in this chapter, we have stated that explicit and
implicit knowledge are qualitatively different. What does it mean for
something to be qualitatively different from something else? Can you explain
what this means for explicit and implicit knowledge of language?

6. Read one of the following from the References and Further Readings
section. Prepare a 1,000 word (maximum) summary of the main ideas.

Hulstijn, “Theoretical and empirical issues in the study of implicit and

explicit second-language learning”

313
VanPatten & Rothman, “What does current generative theory have to
say about the explicit-implicit debate?”

Williams, “Attention, awareness, and noticing in language processing

and learning.”

314
 8
What Are Individual Differences and
How Do They Affect L2 Acquisition?
◈

Individual differences are a fact of life. One person differs from the next on a
range of factors, such as intelligence, height, shoe size, athleticism, and memory.
Even people with the same eye color don’t have exactly the same eye color. The
question of whether such differences come from innate characteristics (nature) or
from their environment (nurture) is an old one in philosophy. At least in the case
of L1 acquisition, the answer seems to be “both.” Children are biologically
predisposed to acquire a language (nature) and they acquire the language they are
exposed to along with the social norms that govern its use (nurture). Child
language acquisition is considered part of normal human development because all
children acquire the language of the speech community in which they are raised.
One of the obvious ways in which L2 acquisition differs from L1 acquisition is
that adults seem to vary a good deal more in rate and ultimate attainment (how far
they get) in L2 acquisition than children acquiring their L1 do. That said, as we
will see shortly, there are individual differences in L1 acquisition as well. For
some time, L2 researchers have looked to individual differences as a way to
explain the apparently wider range of outcomes for L2 learners. In this chapter,
we examine whether individual differences play a role in L2 acquisition, which
factors play a role, and how individual differences interact with the role of input.
First, though, we review the work on individual differences in L1 acquisition.

315
We’ll do this so that we have some context for individual differences in L2
acquisition. In this latter section, we will review the three main factors researchers
have looked at: aptitude, working memory, and motivation. We’ll then end the
chapter with a discussion of what aspects of L2 acquisition individual differences
explain.

316
 Individual Differences in L1 Acquisition
One common assumption about child language acquisition is that it is rapid and
relatively uniform. This assumption is based on two observations. The first is that
children seem to “pick up” their native language without much effort. The second
is that all normally developing children acquire language in a predictable order.
While the second observation is accurate – children do pass through predictable
stages in the acquisition of the L1 – it’s less true that language acquisition is quick
and easy. The fact that, in the absence of major cognitive deficits, human children
acquire their L1 means that it is easy to assume that child language acquisition is
effortless. This assumption masks individual variation in both child language
development and L1 ultimate attainment as well as the thousands of hours it takes
children to acquire the basics of language. One of the major ways in which
individual children differ is rate of acquisition. For example, most children begin
to understand individual words between 8 and 10 months of age. In other words, it
takes some children 25 percent more time to understand individual words.
Another major individual difference in early child language acquisition is
vocabulary size. For all children, vocabulary size increases as they get older.
However, at any given age, there is a wide range in the number of words that
children know. For example, research suggests that at 10 months of age, children
understand an average number of 67 words, with a median of 41 words. In other
words, half of the children have a vocabulary size of less than 41 words, and the
other half have a vocabulary size of more than 41 words. In order to bring the
average number of words to 67, many of the children who know more than 67
words know many more than 67 words. This is quite a bit of individual variation,
and, indeed, the range of the number of common words that those children
understood went from zero to 144. Vocabulary size matters because it predicts
both language comprehension and syntactic development, which means that

317
differences in vocabulary size predict other differences that persist throughout the
course of child language acquisition.

There is variation in both child language development and L1 ultimate

attainment.

One way that differences in vocabulary size persist is in the relationship

between comprehension and production. Production normally follows
comprehension. So children who are slower to comprehend language are also
slower to produce language. There is also a relationship between vocabulary size
and grammatical development. Children start putting two words together when
they have a vocabulary size between 50 and 100 words. Children who are slower
to acquire vocabulary are thus also slower to enter the two-word stage of
grammatical development. Utterance length and complexity start increasing
exponentially when children have a vocabulary somewhere between 300 and 600
words. Again, when this happens depends on the size of the child’s vocabulary.
In addition to the above, children have been observed to fall into two broad
groups regarding how they use language to communicate. One group is called the
referential group. This type of child begins the one-word stage of acquisition by
naming things largely using nouns (e.g., horsie, mommy). The other group of
children is called holistic. This group tends to use more chunks of language to
express emotions, treating the chunks like one-word items. A classic example is
silli-in-it, which is an unanalyzed reproduction of “silly, isn’t it?” Verbs tend to
predominate in the speech of these children. Even with these differences,
however, we still see the predictable ordered acquisition referred to in Chapters 1
and 2. In short, although all typically developing children pass through
predictable developmental stages, there are individual differences among children
in terms of how fast they pass through these stages.
Although it seems like all native speakers of a language have the same
mental representation for that language, adult native speakers can and do diverge

318
from each other. One of the most obvious ways is in vocabulary knowledge. In
addition, there is evidence that there are differences between syntactic and
semantic knowledge in adult native English speakers. For example, some adults
find it difficult to interpret passive sentences correctly, especially when the
syntactic and semantic cues conflict. These speakers find sentences like the dog
was bitten by the man (which means the man bit the dog) more confusing than
sentences like the man was bitten by the dog (which means the dog bit the man).
In the former case, the word order (dog-bite-man) lines up with the real-world
knowledge that dogs are far more likely to bite men than men are to bite dogs,
even though the passive voice gives the alternate interpretation. When passive
sentences have a meaning that conflicts with real-world knowledge, some
speakers misinterpret them. In addition, some adult native English speakers
interpret the sentences every basket has a dog in it and every dog is in a basket as
meaning the same thing while others don’t. Let’s picture, for instance, a dog
kennel that has three dogs and three baskets. Both sentences are true if each
basket has a dog in it. However, what happens when there are three baskets but
four dogs? There could be two baskets each with one dog and the third with two
dogs. This would make both sentences still true. Then there could be three baskets
each with one dog and the fourth dog is lying on the ground. Now only one of the
sentences is true. In other words, the first sentence is a statement about whether
all baskets are filled, and the second sentence is a statement about where the dogs
are. In short, adults can differ from each other in the extent to which certain
syntactic and semantic structures are acquired. These differences are usually not
evident to other native speakers of a language. Thus, it is easy to assume that L1
acquisition is easy, unproblematic, and uniform. However, research from
linguistics suggests that this assumption isn’t accurate. There are individual
differences in L1 acquisition, both in the course of development and in outcomes.
Researchers have suggested that these differences are due to an interaction
between learner-internal factors (e.g., communication styles) and learner-external
factors (e.g., exposure to the target language). However, ultimate attainment

319
seems to depend more on exposure to the language than on learner-internal
factors.

Ultimate attainment in L1 acquisition seems to depend more on exposure

than on learner-internal factors.

The factor that seems to predict L1 syntactic knowledge the best is

education. Ewa Dabrowska has conducted a series of studies investigating
whether a college education makes a difference in how native English speakers
perform on tests of grammatical knowledge. These structures include passive
voice and sentences like those with every mentioned above. Dabrowska has found
that participants with a college education get all of the questions right on these
tests but that there is much more variation among the participants with only a high
school education. Some of these participants perform like the college students and
get all of the questions right, but others perform much worse. Dabrowska argues
that education is the causal factor in these differences in syntactic knowledge and
performance. It’s likely that what education does is give people access to a wider
range of input. Once people get to college, they are exposed to a wider range of
texts and probably read more than people who do not go to college – or they have
read much more prior to going to college. Both the passive voice and the
sentences with every basket discussed above (the target structures in Dabrowska’s
work) are much more frequent in written language than they are in spoken
language. Thus, people who read frequently and widely are more likely to be
exposed to the relevant input. Thus, it’s possible that the differences among native
speakers in Dabrowska’s work have as much to do with quantity and quality of
input as with education, and that education is the vehicle for relevant input. Either
way, it’s clear that (1) there are individual differences in L1 acquisition that
extend beyond vocabulary size, and (2) the usual account of these differences is
that they are due to a combination of exposure, education, and literacy. Individual
differences are also influenced by the interaction between learner-internal factors

320
and the environment. For instance, children growing up in the same household
show variation in rate of acquisition. These differences can’t be attributed solely
to the environment because children who grow up in the same household share
their linguistic environment. In short, individual differences in child language
acquisition are probably due to a mix of learner-internal factors and
environmental factors, such as education and access to input.

321
 Individual Differences in L2 Acquisition
Superficially, L1 and L2 acquisition seem to have major differences in outcomes.
While L1 learners will acquire their L1, L2 learners are not guaranteed to acquire
the L2 and fall within a wide range of non-nativelikeness. Such differences need
an explanation, and research on individual differences in SLA has attempted to
explain why some people progress farther in L2 acquisition than others do. What
is more, individual differences should be able to explain outcomes independently
from the central roles of input, interaction, and general exposure. In this chapter,
we will focus on three main subsets of individual differences in L2 research:
language aptitude, working memory, and motivation. All three of these are
learner-internal factors, but aptitude and working memory can be thought of as
cognitive factors, while motivation is closer to a personality factor that is
sometimes also related to social factors. Although working memory is a factor in
L1 acquisition, neither language aptitude nor motivation is thought to influence
L1 acquisition. Thus, a major role for either aptitude or motivation in adult L2
acquisition would point to a substantial difference between L1 and L2 acquisition.
Before we discuss individual difference factors in more detail, it’s important to
reiterate that when we talk about L2 acquisition we are talking about the
underlying processes that create the abstract, complex, and implicit system we
have consistently called “mental representation” in this book. Let’s begin with
those individual differences that are internal to the learner.

322
 Language Aptitude
Language aptitude is defined as an individual’s ability to acquire an L2. Like
most cognitive attributes, it is hypothesized to be innate and relatively fixed. In
the United States, concern for aptitude for L2 learning can be traced to World
War II and the army’s desire to identify military personnel who would make good
L2 learners. The US military was – and still is – interested in identifying
individuals who have a “special” ability to learn an L2 on the premise that
investing in language instruction for these persons is a more efficient use of
resources than investing in language instruction for people with a low or average
ability to acquire an L2. Thus, from the beginning, work on L2 aptitude assumed
that some adults were more suited to L2 learning than others, and researchers set
out to identify which factors or abilities predicted success in language learning.
The foundation of aptitude research dates to the work of John Carroll and
Stanley Sapon in the 1950s. They set out to identify the abilities that make an
individual a good language learner and then developed the first aptitude test,
which is called the Modern Language Aptitude Test (MLAT). Carroll and Sapon
hypothesized that linguistic aptitude is a relatively fixed set of abilities. In other
words, linguistic aptitude varies among the population in the same way that eye
color and height do. Carroll and Sapon identified four abilities that they
hypothesized underlay a general language-learning ability:

An underlying premise of linguistic aptitude was that it varied among the

population in the same way that eye color and height do.

Phonetic coding ability refers to an individual’s ability to identify

individual sounds, particularly those that are not contrastive in English,
and to form associations between sounds and symbols.

323
 Grammatical sensitivity refers to an individual’s ability to identify the
syntactic function of words or phrases in sentences.

Inductive learning ability has to do with how well an individual can

generalize to a rule based on a set of language materials.

Rote learning ability has to do with how well an individual can learn
arbitrary associations between sounds and meanings.

Carroll and Sapon then developed a test – the MLAT – that tested each of
these subcomponents and refined the test based on how classroom language
learners scored on the test. In other words, they assumed that learners who
perform well on classroom assessments have high language aptitude and refined
the test so that it identifies those learners. Not surprisingly, scores on the MLAT
correlate strongly with classroom language learners’ performance, and thus the
MLAT has good predictive validity in instructional settings. Although other
aptitude tests, such as the Pimsleur language battery, the Hi-LAB, the LLAMA
aptitude test and the Canal-F Test, have been developed in the years since the
MLAT was developed, the MLAT remains the foundational aptitude test. Some of
the more recent tests, such as the LLAMA aptitude test, are closely modeled on
the MLAT, and others test similar abilities. To this day, the MLAT and its close
cousin, the LLAMA, remain the most frequently used aptitude tests in L2
research.

Consider this …
John Carroll and Stanley Sapon initially developed their definition of
linguistic aptitude in the 1950s. At this time, linguistics was just
beginning to undergo the evolution from structural to generative
linguistics described in Chapter 1. To a large extent, linguistics and
language teaching were still governed by the assumptions of structural
linguistics and behaviorism. How might these assumptions have informed
Carroll and Sapon’s definition of aptitude? Do you think if they saw

324
 language differently (i.e., as a complex, abstract, and implicit
representation not resembling textbook rules or patterns) they might have
gone about their research a different way?

Over the years, researchers have investigated whether aptitude scores

correlate with outcomes in L2 acquisition. If aptitude plays a role in L2 ultimate
attainment, we would expect to see a strong correlation between performance on a
measure of linguistic knowledge and scores on the aptitude test. Robert DeKeyser
conducted a study with Hungarian learners of English living in the United States
to see whether such a relationship exists. DeKeyser administered a grammaticality
judgment test and a Hungarian version of the grammatical sensitivity subsection
of the MLAT to L1 Hungarian speakers who were L2 learners of English and had
lived in the United States for at least ten years. DeKeyser found that there was a
strong correlation between aptitude scores and scores on the test for participants
who had arrived in the United States after the age of 16, but not for those who had
arrived before then. Similarly, Niclas Abrahamsson and Kenneth Hyltenstam
investigated the relationship between aptitude scores, age of arrival, and ultimate
attainment in L1 Spanish speakers who were L2 Swedish speakers. They found
that all of their near-native participants who had an age of arrival in Sweden after
the age of 12 had aptitude scores that were higher than the average aptitude score
for the participants who had arrived before the age of 12. The results of these two
studies aren’t that unusual in the literature – successful L2 learners have been
found to have above-average aptitude scores in a variety of studies. Based on
results like these, researchers such as Peter Skehan have argued that aptitude
should be seen as central to explaining individual variation in L2 acquisition.
Recently, however, some researchers have challenged the centrality of
aptitude as a major explanatory variable in L2 acquisition. These challenges have
been both conceptual and empirical. The first conceptual challenge is that the
construct of aptitude, and therefore the design of the aptitude test, rests on
assumptions about language that may or may not be warranted. Bill VanPatten

325
and some of his colleagues have argued that the MLAT assumes that language
acquisition consists mostly of explicit rule learning. This is also how most L2
classrooms treat language – students memorize vocabulary and learn rules – so
it’s not surprising that outcomes on the MLAT should correlate closely with
outcomes in formal language classrooms. It’s also not surprising that scores on
the MLAT should correlate with a lot of the outcome measures used in ultimate
attainment research. For instance, as mentioned above, Robert DeKeyser found a
strong correlation between MLAT scores and scores on a grammaticality
judgment test. DeKeyser’s grammaticality judgment test contained a number of
items that reflected explicit rule learning (as opposed to intuitions when such tests
are used with native speakers), so it is, again, not surprising that two tests of rule
knowledge should correlate with each other. VanPatten and his colleagues have
argued that language acquisition should be conceptualized as fundamentally
involving input processing, and that any regularities in learners’ grammars are a
by-product of processing the input. When acquisition is thought of as being
primarily about input processing and comprehending linguistic input, and not
about learning rules, it makes less sense that learners’ analytic ability should
predict outcomes in language acquisition. In other words, the question changes.
It’s not about whether some learners are better at learning rules than other
learners; it’s about how learners process input in real time.

A premise underlying aptitude is that it involves explicit rule learning.

When acquisition is thought of as being primarily about processing and
comprehending linguistic input, and not about learning rules, it makes less
sense that learners’ analytic ability should predict outcomes in language
acquisition.

VanPatten and his colleagues have published two studies that investigated
the relationship between learners’ processing behaviors and aptitude. The first of
these investigated whether aptitude predicted the speed at which L1 English

326
speakers shifted to a new processing strategy in Spanish, French, German, or
Russian, and is summarized in the Exemplary Study. The second one, conducted
by Bill VanPatten and Megan Smith, investigated whether aptitude scores were
related to naïve learners’ (absolute beginners) processing and acquisition of
complement + head word order in Japanese L2 when their L1 was English (see
Chapter 4 for a discussion of heads and complements). In both of these studies,
individuals differed from each other in terms of how quickly they moved toward
L2-like processing or how much they were able to project beyond the input. In
both of these studies, however, aptitude scores did not predict or correlate with
which learners did what or how fast they came to do it. It’s not that these studies
show no evidence of individual differences. Instead, they suggest that aptitude is
not the source of individual differences in the observed results when the focus is
taken off rule learning.

Exemplary Study

VanPatten, B., Collopy, E., Borst, S., Price, J., & Qualin, A. (2013).
Explicit information, grammatical sensitivity, and the first-noun principle:
A cross-linguistic study in processing instruction. The Modern Language
Journal, 97, 506–527.

Background
Language learners tend to interpret the first noun in a sentence as the
subject or agent.

This presents a problem with interpreting non-canonical word orders,

such as object-verb-subject structures in Spanish or scrambled
sentences in Russian.

Processing instruction (PI) is an intervention designed to push learners

away from relying on the first-noun principle (FNP).

327
In PI, input is manipulated so that both the subject and the object are
capable of performing the action, and learners hear a mix of SVO,
OVS, and OV sentences.

PI includes information about the target structure and feedback so that

learners know when they misinterpret sentences.

The words-in-sentences section of the MLAT, which tests for

grammatical sensitivity, has been found to have strong correlations with
scores on measures of grammatical knowledge in previous research.

Research Questions
Does explicit information (EI) influence how quickly learners of
Spanish, German, French, or Russian move away from the FNP and
correctly interpret sentences in which the first noun is not the agent?

Does grammatical sensitivity correlate with outcomes on the PI task,

regardless of the presence of explicit information?

Participants
42 third-semester Spanish learners, split between a group that got
explicit information (+EI) and a group that did not get explicit
information (−EI)

46 third-semester German learners, split between a +EI and a −EI group

44 third-semester Russian learners, split between a +EI and a −EI group

48 third-semester French learners, split between a +EI and a −EI group

Materials
PI materials with and without explicit information

328
 PI materials had pictures and audio files that provided input on OVS
structures in Spanish, German, and Russian, and the causative faire
construction in French

the words-in-sentences section of the MLAT

Results
Spanish: no effect for explicit information or grammatical sensitivity

German: an effect for explicit information; grammatical sensitivity

correlated with how fast the +EI group progressed but not with the –EI
group’s progress

Russian: no effect for explicit information or grammatical sensitivity

French: an effect for explicit information, but not for grammatical

sensitivity

Conclusion
The results run counter to findings of previous research. The authors
conclude that aptitude as grammatical sensitivity does not relate to how
learners process input, which is assumed to be an integral part of how
acquisition happens.

Other challenges have come from researchers who question the

conceptualization of aptitude. Aptitude, like most other cognitive variables, was
originally hypothesized to be relatively fixed. This means that, regardless of the
context or the type of learning a situation called for, an individual’s aptitude
would be the same. This means that aptitude should be independent of other
cognitive variables, such as intelligence. It also means that linguistic aptitude and
education should not be related to each other. Researchers have begun to
challenge these assumptions. The biggest initial challenges have come from

329
outside of L2 research. Individual differences are an area of research in cognitive
psychology more generally and many psychologists have argued that individual
differences are more related to specific situations and are less fixed than earlier
generations of researchers had thought. Based on this work, some researchers
have argued that L2 researchers need to pay more attention to how individual
learners are situated in their own contexts, as well as to how they respond to those
contexts. In other words, some learners might have a better analytic ability than
others, but that only matters if learners are in a context that calls for analytic
abilities. In addition, some evidence suggests that linguistic aptitude is modulated
by other factors, such as education. The challenges to the basic construct of
aptitude don’t necessarily challenge traditional conceptions of language or of
aptitude as such; instead, they raise the possibility that what looks like an effect
for aptitude actually has other, more diverse, sources. In line with the above, Peter
Robinson has offered a more complex and nuanced view of aptitude in a number
of publications. However, his work is focused almost exclusively on the
relationship between instruction and aptitude (or aptitude complexes) and thus
does not extend beyond what we might consider largely explicit learning
(Chapters 6 and 7). In other words, following in the more general work on
aptitude, the concern is for how instructed learners perform in focused
environments where much of their attention is directed (explicitly or implicitly) to
how language works as well as working with language itself.
In short, much of the research on the role of language aptitude in L2
acquisition has found that learners with higher scores on aptitude tests tend to do
better on tests of L2 knowledge. This is especially true when the tests of L2
knowledge tap learners’ knowledge of explicit rules. On the other hand, other
research has found that when the measures of linguistic knowledge reflect
learners’ input processing, comprehension, or abstract implicit knowledge, the
relationship between aptitude scores and task performance disappears. Like most
things, results tend to reflect researchers’ working assumptions about the nature of
language, the nature of language acquisition, and the tasks they use in their

330
research. Because of this, the role of linguistic aptitude in L2 acquisition more
generally is not clear. What is clear, however, is that it does not play the central
predictive role that Carroll and Sapon initially hypothesized.

Consider this …
Concern for linguistic aptitude originated in the military and is still a
concern today. However, in educational contexts and in language
acquisition outside of the military, aptitude has largely been of interest
only to researchers. That is, the “practical” applications of aptitude
research have not been of concern as they have been in the military. Why
do you think this is so? Do you think, for example, that in schools and
universities there should be tests of aptitude to put students into different
groups or to promote language learning among some students and not
others?

331
 Working Memory
Working memory is another individual difference that has received a significant
amount of attention in L2 acquisition literature. Working memory (WM) is a
construct that was originally developed in cognitive psychology. Working
memory is the part of human cognition that allows us to temporarily store and
manipulate information during comprehension. For example, effective note-taking
involves working memory. To take notes, you have to listen to the speaker, pay
attention to what the speaker is saying, and remember enough of it to write it
down. Because the lecturer often speaks faster than you can write, you have to
have enough processing resources to take in new information while recording
older information. The WM storage component is what allows you to keep things
active in your mind while focusing on something else, and processing capacity is
what allows you to pay attention to the other task. Working memory is thought to
be related to individuals’ ability to process information, and thus may influence
how efficiently learners process an unfamiliar language.
Working memory is a complex phenomenon consisting of various
components that we won’t describe in detail here. However, tests of WM are
important to understand and will describe the most used test: the reading span test.
In the reading span test, a person is asked to read a series of unconnected
sentences aloud. The unconnected sentences are presented in blocks of increasing
size: 3 sentences, then 4 sentences, then 5 sentences, then 6 sentences. To keep
people from simply memorizing the end of a sentence, it is typical to include a
reaction to each sentence’s meaning (e.g., whether it makes sense or not). At the
end of each block, a person is cued to recall the last word of each sentence. At
some point, the person fails to recall all of the items in a block and is assigned a
score to indicate where he or she fell on the reading span test. There are some
variations on the reading span test but all involve the basic components of blocks
of sentences of increasing numbers of sentences per block plus a cued recall at the
end of the block. The idea is that this kind of recall test taps the processing

332
capacity that learners have during comprehension and that this processing
capacity is reflective of working memory. In L1 research, such tests have shown
to correlate highly with reading comprehension.
Reading span tests are not used in child L1 acquisition for obvious reasons,
and so no research has been done to correlate reading span scores in, say, 2- and
3-year olds and their syntactic or grammatical development or their
comprehension of language. Instead, child L1 researchers have looked at one
component of working memory called phonological short-term memory.
Phonological short-term memory governs how much verbal information someone
can remember at any given time. In children learning their L1, phonological short-
term memory is measured by their ability to repeat non-words. Children with a
high phonological short-term memory capacity learn real vocabulary faster, have
a larger vocabulary, and thus usually acquire syntax faster than children with a
lower phonological short-term memory. Overall, children who have a high
working memory capacity also tend to have strong reading skills and are better
able to comprehend complex sentences than children with a low working memory
capacity. This suggests that working memory is related in some way to linguistic
ability in children; the more you can comprehend, the more language you get.
In the L2 literature, researchers have hypothesized that WM is related to the
acquisition of L2 vocabulary and syntax. Researchers have also hypothesized that
WM is related to L2 fluency and reading comprehension skills. With respect to L2
vocabulary, phonological short-term memory has been found to predict learners’
ability to acquire new vocabulary in an L2. In addition, learners with a high
phonological short-term memory capacity tend to do better on tests of
morphosyntactic knowledge. The results of this research suggest that
phonological short-term memory contributes to L2 learning. This may be because
language is primarily a verbal system, and it is processed in phonological short-
term memory first. Better phonological short-term memory capacity then leads to
more linguistic processing (i.e., comprehension), which leads to faster and/or
better L2 development.

333
 In addition to work that looks at the role of phonological short-term memory
in L2 acquisition, researchers have also investigated whether working memory
capacity as a whole predicts syntactic knowledge. For example, Alan Juffs
investigated whether working memory capacity as measured by a reading span
task influenced how non-native English speakers processed complex sentences.
Juffs was interested in whether reading span scores were related to how L2
English speakers from different L1 backgrounds processed different kinds of
sentences with wh-gaps. You might remember from Chapter 5 that wh-gaps in
English involve moving a word from its original position in the sentence to a
position earlier in the sentence. Wh-gaps occur in wh-questions and in relative
clauses (e.g., Which teacher did John say was going to be late? is related to John
said which teacher was going to be late?). In addition to a reading span task, Juffs
gave L1 Chinese, L1 Japanese, and L1 Spanish speakers who spoke English as an
L2 a reading task that asked them to indicate whether sentences containing wh-
gaps, like those in (1), were grammatical.

(1)

a. *Whoi did Ann meet the teacher after she saw ei on Tuesday?

b. Whoi does the nurse know ei saw the patient last week?

A computer recorded participants’ reading times. Slower reading times on

the word immediately following the gap would indicate that participants are
reintegrating the wh-word into the gap position. Overall, Juffs found that L1
background was a factor in the results, with L1 Spanish speakers performing the
closest to native English speakers and the L1 Japanese group demonstrating the
most difficulty processing these kinds of sentences. However, working memory
capacity did not seem to influence performance for any of the L2 learners.
Like the results of aptitude research, the results of research on the effects of
working memory in L2 acquisition are mixed. There is some evidence, for
instance, that phonological short-term memory capacity predicts vocabulary
learning. One limitation of this research is that most of it has been done in explicit

334
or semi-explicit learning conditions. For example, one common design for
vocabulary research is to present participants with L1/L2 vocabulary pairs, ask
them to memorize the vocabulary, and then give them a vocabulary post-test. This
kind of task asks participants to memorize words and it gives them a test of
memory capacity. It could be that the results of these two tasks are correlated
because they are both memory tasks. Memorized vocabulary isn’t necessarily
vocabulary that has become intake and then incorporated into the linguistic
system. Vocabulary is like anything else in language; it is ultimately acquired
through learners’ interaction with aural and written input in communicative
contexts. With respect to working memory capacity more generally, there is some
evidence that working memory is related to L1 sentence processing. However, the
effects of working memory in L2 sentence processing do not seem to be
independent of the influence of the L1.

Like the results of aptitude research, the results of research on the effects
of working memory in L2 acquisition are mixed.

Consider this …
At the conceptual level, the possible role of working memory makes sense
in language acquisition. If working memory is related to comprehension,
and comprehension (processing input) is a necessary part of acquisition,
then working memory should be related to acquisition. Can you think of
any flaws in this argument? Is there a piece of acquisition that research in
working memory might not be considering?

335
 Motivation
Researchers have been discussing the possible role of motivation in L2
acquisition since the early days of L2 acquisition research. Unlike aptitude and
working memory, both of which are cognitive factors, motivation is an affective
factor. Although motivation is usually expressed in thoughts and opinions (e.g.,
“Spanish class is so boring!”), these reflect an individual’s emotions, experiences
in the world, and short- and long-term goals. In addition, although aptitude and
working memory are usually thought of as relatively fixed factors, motivation is
more variable. It changes as a function of a learner’s feelings about something and
their experience of the world. Thus, motivation can be affected by external social
factors. Motivation researchers recognize the fundamental ebb and flow of
motivation and define it as involving a process by which learners direct their
attention toward goals. So motivation has to do with the process that helps
learners reach their goals for acquiring an L2.
Second language motivation research has its roots in Robert Gardner and
Wallace Lambert’s work in the late 1960s. These researchers were interested in
the dynamics of living in a bilingual and bicultural society in Canada. Gardner
and Lambert hypothesized that learners’ social contexts and goals for language
learning would influence individual outcomes. They distinguished between
integrative and instrumental motivation. The term integrative motivation refers
to an individual’s perspective on the target language and culture. In particular,
people with a strong integrative motivation have a positive orientation to the
target culture. This is usually accompanied by relatively higher degrees of
openness to new experiences. It can also be reflected in positive impressions of
speakers of the target language and a desire or willingness to communicate with
speakers of the target language. The other main type of motivation is instrumental
motivation. The term instrumental motivation refers to more of a practical need
the learner has for the language, usually in order to do a job. Instrumental
motivation may not be limited to a job – it can also include things like passing an

336
entrance exam or other high-stakes test or completing some education in the target
language. Both instrumental and integrative motivation are primarily about social
contexts: affinity for culture, relationship building and maintenance, and success
at work and school. Over the years, this distinction became less pronounced.
Gardner, for instance, argued that, at some level, instrumental motivations are
also integrative because they indicate some willingness to interact with other
communities. Instrumental and integrative motivation are important because they
influence what language(s) learners study and whether they seek out opportunities
to interact in the target language and develop their skills. Learners without some
basic level of motivation won’t engage the learning process.

Consider this …
Gardner and Lambert were working in Canada where there were two
relatively homogeneous communities – French-speaking and English-
speaking. They were also working in a context in which bilingualism was
a legal requirement, if not a social norm. At the same time, due to
historical and social factors, the two communities were largely
independent of each other in that people went to different churches and
sent their children to different schools. In this context, the idea of
integrative motivation makes a lot of sense: There are political and social
factors that make the idea of “crossing cultures” seem possible or
desirable. Can you think of cases in which learning an L2 might be
important but in which integrative motivation would not be a major
factor? What about cases in which learning an L2 might depend primarily
on instrumental motivation? How might different types of motivation
influence how much of the L2 learners acquire?

By the early 1990s, research on motivation had shifted from focusing

primarily on the social context for language learning to focusing on classroom

337
contexts. There was also an effort to bring the constructs of motivation more in
line with the research in psychology on motivation. This body of research focused
on the interaction between social context, learner-internal factors (such as self-
efficacy), and classroom and course contexts (e.g., what the classroom incentives
are). One of the insights from this period of research is that motivation ebbs and
flows over time. On any given day, and in any given class period, a learner’s
motivation may be low or it may be high. One question is how to measure the
relationship between motivation at any given point in time and learning of one
particular aspect of the language. At the same time, however, because motivation
ebbs and flows, and because language acquisition is a long-term process, what
matters most is not the moment-by-moment motivation but the long-term
commitment to language acquisition goals.
As more researchers with different backgrounds became interested in
motivation, the focus shifted again. This shift is the result of a number of factors.
One of these is that the world changed significantly between the 1960s, when
Gardner and Lambert started their research, and the early decades of the twenty-
first century. As the pace of globalization increased and English became a global
lingua franca, the construct of integrative motivation made less sense. For
instance, English is the language of scientific publication, and it’s the language of
many science departments in universities across the world. This means that
Japanese scientists might never leave Japan, and they might speak Japanese at
home and in some aspects of their professional lives, but they would also need
fairly technical English skills to conduct other aspects of their professional
careers. It’s not clear what the integrative motivation would be here – the
Japanese scientist doesn’t need to have a particular affinity for England or the
United States, for example, to need English skills. Consequently, Zoltan Dörnyei
and other L2 motivation researchers have argued that the construct of integrative
motivation should be expanded and reconceptualized so that it better fits with
constructs of self-conception in psychology research. They have done so by
introducing the idea of the learner’s self-concept.

338
 Self-concept has to do with how learners see themselves, including who they
are today (both positive and negative attributes), and who they might become.
Dörnyei calls this the L2 motivational self-system, and it includes three aspects:
the Ideal L2 Self, the Ought-to L2 Self, and the learning environment. The Ideal
L2 Self draws on ideas from psychology about how people grow and change. The
basic idea is that who individuals imagine themselves to be in the future is
important for directing their attention to those goals. So the Ideal L2 Self is
essentially who learners consider themselves to be with respect to the L2. It’s a
way to incorporate L2 skills into learners’ conceptions of themselves. If they
don’t currently have the skills they want, the Ideal L2 Self helps orient learners to
different aspects of the language-learning task. The Ought-to L2 Self includes the
beliefs or expectations that learners have about who they should be or how they
should behave. Ideally, the Ideal L2 Self and the Ought-to L2 Self should work
together, but this may not always be the case. Both of these self-concepts are
influenced by learners’ experiences, learning environments, and social contexts.
Overall, motivation research today considers motivation on a number of different
levels: learners’ goals, learners’ self-concept, the social context for learning, and
the interaction between learner-internal factors and their educational and social
contexts. The key insight from L2 motivational research is that learners’ self-
concept is a key factor in a larger constellation of factors that influences how far
learners get in L2 acquisition.

L2 motivation research finds that learners’ self-concept is a key factor in a

larger constellation of factors that influence how far learners get in L2
acquisition.

339
 What Do Individual Differences Explain?
Many L2 researchers assume that individual differences, such as working
memory, aptitude, and motivation, play an important role in determining
outcomes for L2 learners. It seems to us that individual differences may help
explain the rate of acquisition and why some learners acquire more of an L2 than
other learners, but they do not really explain how language gets into the learner’s
head, and they don’t really explain what language gets into the learner’s head.
Throughout this book, we have focused on four central observations about L2
acquisition. These observations are: (1) L2 development is ordered and relatively
impervious to outside intervention; (2) input is central to language acquisition; (3)
instruction has a limited role to play in influencing the L2 acquisition process or
how far learners get; and (4) explicit knowledge and learning play little to no role
in L2 acquisition. None of the research on individual differences changes any of
these observations. For instance, stages of development and acquisition orders
show up in different populations of L2 learners and are not affected by a learner’s
L1 or by instruction. There is no evidence that aptitude or working memory
changes acquisition orders. Aptitude, it is true, has been found to predict learners’
outcomes in two areas: classroom performance and scores on certain kinds of
grammaticality judgment tasks. This sounds like aptitude might influence the
development of the internal linguistic system. However, one of the problems with
the research that supports the role of aptitude in L2 acquisition is that it tends to
rely on tests of explicit knowledge or material learned explicitly. And, as we
discussed in Chapter 7, explicit knowledge and learning play little to no role in
the development of the linguistic system. When aptitude is studied in conjunction
with tasks that require learners to process input, the relationship between task
performance and aptitude scores disappears. Similarly, there is no evidence that
these individual difference factors influence whether learners show poverty of the
stimulus effects in the L2. Remember that poverty of the stimulus effects

340
(discussed in Chapters 3 and 5) show up when people show that they know
something about the language that they could not have acquired based only on the
information in the input. They are also usually subtle enough that they are not
explicitly taught in language classrooms. Poverty of the stimulus effects show up
when learners project beyond the input to which they were exposed in order to
create a mental representation for the target language. Most studies that have
investigated poverty of the stimulus effects in L2 acquisition find that L2 learners
have the same judgments as native speakers. This suggests that poverty of the
stimulus effects are not subject to individual differences. So while individual
differences do not influence the type of system learners acquire, they do play a
secondary role.

341
 Individual Differences and Rate of Acquisition
Individual differences such as aptitude, working memory, and motivation are
usually invoked, alongside the apparent influence of the L1 on L2 acquisition, to
explain the apparent failure of L2 learners to converge on target-like L2
knowledge. We have seen that individual differences do not actually have much
explanatory power when it comes to explaining the process of language
acquisition and the nature of the system that learners acquire. That said, individual
differences do have some explanatory power in terms of outcomes. This is
because learners’ individual circumstances and choices influence how they
interact with the input and what input they have access to. In other words,
motivation matters for language learners not because it changes how they acquire
language but because it keeps them oriented to long-term goals and influences
whether they choose to seek out opportunities to learn and use the language
beyond the classroom. In short, motivation keeps learners engaged in the (largely
unconscious) language acquisition process. For example, one of the authors of
this book was motivated to learn Japanese. She sought out short- and longer-term
study-abroad opportunities, spending about a year in Japan as a high school
student. She also worked there for a while as an adult. She acquired more
Japanese than her classmates in her high school Japanese classes did. Without the
motivation to learn Japanese, she probably would not have sought out those
opportunities. That said, this motivation did not change the fact that she acquired
her Japanese through interaction with the target-language input. In other words,
the exposure to Japanese language input in communicative contexts was essential
for her acquisition of Japanese. Her motivation had nothing to do with the internal
mechanisms that work on language – but it was vitally important in that it helped
her create the opportunities to have better access to input and then take advantage
of them (and it kept her from giving up when the process was hard). This stands
in contrast to another author of this book whose approach to the acquisition of
French was the “good enough” approach. That author knows and speaks some

342
French but readily admits he has no desire to learn more and has no desire to be
part of the French-teaching profession. He does not identify with French speakers
and does not want to be like them and his interaction with the French language-
teaching profession has been negative in many ways, which in turn has impacted
his motivation to seek out input and interaction with speakers. He does not read in
French and seldom watches French movies any more. Interestingly, his
motivation to learn French at the beginning was high. But as he interacted more
and more with the French language-teaching profession, his attitude toward
learning French changed. In essence, he has stopped learning French and is far
short of any nativelike ability.

Motivation matters for language learners not because it changes how they
acquire language but because it keeps them oriented to long-term goals
and influences whether they choose to seek out opportunities to learn and
use the language beyond the classroom.

Similarly, working memory may influence how much of the input a learner
can process at any given time, and thus may influence the rate at which learners
acquire vocabulary. It’s likely that vocabulary acquisition also influences
syntactic development. This is because learners process input for meaning first, so
the more they understand from the input, the more resources they have available
for syntactic processing. Thus, learners with a higher working memory capacity
may have more cognitive resources available to allocate to language processing
and may acquire language somewhat faster than learners with less working
memory capacity. Still, though, working memory capacity doesn’t change the
basic processes that underlie language acquisition. Learners need access to input,
they have to process that input, and then their internal mechanisms have to
“organize” the data.
Given that individual differences influence rate of acquisition and access to
input but not the actual processes of acquisition, why have they received so much

343
attention in L2 acquisition research? The answer to this question involves a
couple of factors. First, two of the major individual difference factors –
motivation and aptitude – emerged in specific social and historical contexts. In the
case of motivation, this work came out of the Canadian context, where
bilingualism is a social and political goal, but where the language communities
themselves may have relatively little spontaneous interaction with each other.
Aptitude research emerged in an era when linguistics and language teaching was
dominated by behaviorism and structural linguistics, with its focus on rule
learning and habit formation. Aptitude thus received a good deal of attention
because the military was (and still is) looking for a way to identify good language
learners.
The second, and probably more important, factor is that many SLA
researchers assume that children and adults have access to the same kinds of input
and spend the same amount of time on task. When we compare a child L1 learner
after four years of input and interaction to an adult who has had four years of
language classes, we find that the child has acquired much more of the target
language. Because we assume that the acquisition contexts are the same, this
seems to suggest that target-like L2 acquisition is the exception, not the rule, for
L2 learners. Individual differences are thus proposed as an explanation for this
fact. But let’s examine this observation about time a little more closely. We know
that children have acquired most of the target language by about the age of 5. By
this point, children have been engaged in meaningful interaction with the target-
language input at least eight hours a day for five years. This means that they have
spent at least 14,600 hours acquiring their native language. Even then, children
don’t have adult-like knowledge of passive voice or of the semantic factors that
are encoded in words like each, every, all, and none. In addition, children can’t
read and their vocabulary is limited to their immediate world. In contrast, an adult
who takes four years of college-level Spanish has fewer hours of exposure to
Spanish. Assuming that adult takes one Spanish class every semester for four
years, that that class meets five days a week for an hour at a time, that the input is

344
good, and that classes run for 15 weeks at a time, the adult has had about 600
hours interacting with target-language input. If we add some out-of-class
exposure (not fill-in-the-blank-type homework), we could be generous and add
another 100 hours. If the input isn’t good, or if the class doesn’t promote
meaningful interaction with the input, the adult has even fewer hours of exposure
to the target language. In other words, after four years of college-level Spanish,
the adult has spent the same amount of time interacting with the target language
as a child who is about 10 weeks old. Unlike the 10-week-old, however, the adult
is expected to be a fluent speaker and reader of the target language. We wouldn’t
expect that of a 10-week-old, so it’s likely not a fair expectation of the adult,
either. In fact, given that input and time on task are necessary conditions for both
L1 and L2 acquisition, it is not surprising that outcomes for L2 learners are more
variable than they are for L1 learners. In fact, in many cases, it’s surprising that
L2 learners acquire as much as they do. And it is only the long-term learners, the
ones who accumulate thousands of hours of good input and interaction with that
input who approach nativelike abilities in some of the domains of language and
language use. To be sure, it may not be just time on task that is the difference we
see between child L1 outcomes and adult L2 outcomes. And to be sure, we noted
that there are individual differences in L1 acquisition and outcomes. But the point
here is that whatever individual differences are and whatever role they play in L2
acquisition, they have to be situated in a more complex picture of input and
interaction over time – including time on task.

What about Social Factors?

Social factors, like community expectations of bilingualism, access to

education, and access to language-learning opportunities, all influence
outcomes for L2 learners. But they do so because they influence access to
input and they influence how much time learners can spend interacting
with the target language. For instance, in many parts of the United States,
being a monolingual English speaker is the norm. Some communities may

345
place a value on learning an L2, but it is not expected and opportunities
for interaction with native speakers of other languages are rare. Thus, L2
learners have to seek out other opportunities for input and interaction and
will be more highly motivated.
In contrast, in countries like Belgium where multilingualism is the
norm, the social world is different. Signs may be in multiple languages,
shopkeepers may speak multiple languages, and people have access to
better input and more opportunities for meaningful interaction with the
target-language input, and are therefore more likely to acquire the L2 with
a greater degree of proficiency. One of the authors recalls a visit to
Denmark where he discovered how important it was among teenagers to
speak English. There was peer pressure to learn English. As he heard from
one person, “If you can’t speak English by high school, you’re a nerd.”
When he probed, he found that kids in Denmark developed a lot of
strategies outside of the classroom to get access to English. Music (largely
from the US) seemed to be a number one source of input for this group.
That author makes no claims about the reality of this situation at the time
of writing this book, but this was his experience some 20 years ago when
he visited. The anecdote reminds us, though, that it is good for us to ask,
as we travel, about people’s attitudes toward language acquisition to
discern just what the social context is that fosters acquisition or doesn’t
foster it.

346
 Recap
Here is a summary of the major ideas covered in this chapter.

Individual differences are evident in L1 acquisition. They usually show up

in vocabulary knowledge and in some types of syntactic knowledge.
Researchers trace individual differences in L1 acquisition to education and
socio-economic status.

Individual differences, such as aptitude, working memory, and motivation,

have been proposed to account for the variation in outcomes for L2
learners.

Aptitude is hypothesized to be fixed and to vary among people in the same

way physical traits vary.

Working memory has to do with how much information people can keep
active and process at any given time.

Motivation has to do with learners’ long-term goals and orientation to the

target language. Early motivation researchers distinguished between
integrative and instrumental motivation, and researchers today see
motivation as more fluid. Either way, it is the factor that keeps learners
oriented to long-term goals.

Individual differences do not play a significant role in what gets into

learners’ heads or how it gets there.

Individual differences do influence how much access to input learners

have, how much time they spend interacting with the input, and the quality
of the input learners receive. Thus, they can influence rate of acquisition
and overall outcomes. But individual differences do not change the
fundamental linguistic and cognitive processes involved in either L1 or L2
acquisition.

347
 References and Readings

Abrahamsson, N. & Hyltenstam, K. (2008). The robustness of aptitude effects in

near-native second language acquisition. Studies in Second Language Acquisition,
30, 481–509.

Carroll, J. & Sapon, S. (1959). Modern language aptitude test. San Antonio, TX:
Psychological Corporation.

Dabrowska, E. (2012). Different speakers, different grammars: Individual

differences in native language attainment. Linguistic Approaches to Bilingualism,
2, 219–253.

Dabrowska, E. & Street, J. (2006). Individual differences in language attainment:

Comprehension of passive sentences by native and non-native English speakers.
Language Sciences, 28, 604–615.

DeKeyser, R. (2000). The robustness of critical period effects in L2 acquisition.

Studies in Second Language Acquisition, 22, 499–533.

Dörnyei, Z. (2009). The L2 motivational self system. In Z. Dörnyei & E. Ushioda

(Eds.), Motivation, language identity and the L2 self (pp. 9–42). Bristol:
Multilingual Matters.

Dörnyei, Z. & Ushioda, E. (2009). Motivation, language identities and the L2

Self: A theoretical overview. In Z. Dörnyei & E. Ushioda (Eds.), Motivation,
language identity and the L2 self (pp. 1–8). Bristol: Multilingual Matters.

Gardner, R. & Lambert, W. (1959). Motivational variables in second-language

acquisition. Canadian Journal of Psychology/Revue canadienne de psychologie,
13, 266–272.

Gardner, R. & Lambert, W. (1972). Attitudes and motivation in second-language

348
learning. Rowley, MA: Newbury House.

Juffs, A. (2005). The influence of L1 on the processing of wh-movement in

English as an L2. Second Language Research, 21, 121–151.

Robinson, P. (2007). Aptitudes, abilities, contexts and practice. In R. M.

DeKeyser (Ed.), Practice in a second language: Perspectives from applied
linguistics and cognitive psychology (pp. 256–286). Cambridge: Cambridge
University Press.

Skehan, P. (2012). Language aptitude. In S. M. Gass & A. Mackey (Eds.), The

Routledge handbook of second language acquisition (pp. 381–395). New York:
Routledge.

VanPatten, B., Collopy, E., Price, J., Borst, S., & Qualin, A. (2013). Explicit
information, grammatical sensitivity, and the First-noun Principle: A cross-
linguistic study in processing instruction. The Modern Language Journal, 97,
506–527.

VanPatten, B. & Smith, M. (2015). Aptitude as grammatical sensitivity and the

initial stages of learning Japanese as a L2: Parametric variation and case marking.
Studies in Second Language Acquisition, 37, 135–165.

349
 Following Up
1. One of the ideas that emerged from this chapter is that individual
differences in L2 acquisition may help explain ultimate attainment but not
affect underlying processes. See if you can explain in your own words what
this means.

2. In the chapter we briefly discussed individual differences in L1

acquisition. Do you think that perhaps individual differences in L1
acquisition carry over to L2 acquisition? That is, faster learners in L1
acquisition are faster learners in L2 acquisition? How might you research
this? Consider doing a literature search on this topic and prepare a short
summary.

3. Read the VanPatten and Smith article from Studies in Second Language
Acquisition listed in References and Readings. Prepare a summary similar to
how exemplary studies are done in this book. Then discuss any implications
for additional research and/or for teaching that you think emerge from this
kind of research.

4. Review the nature of reading span tests for working memory briefly
described in this chapter. Then do an internet search on these other WM
tests: digit span and word span. How do the tests differ? Do you think one is
more valid than the other? Note that the reading span test or some version of
it has predominated in L2 research. Why do you think that is so?

5. Conduct a survey among ten people who are not in the field of language
acquisition or language teaching but who have studied another language. Ask
them “how far” they got and then ask them to what extent “motivation” has
played a role in their outcome. What ideas emerge from what they tell you?
Do they mention any factors other than motivation? When they talk about
motivation, how do they talk about it?

350
 Epilogue
◈

As promised at the outset of this book, in this epilogue we return to the

fundamental question that launched the field of L2 research in the 1970s and lurks
in the background of contemporary research: Is L2 acquisition like L1
acquisition? By now, the astute reader may have guessed that the way this
question is phrased is deceptively simple. Language acquisition is a complex
phenomenon and to answer such a question we need to tease apart some aspects
of acquisition covered in this book.
We can begin by thinking about language acquisition as an internal process
and set of internal mechanisms as well as an external process or set of external
mechanisms. For the internal processes we can ask ourselves this: Do L1 and L2
share the same internal process and mechanisms for the development of
language? Our conclusion is that, to a large degree, they do. What is the evidence
for this? We list it here.

Both contexts (if not all contexts of language acquisition) require the same
kind of raw data: communicatively embedded input (CEI). That is, both
children learning an L1 and children, adolescents, and adults learning an
L2 need to engage in contexts in which they hear and see language in
communicative contexts. They must actively engage in comprehension
and interaction with that input. Researchers have recognized the
fundamental role of CEI in language acquisition since the 1960s with L1
acquisition and the 1970s with L2 acquisition. Nothing has surfaced in the
research to challenge the critical role of CEI as the data from which
internal mechanisms draw in order to create language in the mind/brain.

351
That the data for both L1 and L2 acquisition are the same suggests that
something internal to all learners is the same. That is, CEI is required
because the mechanisms in the mind/brain require it regardless of context.
This suggests a substantial similarity in the raw data for both L1 and L2
acquisition.

In both L1 and L2 acquisition, we see ordered development. Neither child

L1 learners nor child, adolescent, or adult L2 learners create idiosyncratic
languages in their heads. Allowing for individual differences in rate and
attainment, learners in both contexts follow predictable paths in the
acquisition of language. In many cases, the paths for both L1 and L2
acquisition overlap or are very similar. We see that in both L1 acquisition
and L2 acquisition, the same factors seem to impact development:
markedness, universals of language, Universal Grammar, and frequency in
the input, among others. Something in both contexts constrains linguistic
development in similar ways. This suggests something is the same about
L1 and L2 acquisition. However, there is the issue of L1 influence in the
L2 context, which we take up in the next bullet point.

L2 learners have something internal to them that L1 learners do not:

another language (or other languages). We know that L1 influence
happens in L2 acquisition. At the same time, we know that this L1
influence is constrained or quickly attenuated when universals of language
and acquisition come into play. This aspect of language transfer or
influence would seem to be a big internal difference between L1 and L2
acquisition. At the same time, the first two points above hold. The
presence or absence of another language in the mind/brain does not
obviate the role of input and does not compromise ordered development. It
might affect ordered development, but again, as we’ve seen, the effect is
constrained. So as we see it, the presence of an L1 inside the mind/brain of
the L2 learner does not alter the processes or block what the internal
mechanisms responsible for language development do or must do. The L1

352
can, however, “gum up” the works and slow down the process in some
areas or some cases. In short, the presence of an L1 in L2 acquisition does
not cause the underlying processes and mechanisms to be different for the
L2 learner; it simply makes them (in some cases) not as efficient.

Neither L1 nor L2 acquisition seems particularly responsive to outside

manipulation. Attempts at overtly correcting child L1 grammar results in
sometimes hilarious and unexpected outcomes, as the example from David
McNeill in Chapter 1 shows. Chapter 6 on the effects of instruction in L2
acquisition suggests that such effects are limited. Some L2 scholars would
claim the effects of external manipulation via explicit teaching and
practice are non-existent. Thus, in both contexts, acquisition proceeds
along its own path with the only external influence being the actual input
itself. This suggests something is the same about L1 and L2 acquisition.

In both L1 and L2 acquisition, output as part of communicative interaction

is not the causal factor in how language unfolds over time. In either
context, when learners engage in communication and express their own
meaning, they draw from a linguistic system already in their heads.
However, in both contexts, being part of communicative interactions gets
the learner more appropriate-level input. This limited or indirect role of
output suggests that something is the same about L1 and L2 acquisition.

From the above and other observations made in this book, we know that
both L1 and L2 acquisition involve largely implicit internal processes.
Although it appears that L2 learners tend to deliberately and explicitly go
about learning another language (except for L2 children whose parents
migrate to a new language culture), their attempts to do so do not directly
impact all of the other things we’ve noted above. Explicit learning may
serve more of an affective factor, a point we’ll touch on below. For now,
we note that in most contemporary L2 research circles, language
acquisition happens to learners as they get involved in communicative
events. And because the product – mental representation – is far richer,

353
 more complex, and more abstract than any explicit learning or processing
could achieve, implicit processing and organization of linguistic data must
underlie acquisition. This would suggest that something is the same about
L1 and L2 acquisition.

With the above said, we do not suggest at all that the outcomes of L1 and L2
acquisition are the same. We have touched on ultimate attainment and
nativelikeness at various points in this book, and most directly in Chapter 5. We
would also point out, though, that the fact that learners attain different proficiency
levels does not mean that the underlying internal processes that get them there are
different. Now we turn our attention to external aspects of acquisition.
The flip side of the internal is the external. More specifically, do L1 and L2
acquisition share the same external process and mechanisms? And by external, we
mean things that are not language specific: the social environment, motivation,
aptitude, and so on. In this book we have not focused on the social environment,
save for the recurring boxes labeled “What about Social Factors?” We did touch
on individual differences such as motivation and aptitude in Chapter 8. In this
next set of bullet points, we list both what we believe the research tells us as well
as our own speculations about external processes and mechanisms.

Motivation does not seem to affect underlying processes in L2 acquisition.

Ordered development, the limited effects of instruction, the role of the L1,
the role of universals and Universal Grammar, and other language and
learner-internal aspects of acquisition are not altered or impacted in any
discernible way depending on a learner’s motivation or orientation toward
the language or the culture, for example. Motivation seems to affect two
things: (1) how far learners get (i.e., ultimate attainment), and (2) the
explicit and external strategies learners use to go about the task of
acquisition (e.g., taking classes, seeking out native speakers, watching
movies, taking notes, making flash cards). Clearly, L1 learners do not
engage such strategies nor is motivation implicated in L1 ultimate
attainment. So there is some external difference between L1 and L2

354
acquisition regarding motivation as part of an individual’s profile,
although this difference doesn’t affect internal underlying processes and
mechanisms.

There is little in the research on language aptitude that leads us to

conclude that it affects the underlying implicit processes in acquisition,
although there is clear evidence that it impacts explicit learning. Because
there is no theoretical reason and no empirical evidence to connect explicit
learning to underlying implicit processes in acquisition, it is safe to
conclude that the role of aptitude in acquisition is, at best, minimal. But
because aptitude is related to explicit learning and because for many
learners this is standard fare for “getting into a language,” aptitude is an
important part of the external processes and mechanisms for learners.
Again, this sets L1 and L2 acquisition apart from each other, as aptitude in
L1 learning is not considered a critical component implicated in the
acquisition process. At the same time, aptitude does not affect underlying
L2 processes, which is why we do not see aptitude affecting the role of
input, ordered development, the role of the L1, the role of universals and
Universal Grammar, and so on, as we have stated previously.

None of the three authors of this book is a researcher in social factors in

L2 acquisition although we are familiar with the scholarship in this area.
So here we offer some speculation based on our reading. Social factors
such as identity, power relationships, contextual roles, and others do not
affect the underlying processes of acquisition. That is, social factors do not
show up as variables that alter or affect ordered development, the role of
input, the role of the L1, the role of universals and Universal Grammar,
and so on. Instead, social factors are implicated in learners’ willingness
and ability to interact with level-appropriate input, as well as their
expectations about their own and others’ behavior. Social factors are
implicated in learners’ willingness and ability to seek out contexts that aid
rather than hinder their acquisition. These are powerful external factors, to

355
 be sure. As such, this makes L2 acquisition externally different from L1
acquisition in that social factors are not implicated in L1 acquisition, save
for socio-economic factors that affect dialects acquired and by-products of
language acquisition (e.g., literacy, discursive skills). L1 learners,
regardless of the social contexts into which they are born and regardless of
the social factors involved in these contexts, acquire their L1 (assuming
non-extreme cases such as “wild children” or abused children who are
isolated from normal human interaction). In short, social factors are more
important in the L2 context than they are in the L1 context even though,
again, they do not affect underlying and internal processes specific to
language.

From this discussion, we suggest to the reader that at their core (i.e., when it
comes to internal linguistic aspects of language acquisition) L1 and L2 acquisition
are fundamentally similar. However, outside of this core, when it comes to
external factors that are not linguistic in nature, L1 and L2 acquisition may be
fundamentally different. These differences do not affect underlying processes, but
they will affect such things as motivation, rate of acquisition, eventual outcome,
and in general how the L2 learner “feels” about the whole process of acquisition.
Such things, in turn, affect the quantity and quality of CEI that learners are
exposed to and their interactions with that input. Given the fundamental role of
input in language acquisition, the effect of external factors is not trivial when it
comes to both rate and outcome. At the same time, given the fundamental role of
learner-internal factors responsible for the processing of language, these external
factors are in a certain sense unimportant for how language itself comes to be
represented over time in the learner’s mind.
The point here is that how one addresses similarities and differences between
L1 and L2 acquisition depends on what one is looking at. And the relative
importance of internal and external factors is also dependent on whether one is
looking at internal linguistic development or something else, such as outcome and
attitudes toward L2 acquisition itself. As we see it, there is no problem in saying

356
something such as L1 and L2 acquisition are both fundamentally similar and
fundamentally different.
To be sure, there are some scholars out there who disagree with our
assessment – while others agree. Such disagreement hinges on definitions of
similarity and difference. For example, some scholars might argue that because
there is an L1 present in L2 acquisition and because we know that the L1
influences L2 acquisition, L1 and L2 acquisition must be different. As we said
above, it depends on how you examine such things. The presence of the L1 as a
factor in L2 development does not obviate the role of input, the role of universals
and Universal Grammar, the role of markedness, and it does not change the fact
that L2 acquisition is ordered and follows developmental stages. Our view is that
the L1 may slow things down or make L2 acquisition less efficient in some
aspects of acquisition, but it does not change the fundamental underlying
processes. To use a metaphor used elsewhere (see, for example, Bill VanPatten’s
2017 book, While We’re on the Topic, published by the American Council on the
Teaching of Foreign Languages), let’s imagine a woman going up six steps to a
landing. She puts one foot on one step, then the other foot on the next step, and so
on until she reaches the top. As long as she is physically unimpaired, she (and
anyone like her) will perform the same procedure to get to the landing. Now let’s
imagine that same woman carrying a 30-pound suitcase. She will probably
struggle somewhat to go up the steps – and she most likely will take more time.
However, she still has to put one foot on one step, then get to the next step, then
the next as she would without the suitcase. In both cases, she is fighting gravity
and there is no way around it. The suitcase-less scenario would be akin to L1
acquisition while the suitcase scenario would be like L2 acquisition. The woman
does the same thing in each scenario; she merely has extra baggage to contend
with in the second situation. Are the two scenarios fundamentally similar or
fundamentally different? We see them as fundamentally the same at their core
(e.g., going up steps, the basic body/leg movements, fighting gravity).

357
 Scholars involved in instructed L2 acquisition may also disagree with us.
Many scholars in this line of research place a high emphasis on explicit processes
in learning, and because of these explicit processes, they would argue that L1 and
L2 acquisition are fundamentally different. But as we have pointed out in this
book and as summarized above, it is not at all the case that explicit processes
(including explicit learning) affect L2 acquisition in any significant way. We
know that any observed short-term effects of instruction disappear with time. We
know that ordered and staged development is not altered in any observable or
significant way. We have also argued that explicit processes can’t be at the core
of input processing. So looking at the short term, a scholar might argue that L1
and L2 acquisition are different. Looking at what we have examined in this
volume (e.g., ordered and staged development, long-term effects, the limited to
non-role of explicit learning/processing), other scholars might conclude that L1
and L2 acquisition are basically the same because they are resistant to external
attempts to alter the processes. We fall into the latter group.
Although our position is that, at their core, L1 and L2 acquisition are
fundamentally similar processes, we understand that such a position is not popular
among certain sectors of L2 research (as briefly mentioned above) and is certainly
concerning to many teachers who look to L2 research for insights. But we think it
is important to underscore fundamental similarity because, in our experience, both
the field of SLA and the field of language teaching have tended to focus on the
differences. When we let the differences drive how we view the world, we lose
sight of the similarities. In the case of research, scholars may spend a good deal of
time (and publishing) on things that, in the end, don’t really tell us much about
how acquisition actually happens or what it means for an L2 learner to “know” a
language. They may not be able to link a narrow research agenda into the bigger
picture of what it means to acquire a language. In the case of teaching, by
focusing on differences, the profession may develop pedagogies and strategies for
the classroom that work against what the mind/brain wants to do naturally and,
indeed, what the mind/brain needs to do. So we believe that focusing on the

358
fundamental similarity between L1 and L2 acquisition is a good reminder about
the special nature of the human ability to acquire language. What this means for
teaching is that teachers (and learners!) can relax: Given access to input,
interaction with that input, and time, learners will acquire an L2. There are
important differences between the two, but these should not be the determining
factor in what we do. Finally, focusing on the similarities between L1 and L2
acquisition helps us see the complexity – and wonder – of language and language
acquisition more clearly and helps us to contextualize the differences better.

359
 Glossary

Acquisition.
Steven Krashen’s term for implicit learning. He defines L2 acquisition as
something similar if not identical to what happens in child L1 acquisition.
Learners are not aware of the processes they are engaged in or what they are
acquiring. The result is implicit knowledge.

Automatized knowledge.
In skill theory, knowledge that can be used in relevant contexts accurately and
efficiently without conscious awareness of doing so.

Backsliding.
The phenomenon by which a learner regresses to a developmental stage prior to
the one he or she is in.

Behaviorism.
A theory of learning prevalent in the twentieth century up until the 1970s.
Learning was seen as the acquisition of behaviors (observable phenomena) in
any living being (see also operant conditioning).

Blocking.
A term used in usage-based approaches to describe how previous knowledge
(either from the L1 or the L2) can inhibit the learning of something new.
Example: The strong association between subject (a grammatical notion) and
agent (a semantic notion) as in Bill carried John (Bill = both subject and agent)
is forged early on in acquisition. This can impede the learning of passives

360
where the subject is not the agent but the theme of the verb, as in John was
carried by Bill (John = subject but he is not the agent of the verb carry; he is
the theme of the verb carry).

Communicative output (communicatively embedded output).

Language that the learner produces that is part of the expression and
interpretation of meaning in a given context with a given (communicative)
purpose. Contrast with communicatively embedded input.

Communicatively embedded input.

See input.

Complementizer.
A word that introduces a clause (e.g., Bill said that Megan is smart, Alessandro
is the guy who speaks Italian, I don’t know if he will join us for dinner).

Comprehensible input.
That part of input that learners can make sense of in terms of meaning (i.e.,
what they can actually comprehend).

Critical period.
A specific window of time during biological development in which stimuli for
a particular ability have to be provided. If critical stimuli are not provided
during this window of time, the ability cannot be acquired.

The Critical Period Hypothesis.

An idea in language acquisition stating that if linguistic input is not provided
before a child is five, the child will not be able to learn the language. Applied to
L2 acquisition, it states that adults will not be able to acquire an L2 if they have
not been exposed to appropriate input before adolescence, if not earlier.

Declarative knowledge.

361
In skill theory, “knowledge that.” Knowledge that can be verbalized and
explained and that usually involves facts and details.

Developmental sequences.
Stages in which a particular structure or feature of language is acquired over
time. Examples include the developmental sequence for negation in English,
for linking verbs in Spanish, and for basic sentence word order in German.
Same as staged development or developmental stages.

Domain-specific learning mechanisms.

Learning mechanisms that are specifically designed to process (and organize)
linguistic input independently of other kinds of stimuli. These mechanisms are
central to generative linguistics.

Explicit knowledge.
Knowledge of language that is conscious and can be articulated by a person.
Example: A learner of English can say “Third-person verbs in English end in -
s.” A learner of French can say “With partitives, don’t use the article if the
sentence is negative.”

Explicit learning.
Processing linguistic input with conscious attention paid to the formal
properties of language.

Form–meaning connections.
How a meaning is expressed linguistically (e.g., the connection between “cat”
and its meaning as a feline, or the connection between -ed on a verb and the
concept of “pastness”).

Fossilization.
A term coined by Larry Selinker in 1972 to refer to the process by which L2
learners’ internal linguistic systems stop evolving. Fossilization results in a
non-native linguistic system or parts of the system.

362
Full Access.
The position claiming that the initial state of acquisition for L2 is just UG.
There is no L1 transfer at the outset of L2 acquisition (similar to the Initial
Hypothesis of Syntax).

Full Transfer/Full Access.

The position claiming that the initial state of L2 acquisition is the L1 mental
representation, but that learners have complete access to the contents of
Universal Grammar that subsequently guide and constrain acquisition.

Functional categories/features.
Refers to the abstract features of language provided by Universal Grammar
such as Case, Number, Aspect, Tense, Complementizers, and so on.

Garden-path effects.
Those “Huh?” moments that happen when comprehending sentences when the
intended sentence structure is not the structure that the parser originally
assigns to the sentence. The person usually has to stop and “reanalyze” the
sentence.

General learning mechanisms.

Cognitive structures designed to process information in the world and learn
something from it. Unlike domain-specific learning mechanisms, which apply
to language only, general learning mechanisms are used in all learning contexts
(see Usage-based approaches).

Generative linguistics (generative revolution).

The scientific study of language founded by Noam Chomsky in the late 1950s /
early 1960s. Two basic ideas lay at the heart of generative linguistics (and still
do): (1) only humans have language, so there is probably something innate
about language; (2) people can generate an infinite number of sentences using a
finite system of “rules,” including sentences they’ve never heard before and

363
couldn’t be imitating from their environment. For example, I know the man. I
know the tall man. I know the tall man over there. I know the tall man over
there in the brown coat. I know the tall man over there in the brown coat who is
smoking a cigarette. I know the tall man over there in the brown coat who is
smoking a cigarette and probably shouldn’t be. Minimalism is the current
version of generative linguistics. Universal Grammar (UG) is a major
construct in generative linguistics.

Generative tradition.
See generative linguistics.

Grammatical sensitivity.
The component of linguistic aptitude that is hypothesized to govern someone’s
ability to identify grammatical constructions.

Ideal L2 Self.
A component of the L2 motivational self-system that includes how learners see
themselves as users of the L2.

Implicit knowledge.
Knowledge of language that exists outside of one’s awareness whose contents
can’t be articulated by a person. Example: Except for trained linguists, people
can’t say why they prefer “Me and Alex went” compared with “I and Alex
went.” They simply say, “It sounds better.” (See also poverty of the stimulus
for examples of implicit knowledge.)

Implicit learning.
Processing linguistic input without paying conscious attention to the formal
properties of language.

Indeterminate intuitions.
Intuitions about whether something is grammatical in a language that are
“fuzzy” or not especially strong. For example, we might have a strong intuition

364
that undrink is not a possible word (I undrank the soda) but we might not be
sure about something like strived (is it strove or strived as in He strove/strived
to do it? Hmmm. I’m not sure).

Inductive learning ability.

The component of linguistic aptitude hypothesized to relate to how well an
individual can “formulate a rule” based on a set of language stimuli.

Inflectional morpheme.
Morphemes that are added to nouns and verbs that don’t change the
fundamental meaning of the root word and don’t change its function (i.e., the
noun remains a noun and the verb remains a verb). Example: A form of -s is
added to the end of nouns to indicate plurality (dog/dogs, house/houses,
snack/snacks).

Initial Hypothesis of Syntax.

A position claiming that the initial state of acquisition for L1 and L2 is the
same; that is, that both L1 and L2 learners begin the process of acquisition with
just Universal Grammar. There is no L1 transfer at the outset of L2
acquisition.

Initial state.
Refers to what learners bring to the task of acquisition at the outset. The
concern is largely with whether they transfer all the properties of the L1 at the
outset or not (see L1 transfer, the Initial Hypothesis of Syntax, Full
Transfer/Full Access, and Minimal Trees).

Input.
Refers to the language (in spoken, written, or signed forms) the learner is
exposed to in communicative contexts. Also called communicatively
embedded input.

Input Processing.

365
Refers to how learners make form–meaning connections during
comprehension/acquisition and how they process sentence structure during
real-time comprehension.

Instrumental motivation.
A desire to learn a language because it will be useful, usually for a professional
or educational reason.

Intake.
That subset of the input that learners can actually process at a given time and
make use of for acquisition.

Integrative motivation.
A desire to learn a language based on an interest in or affinity for aspects of the
target language culture(s).

Interlanguage.
A term coined by Larry Selinker that refers to an L2 learner’s mental
competence at any point in time during development.

Intervening variable.
Refers to a variable that researchers do not account for or cannot account for
during experimentation, which may affect results. Example: In research on the
outcomes of classroom learning, very often researchers cannot account for what
learners do outside of classrooms, which may affect the results.

L1 influence.
See L1 transfer.

L1 transfer.
Refers to the influence that the L1 has on L2 acquisition. It is not clear what the
influence is (for example, whether it’s on mental representation or input
processing or on both) or exactly where it happens, and different theories

366
define it in slightly different ways. It is a concept especially relevant to the idea
of the initial state.

L2 motivational self-system.
The ways that learners think about themselves in the present and who they want
to be in the future as related to the L2. This term has replaced integrative
motivation.

Language-related episode.
Refers to when learners engage in explicit discussion about language during an
interaction.

Learning.
Steven Krashen’s term for the explicit learning of rules and properties of
language typical in many classrooms. The result is explicit knowledge.

Lexical categories/features.
Refers to the concrete features of language provided by Universal Grammar
such as nouns, verbs, adjectives, prepositions, and so on.

Linguistic aptitude.
An innate ability to learn an L2 that varies among the population.

Long-term studies/short-term studies.

Refers to research on instructed L2 acquisition and how long after a treatment
the research measures the effects of instruction. Typically, short term refers to
tests following immediately after treatment or up to a month or two later. Long
term refers to testing learners many months or even a year after treatment.

Long(er)-term effects.
Refers to whether or not the effects of instruction are evident in long-term
studies.

367
Marked and unmarked structures (lesser/more marked).
See markedness.

Markedness.
A concept in linguistics with differing definitions. In general, it refers to how
typical/frequent/simple or not typical/infrequent/complex a linguistic structure
or form is. The most typical, most frequent, and/or simplest are termed
unmarked or the least marked structures/forms in a language. Example: In most
languages that mark plurality, the plural form is more marked than the singular,
which is considered unmarked. Thus dog is unmarked and dogs is more
marked. The latter is more complex and is less frequent than the singular form.

Meta-analysis.
A large-scale analysis of many studies focused on the same question or
problem. For an example, see Spada, N. & Tomita, Y. (2010). Interactions
between type of instruction and type of language feature: A meta-analysis.
Language Learning, 60, 263–308.

The Minimal Trees Hypothesis.

The position claiming that learners transfer only the lexical categories from the
L1 at the outset, but not the functional categories. In other words, L1 transfer
is constrained in the initial state.

Modified input.
Input that is adjusted by another speaker to a language learner based on a
perceived communication problem.

Monitored output
occurs when a learner has formally studied grammar and can apply consciously
learned rules and formal features to edit speech or writing as he or she produces
it.

368
Morpheme.
The smallest unit of language that carries meaning. The simple word book is
one morpheme. The word books consists of two morphemes: the morpheme
book and the plural marker -s that indicates “more than one.” Another example
of a morpheme is the prefix un- which denotes negation, as in invite and
uninvite (and of course, invite as a simple word is also a morpheme).

Morpheme orders.
The sequence in which morphemes are acquired over time. Example: In
English, verbal inflectional morphemes are acquired in this order: -ing before -
ed (regular past tense) before irregular past tense before 3rd-person -s.

Movement (Syntactic movement).

An operation that takes a word or phrase from one part of the sentence and
moves it to another part of the sentence. Movement happens without us
knowing (e.g., “What did you eat?” represents movement of the wh-word what
from its position as object of the verb eat : You eat what? → What did you
eat?).

Nativelikeness.
A mental representation for the L2 that is largely indistinguishable from that of
native speakers of that language.

Negative evidence.
Information about what is not possible in the target language, usually provided
through error correction but also through recasts, for example.

Negotiation of meaning.
The act of resolving comprehension problems during a communicative
interaction.

No interface position.

369
This is the position that implicit and explicit knowledge are distinct from each
other and do not influence each other.

Noticing Hypothesis.
Richard Schmidt’s proposal that L2 learners need to be aware of formal
properties in the input for acquisition to happen.

Noticing the gap.

The idea that learners can compare what they produce to what they hear/read to
see if there are differences.

Operant conditioning.
A major construct in behaviorism that attributed learning to a combination of
various factors, including stimulus, response, and feedback/reward. B. F.
Skinner was one of the major twentieth-century figures and scholars in the field
of behaviorism.

Ordered development.
Refers to morpheme orders, developmental sequences and other phenomena
suggesting that language follows particular paths as it develops in the learner’s
mind/brain over time. Example: For learners of English, the verbal morpheme -
ing is always acquired before past-tense verbal morphemes, which in turn are
always acquired before 3rd-person -s.

Ought-to L2 Self.
The component of the L2 motivational self-system that includes how learners
see themselves using the L2 in the future.

Output.
Language that the learner produces (compare with input).

Output as practice.

370
Language that the learner produces for the sake of producing the language or to
explicitly try to learn something.

Parameter resetting.
Refers to the “switching” of L1 values to L2 values for functional and lexical
features during the course of acquisition.

Parser.
The part of the mind that constructs a moment-by-moment interpretation of
spoken or written language as it encounters it. Example: The moment the parser
“hears” the word the it projects a noun phrase and if it “hears” it at the
beginning of a sentence in English will most likely posit it as the subject noun
phrase (until told otherwise as in garden-path experiences).

Phonetic coding ability.

A component of linguistic aptitude referring to an individual’s ability to
identify individual sounds and to form associations between sounds and
symbols.

Phonological short-term memory.

People’s capacity for recalling verbal information for short periods of time.
Phonological short-term memory is thought to be related to vocabulary
acquisition.

Planned discourse.
Refers to non-interactive language that is written out or thought of prior to
production, such as literary works, formal speeches that are read, or essays on
subject matter. Unplanned discourse would be the spontaneous use of language
found in interactions between two or more speakers.

Poverty of the stimulus.

The situation that arises when people come to know more about language than
they could have learned from exposure. Example: Speakers of English know

371
that you can add re- to verbs like paint, record, and decorate to make new
verbs: repaint, rerecord, redecorate. They also know that you cannot add re- to
verbs like pet (a dog), bake, and sleep : *repet, *rebake, *resleep. Restrictions
on the use of re- with verbs are not evident in the input English speakers are
exposed to and no one ever tells them what they can and cannot do with re-.

Procedural knowledge.
In skill theory, “knowledge how.” It refers to the ability to execute a task
without explicitly thinking about it even if the person has not yet developed
automatized knowledge.

Process.
The act of linking meaning with its form in the input during comprehension.

Processability Theory.
A theory concerned with universal constraints on how learners process
communicative output over time; that is, how they put words in serial order
during real-time spontaneous speech.

Rate of development.
Refers to the relative speed with which learners acquire language. Some are
faster than others, for example.

Recast.
When a speaker reformulates what a learner has just said in a more appropriate
or nativelike way.

Rote learning ability.

A component of linguistic aptitude hypothesized to relate to how well an
individual can learn/memorize arbitrary associations between sounds and
meanings.

Rule-like knowledge.

372
Linguistic knowledge that shows a pattern or a systematic nature whether that
system is adult-like or nativelike. For example, the child who consistently says
ated and wented is showing systematicity with regular past-tense endings. This
would be rule-like knowledge where the “rule” of regular past-tense endings is
used with irregular verbs.

Salience.
Not always well defined, it tends to refer to how a linguistic form or word
“stands out” in the input. Salience can involve stress or position in a sentence,
among other features. Example: In the sentence Where did you go? the wh-
word where is salient as it carries stress and is in sentence-initial position. The
auxiliary did is less salient as it does not carry stress and can even be contracted
as in Where’d you go?

Self-concept.
How learners see themselves; includes both positive and negative attributes.

Self-paced reading.
A research tool to examine people’s unconscious processing of sentences as
they read for meaning. The participant pushes buttons to reveal parts of a
sentence as they read, thus controlling the speed at which they read. For an
example, see VanPatten, B., Keating, G. D., Leeser, M. J. (2012). Missing
verbal inflections as a representational issue: Evidence from on-line
methodology. Linguistic Approaches to Bilingualism, 2, 109–140.

Staged development.
See developmental sequences.

Strong interface position.

This is the position that explicit knowledge can become implicit knowledge in
some way.

Structural linguistics.

373
The major school of linguistic analysis prior to the Chomskyan revolution.
Structural linguistics focused on patterns in order to state observable rules of
language.

Teachability hypothesis.
Refers to the claim that learners cannot be taught that which they are not ready
for in terms of developmental sequences. To say this another way, instruction
can only make a difference if learners are at the stage where the focus of
instruction would be acquired anyway. Example: If a learner cannot produce
verb separation in German (stage 3 of word order acquisition), that learner
cannot be taught subject–verb inversion (stage 4) (see Chapter 2 for details on
the acquisition of German word order).

U-shaped development.
The phenomenon by which a learner begins to do something correctly, then
seems to do it incorrectly, and then over time does it correctly again.

Ultimate attainment.
Refers to how far learners get along the acquisition continuum or how close
they get to nativelikeness.

Universal Grammar (UG).

An innate system in humans that constrains the nature of all languages. It
consists of an inventory of basic features (e.g., Tense/Finiteness, Aspect, Case,
Number) and principles (e.g., the nature of phrase structure). Example: All
languages consist of hidden phrases that exist in a hierarchical manner (they
stack on each other to make sentences) and all phrases consist of a head and its
complement. There is no human language that does not exhibit this universal
trait.

Universals (of language).

374
Refers to the approach called typology of language in which languages are
surveyed to see what features consistently co-occur (absolute universals) or are
a strong tendency (weak universals). Example: An absolute universal is that all
languages that have VSO word order also have prepositions (and not
inflections, for example, to carry the same meaning as prepositions). A weak
universal would be that SOV languages tend to have case marking (about 70
percent of SVO languages do, while only 14 percent of SVO languages have
case marking).

Uptake.
Refers to that moment in an interaction when a learner repeats something he or
she just heard.

Usage-based approaches (UBAs).

Approa-ches from the field of psychology that tie all human learning to an
interaction between external stimuli and internal general learning mechanisms.
A major component of UBAs is frequency and how it affects what gets picked
up by the learner and in what linguistic environments.

The Valueless Features Hypothesis.

The position claiming that L1 lexical and functional categories are transferred
to the L2 in the initial state. However, the strength of the features associated
with L1 functional categories does not transfer (see also L1 transfer).

Variation in L2 development.
Refers to learners’ use of different versions of a structure at a particular stage of
acquisition. Example: Learners of Spanish in Stage 2 of the acquisition of the
verbs ser and estar (both mean ‘to be’) make ser the default verb. However,
estar is required in progressive constructions (e.g., Juan está corriendo ‘John is
running’). So learners use ser instead of estar for progressives. Variation
occurs because learners can vacillate between Juan es corre, Juan es correr,
and Juan es corriendo. And variation occurs across learners because learner A

375
might have a “preference” for one version while learner B has a preference for
another. Yet both are in the same stage: ser is the default verb.

Weak interface position.

This is the position that explicit knowledge can somehow influence the
development of implicit knowledge.

Working memory.
The amount of information people can store and manipulate at any given time.
Working memory varies among the population. Working memory is usually
discussed in research on language comprehension.

376
 Index
Abrahamsson, Niclas 104–105, 174
abstract concepts, (problems/process of) definition 149
accents, “foreign” 28, 51, 90 see also dialect(s)
active vs. passive sentences 12
Alexopoulou, Theodora 21
ambiguity 34
Andersen, Roger 19–20, 21, 24
Andringa, Sible 159–161
animals
communication systems 68
critical periods in infancy 102
aptitude, linguistic 171–178, 188, 193–194
(alleged) faulty conceptualization 177
correlation with L2 outcomes 173–174, 184
; challenges to 174–177
defined 172
determining abilities 172–173
exemplary studies 175–176
and input processing 174, 177
military value 172, 178
socio-historical context of research 186
tests for 172–173
underlying assumptions 172, 174
Arabic
resumptive pronouns 27, 29–30
studies of L2 acquisition 104, 105–107

377
Army Method 8
Audiolingual approach 8
automatized knowledge 151–152

backsliding 26
behaviorism 2–3, 30
application to language acquisition 2, 8
critiques/objections 2–3, 6, 8–9, 30
decline in popularity 7
Behney, Jennifer 156
Belgium, multilingual culture 187
Bellugi, Ursula 6, 10, 22
Berko (Gleason), Jean 2, 6, 8
“this is a wug” experiment 3–4, 12
bi/multilingualism 186, 187–188
Bley-Vroman, Robert 11, 93–96
blocking 38
Brown, Roger 6, 8, 9, 17–19
Burt, Marina 10, 12, 19

Canada, biculturalism/L2 acquisition 181–182, 186

Canal-F test 173
Carroll, John 172–173, 177
case endings 37
children, language acquisition 2
active learning 2
“blank slate” theory 67, 88
correction/acceptance of errors 6–7, 95, 192
CPH applied to 102–103
data collection 11–12
degree of exposure, contrasted with adult 186–187
deprivation of opportunity 103, 194
development of cognitive structures 148–149

378
 developmental stages 6, 14, 15–16
and dialect 7
grammatical errors/mispronunciations 92
individual variations 168–170
initial state 67, 69
linguistic experiments involving 3–4
memory tests 179
mistakes avoided 5
parental/caregiver role 6–7
referential vs. holistic 169–170
relationship of comprehension and production 169
second language 73–74, 102, 103–104
social factors 7
speed of acquisition 46–47, 52, 168–169, 186
universal features 7–8, 18–19, 29–30, 168
vocabulary size 169
see also morphemes
Chinese
acquisition as L2 74–77
acquisition of English by L1 speakers 38
copular verbs 88–89
reflexive pronouns 75
Chomsky, Noam 9, 67, 151
Aspects of the Theory of Syntax 5
critique of Skinner 3
promotion of generative linguistics 4–5, 49
Clahsen, Harald 112
classroom vs. informal acquisition, compared 145
ambiguity of outcomes 127
(apparent) inferiority of instruction 127, 133–134
(claimed) superiority of instruction 130, 132–133, 136–137
divergent outcomes 126–127, 132–134
indeterminacy of outcomes 141, 142–143

379
 linguistic differences 130
motivational differences 130–131
rates of development 130–131, 143
shortage of research 131–132
similarity of outcomes 31–32, 42, 44, 105–107, 119, 122, 125–127, 128–129,
140
social factors 130–131
cloze-selection testing 133
cognitive psychology 178
communication
defined 46, 57
distinguished from linguistic knowledge 92
embedded factors 66
purpose of 66
communicative output, defined/distinguished from output as practice 53, 66
communicative output, role in L2 acquisition 53–63, 64, 192
as beneficial 56–63
exemplary studies 59–60
and formal structures 58–59
as having no role 56
and increase of input 63
as necessary 54–56
range of views on 54–56, 64
social factors 63, 66
triggering of learning processes 57–59
communicatively embedded input (CEI) see input
complementizer phrases 36, 69, 110
in Minimal Trees Hypothesis 78, 79–80
comprehensible input 45, 56–57, 64
conferences 14
constraints (on acquisition) 35–39, 41, 58
defined 35
types of 35–38

380
 see also markedness relationships
contractions 33, 34
copular verbs 88–89
Corder, S. Pit 8–9, 10–11, 13–14, 15, 47
Critical Period Hypothesis (CPH) 102–107, 118, 120
central premise 102
child/adult distinction 103
compared with FDH 122
complicating factors 104
counter-evidence 104, 118
difficulty of testing 103, 104–105, 118
empirical evidence 103–104
exemplary studies 104–107
theoretical exposition 102–103
Curcic, Maja 159–161

Dabrowska, Ewa 171

data collection 15
oral/comprehension 11–12
range of methods 12–13
declarative knowledge 151–153
DeKeyser, Robert 54, 151, 152, 153, 173–174
Denmark, cultural/linguistic mores 187–188
development (in language learning) xiii
rates of
see rates of development
developmental sequences 22–24, 29, 41
defined 22
individual variations 24
(limits of) L1 influence 23
linguistic variations/similarities 23
and negation 22–23, 24
overlap among stages 24

381
 universality 23, 24
dialect(s) 7
domain-specific learning mechanisms 93
distinguished from general learning mechanisms 96
Dörnyei, Zoltan 183
Doughty, Catherine 137
driving, types of knowledge involved 151–152
Dryer, Matthew 37
Dulay, Heidi 10, 12, 19
Dyson, Bronwen Patricia 127

education 171
Ellis, Nick 38, 48, 83, 148
Ellis, Rod 44, 125, 154
embedded clauses 99–100
English
(absence of) resumptive pronouns 27, 29–30
acquisition of negation 22, 29
acquisition of verbal forms 6, 29, 70, 128–129
childhood acquisition as L2 73–74
classroom vs. informal acquisition 125–127
contractions 33, 34
as focus of studies 20
interrogative forms 95, 109–111, 179–180
as L1, influence on L2 acquisition 38, 74–77, 114–115, 133–134, 175–176
overt subject, requirement of 97–101, 114–115, 116–117
reflexive pronouns 75–76
relative clauses 26–27, 132–133
similarities of L1/L2 acquisition 42
variation between L2 learners 113
word order in phrases 36, 69, 72–73
errors
correction/acceptance 6–7

382
 role in learning process 8–9
self-correction 32
Esperanto 159–161
Eubank, Lynn 80–81
explicit knowledge/learning xiii, 58, 124–125, 146, 150–166
(absence of) link with acquisition 60–61
assumed link with aptitude 174, 177, 184
confused with/distinguished from explicit teaching 148
defined 124
distinguished from implicit see under implicit knowledge
failure to define 148–149
influence on development of implicit 150, 154–161
(limited) role in acquisition 161, 163, 164–165, 166, 184, 192, 195–196
relationship with processing 157–161
researchers’ bias towards 137–138, 141, 143, 154
role of instruction 142–143
social factors 165
transformation into implicit 150–153
; alleged impossibility 150–153
eye-tracking, studies involving 156, 159–161

Felser, Claudia 112

feral children 103, 194
Fernández, Claudia 142
first language (L1) acquisition
compared with L2, xii, xiv see L1/L2 acquisition, compared
complexity of processes 3, 14
data collection 11–12
history of research 1–8
increased availability of data 6–8
individual variations 39, 121–122, 168, 170–171
influence on L2 acquisition see language transfer
knowledge beyond what one has been exposed to 32–34

383
 nature vs. nurture 168
role in processing 164
role of education 171
social factors, limited role 194
see also children
first-noun strategy 83–84, 157–159
Flynn, Suzanne 71–72
form–meaning connections 47–48
fossilization 9, 94
French
interrogative forms 95
motivation to learn 185
negative forms 23, 71
noun genders 49, 108–109
sentence structure 157–159
subject–verb agreement 107–108
“frequency tabulator” 38
Full Access Hypothesis 71–72
Full Transfer/Full Access Hypothesis 72–77, 88–89, 114–115
arguments for 73–74, 76–77, 82, 86, 115
central questions 114
constraints on 115
exemplary studies 74–77
“full access” aspect 115
“full transfer” aspect 114
“overwriting” process 73, 89
primary focus of studies 74
functional features 108
L2 learners’ (alleged) failure to adapt to 108–109
functional phrases 68–69
distinguished from lexical 35–36
role in Minimal Trees Hypothesis 78–79
structure/L2 acquisition 36–37

384
Fundamental Difference Hypothesis (FDH) 11, 93–101, 118, 120
central premise 96
compared with other systems 109, 115, 122
empirical flaws 11, 97–101, 118
logical flaws 96–97
observations on L2 acquisition 93–96

García Mayo, María del Pilar 59, 60–61

garden-path effects 112
Gardner, Robert 181–182
Gass, Susan M. 57, 58, 59, 63, 156
gender, of nouns 49, 108–109, 110
generative linguistics 4–5, 16, 35, 49–50, 146, 151
aims 5
evolution/constant features 5
increased prominence 173
and initial-state theories 67–68, 82, 86
(rise to) dominant position xv, 3
German
acquisition as L2, research into 31–32, 79
sentence structure 157–159
word order 23, 31–32, 70–71, 127
globalization 182–183
glosses 98
Goo, Jaemyung 136, 141
grammar, teaching of 144, 145, 167
grammatical information, exchange of 84–85
grammatical sensitivity 172
tests for 173–174
Greek, as L1 113
Greenberg, Joseph H. 37, 43

Håkansson, Gisela 127

385
Hatch, Evelyn 45–46, 56, 65
Haznedar, Bella 73–74
HiLAB test 173
Hulstijn, Jan 167
Hungarian, as L1 173–174
Hyltenstam, Kenneth 104–105, 174
“hypothesis space” xiii

identity, role in language acquisition 135

immersion programs 55
implicational hierarchy 26
implicit knowledge/learning xiii, 124–125
(alleged) total detachment from explicit 150–151
defined 124–125
(difficulty of) research 138–141, 143
distinguished from explicit 125, 141–142, 143, 144, 146, 167
impact of instruction 125, 127–128, 131, 140
L2 acquisition as system of 146–148, 161, 163, 192–193
mode of operation 157
impossible sentences 91, 95, 122–123
indeterminate intuitions 94–95
individual variations (in L2 acquisition) xiii, xiv, 93–94, 168, 171–172, 188, 189,
193
abilities linked to see aptitude; motivation; working memory
(lack of) explanatory factors 183–184, 188
rates of acquisition 184–187
researchers’ focus on 186–187
social factors 187–188
inductive learning ability 173
Initial Hypothesis of Syntax 70–71
and negation 71
and word order acquisition 70–71
initial state, in L2 acquisition 67–87

386
 compatibility of theories 86
first language as 72–77, 82, 86, 88–89 (see also Full Transfer/Full Access
Hypothesis)
generative perspectives on 67, 86–89
partial transfer theories 77–82, 86 (see also Minimal Trees Hypothesis;
Valueless Features Hypothesis)
social factors 87
Universal Grammar as 70–72, 86 (see also Full Access Hypothesis; Initial
Hypothesis of Syntax)
input xiii
as basis of acquisition theory 51, 184
central role in L1/L2 acquisition 191, 194–195
communicative context 44–45, 46, 63–64, 191
conversion to intake 47, 48–49
defined 44, 46
development of theory 49
exemplary studies 59–60
frequency in 48 (see also “frequency tabulator”)
increased by communicative output 63
methods of communication 45
necessity 49–52, 66
processing 83–84, 86, 125, 149, 157–161, 165, 174, 177, 180
(proposed) insufficiency for L2 acquisition 51
; flaws in theory 51–52
reliability 48
role in nativelikeness 119, 187
role in ultimate attainment 132–133, 134
scale of exposure to 46–47
simplified 65
SLA-specific usage 44
types of 45–46 (see also comprehensible input; modified input)
instruction 124–143
(absence of) long-term studies 134, 135, 142

387
 challenges to learners 130
compared with informal acquisition see classroom vs. informal acquisition,
compared
and the Fundamental Difference Hypothesis 95–97
impact on ultimate attainment 132–134, 143
implicit/explicit impact see under implicit knowledge
lack of impact 140, 184 (see also under ordered development)
negative effects 127, 133–134
role of xiii, 51, 66, 195–196
short-term studies 135–142, 143
intake 47
learners’ control of 47
interaction
and acquisition of formal structures 58–59, 61–63
benefits of 64
and explicit learning 60–61
interface
defined 116
“no interface” position 150, 165
weak 154, 165
see also Interface Hypothesis
Interface Hypothesis 114, 115–117, 118, 120
central proposition 116
examples 116–117
interlanguage 9
intervening variables 134
Ioup, Georgette 104
Italian
acquisition of English by L1 speakers 125–127, 132–133
null subjects/subject pronouns 116–117

Japan, English-speaking expatriates in 119–120

Japanese 21, 37

388
 as L1, influence on L2 acquisition 74–77, 180
motivation to learn 185
null subjects 97–101
social factors in acquisition 119–120
word order in phrases 36–37, 69, 72, 175
Johnson, Jacqueline 103–104
journals 14
Juffs, Alan 179–180

Kanno, Kazue 101

Keating, Gregory 138–141
Klima, Edward 22
Korean 112
Krashen, Stephen 10, 12, 19, 20, 21, 45, 56, 150

L1/L2 acquisition, compared xiii, 9–10, 15, 191–196

data-collection methods 11–13
differences of opinion 195–196
differing importance of social factors 194
differing outcomes 97
external mechanisms 193–195
faulty assumptions regarding 186–187
internal mechanisms 191–193, 194–195
(perceived) differences 11, 91, 93, 95–96, 108–109, 112, 117–118, 168,
171–172, 186–187, 193–195
(perceived) similarities 10–11, 42, 191–193, 194–195, 196
relative importance of internal and external factors 194–195
Lado, Robert 8
Lambert, Wallace 181–182
language, (debate on) nature of 165
language acquisition
defined 54
piecemeal nature 30

389
 speed of 30
see also first language acquisition; second language acquisition
language-related episodes 61
language transfer 9, 24, 184, 192
constraints on 30, 43, 192
developmental sequences 23
and markedness 27–28
markedness relationships 27–28
morpheme acquisition 20–22
(problems of) definition 28
reduction of efficiency of acquisition 192, 195
see also Full Transfer/Full Access Hypothesis; Minimal Trees Hypothesis; Valueless
Features Hypothesis
Lardiere, Donna 115
Larsen-Freeman, Diane 19
Latin 37
learning
complexity of process 161–164, 166
distinguished from acquisition 150
explicit/implicit see explicit knowledge; implicit knowledge
(problems of) definition 148, 165, 166
Leeser, Michael J. 66, 138–141
Lenneberg, Eric 102–103
lexical phrases 35–36
role in Minimal Trees Hypothesis 78–79
Lightbown, Patsy 127, 128–129, 140
linguistics 1–2 see also generative linguistics; structural linguistics
LLAMA aptitude test 173
Long, Michael H. 45, 56, 66, 130, 131
Luk, Zoe Pei-su 21

Mace-Matluck, Betty 19
Mackey, Alison 57, 60, 61–63

390
Marinis, Theodore 112–113
markedness relationships 26–28
constraints on language transfer 27–28, 29, 35
implicational hierarchy 26
role in L2 acquisition 27
see also relative clauses
McNeill, David 6–7, 192
McWhinney, Brian 48
memory see phonological short-term memory; working memory
mental representation
construction process 93, 115
data processing 149
implicit nature 147–148
indeterminate intuitions 94–95
relationship with nativelikeness 91–92, 100–101, 107–108
meta-analyses 136–137, 141–142
Minimal Trees Hypothesis 77–80
limitations 80
terminology 77–78
mistakes see errors
Modern Language Aptitude Test (MLAT) 172–173, 174
modified input 45–46, 56–57, 64
defined 45
monitored output 20
morphemes
defined 1, 17
inflectional 17–18
morphemes, order of acquisition 10, 17–22, 29, 128–129
individual variations 39
linguistic variations/influence of L1 20–22
monitored output 20
noun-related vs. verb-related 19, 21
relationship with spontaneous speech 20

391
 second-language 19–22
universality 18–19, 41
morphology xiv
L1 vs. L2 acquisition 108–109
overt 107–108
motivation 97, 130–131, 171–172, 181–183, 186, 188, 190, 193
ebb and flow 182
“good enough” approach 185
importance to L2 learners 185–184
instrumental 181
integrative 181, 182–183
and the self-concept 183
shifting focus of research 182–183
social factors 130–131, 181, 182, 183
movement 109–111
multilingualism see bi/multilingualism
Murakami, Akira 21

nativelikeness xiii, 90–120, 193

absence in L2 learners 51, 52, 90, 92–107, 117–118 (see also Fundamental
Difference Hypothesis)
age of acquisition, relationship with 90 (see also Critical Period Hypothesis)
defined 91–92
difficulty of definition 120, 121–122
partial acquisition by L2 learners 92–93, 107–113, 118–119 (see also
Representational Deficit Hypotheses; Shallow Structure Hypothesis)
(potential) acquisition by L2 learners 51–52, 90, 92–93, 104, 105–107,
114–117, 118–119, 120 (see also Full Transfer/Full Access
hypothesis; Interface Hypothesis)
social factors 119–120
nativism 49–50
negation, stages of acquisition 6, 22–23, 29
pre- vs. post-verbal 23, 71

392
 variations 24, 39, 40
negative evidence 95–97
negotiation of meaning 57–58
Newport, Elissa 103–104
no-transfer hypothesis see Initial Hypothesis of Syntax
non-native theories 49, 50–51
Norris, John 136, 137
Norton, Bonnie 135
Noticing Hypothesis 154–157, 165
criticisms 157
limitations 155–156
“noticing the gap” 57–58
noun–adjective agreement 84–85
noun phrases 68
nouns see gender; plurals
null subjects 33–34, 50, 97–101, 114–115
and the Interface Hypothesis 115–117

online studies 159–161

operant conditioning 2
flaws in theory 3
ordered development 9–10, 11, 17, 28–29, 40
individual variations 29
L1/L2 similarities 42, 191–192
(lack of) impact of instruction 125–127, 128, 144, 147–148
unaffected by individual abilities 184
universality 29–30, 125, 143, 168–169, 170
see also developmental sequences; markedness relationships; morphemes, order of
acquisition; U-shaped development
Ortega, Lourdes 136, 137
output xiii, 52–63
communicative see communicative output
defined 53

393
 in practice 53–54, 66
types of 53
“Output Hypothesis” 55–56
overt pronoun constraint (OPC) 100–101, 115–116

parameter resetting 73
parsers
dependence on syntactic information 112
L1/L2 differences 112
mode of operation 111
passive voice 137–138, 170
past tenses 16, 115
U-shaped development 25
Pavesi, Maria 119, 125–127, 130, 132–133, 145
Philp, Jenefer 63
phonetic coding ability 172
phonological short-term memory 179, 180
phonology xiv, 28
phrase structure 68–69, 70
“pidginization process” 135
Pienemann, Manfred 84, 85, 127
Pimsleur Language Aptitude Battery (PLAB) 173
Pinker, Steven 67, 88
planned discourse 126
Platzack, Christopher 70–72
Pliatsikas, Christos 113
plurals, formation of
learning of rules 3–4
U-shaped development 25
poverty of the stimulus 34, 43, 97, 118, 140, 147, 184
research on 35
procedural knowledge 151–152
distinguished from declarative 151–153

394
 relationship with automatized 152
process 47–48
Processability Theory 84–86
central principles 84–85
set of procedures 85
pronouns
overt/null subject 50, 99–100
reflexive 75
resumptive 27, 29–30, 126–127, 132
see also overt pronoun constraint
pronunciation, teaching of 145

questions, formation of 95, 179–180

and RDHs 109–111
and the SSH 111, 112

Ravem, Roar 10, 22

reading span tests 178–180, 189–190
recast, conversational 58–59
relative clauses 26–27, 126–127, 132–133
defined 26
resumptive pronouns 27
and the SSH 112–113
types of 26–27
Representational Deficit Hypotheses 107–111, 118
compared with other systems 109, 113
research methodology 140–141, 145
criticisms 72
Robinson, Peter 177
rote learning ability 173
Rothman, Jason 133–134, 141, 144, 167
rule-like knowledge 2, 7
Russian, sentence structure 157–159

395
Sagarra, Nuria 38
salience 38
Sapon, Stanley 172–173, 177
Savignon, Sandra 46, 54–55
Schmidt, Richard 155
Schumann, John 10, 22, 135
Schwartz, Bonnie 72, 114, 150–151
second language acquisition (SLA)
central features 184
classroom vs. informal see separate main heading
data collection 12–13, 30–31, 35
distinguished from learning 150
exemplary studies 31–32, 42
handbooks xiv
history of research xii, 1, 8–10, 13–14, 19–22, 44, 124
as implicit system 146–148, 161, 192–193
individual variations see separate main heading
knowledge beyond what one has been exposed to 32–35, 147–148, 184 (see
also poverty of the stimulus)
resistance to outside influence 192
social factors 40, 135, 194
speed of 46–47, 52, 130–131, 184, 186–187
underlying questions xii, xiv
see also SLA, compared with L1 acquisition
Second Language Research Forum (SLRF) 16
self-concept 183
self-paced reading 138–141
Selinker, Larry 9, 10
sentence structure(s)
impact of interaction on development of 58–60
phrasal 35–37
SVO/SOV 37, 43
see also active vs. passive sentences

396
Shallow Structure Hypothesis (SSH) 107, 111–113, 118
compared to RDH 113
research, mixed results 112–113
Shirai, Yashuro 21
Skehan, Peter 174
skill, linguistic 153
distinguished from acquisition 92
skill acquisition theory 54–55, 151–152
defined 151
(limited) application to L2 acquisition 152–153
limits of scope 153
role of explicit rules 152–153
Skinner, B. F. 2
critiqued by Chomsky 3
Slabakova, Roumyana 82
Smith, Megan 36–37, 175, 189
snowflakes, metaphor of 39–40
Sorace, Antonella 115–116
Spada, Nina 136, 141
Spanish
(absence of) resumptive pronouns 27, 29–30
(acquisition of) verbal forms 21–22, 23, 38, 133–134
copular verbs 88–89
implicit knowledge study 138–141
interrogative forms 112
as L1, influence on L2 acquisition 21, 180
negative forms 23
noun–adjective agreement 85
noun genders 49, 108–109
null subjects/subject pronouns 33–34, 50, 98, 114–115, 117
“personal a” 146–147, 150, 155
sentence structure 157–159
word order in phrases 72–73

397
Spinner, Patti 156
Sprouse, Rex 72, 114
structural linguistics 1–2, 16
decline in popularity 173
subject–verb agreement 84, 107–108, 110
learners’ ability to observe 140–141
subordinate clauses 32
Swain, Merrill 55–56, 57, 61
Swedish
acquisition as L2 104–105, 174
word order 70–71
syntactic information 77, 108
parser’s dependence on 112
syntax xiv

Tarone, Elaine 63
“teachability hypothesis” 127
teachers, native/non-native speakers as 90
tense phrases 36
in Minimal Trees Hypothesis 78
“this is a wug” experiment 3–4, 12
Tomita, Yasuyo 136, 141
Truscott, John 157, 164
Turkish (as L1), influence on L2 acquisition 73–74

U-shaped development 24–26, 29, 41

definition/derivation 24–25
in L2 acquisition 26
structures subject to 25
studies 26
ultimate attainment 91, 132–134, 143, 193
(absence of) long-term studies 134
effect of motivation 193

398
 in L1 acquisition 170
social factors 120
United States
Army, recruitment for linguistic aptitude 172, 186
monolingual areas 187
Universal Grammar 50, 67–72, 83
abstract features 68
constraint/guidance on language acquisition 69–70, 73, 86
defined 35
defining characteristics 68
and the FDH 93, 96, 101
and Full Transfer/Full Access 114, 115
human-specificity 68
as initial state (in L2 acquisition) 67, 70–72, 82, 86 (see also Full Access
Hypothesis; Initial Hypothesis of Syntax)
and the Interface Hypothesis 116
principles/constraints 68–69
properties not governed by 116, 118
treatment of sentence structure 35–37, 43
typological approach 37
uptake 58–59
usage-based models 38, 50–51, 82–83, 86, 146

Vainikka, Anne 77–80

Valueless Features Hypothesis 80–81
flaws 81
and word order 80–81
VanPatten, Bill 19–20, 21, 36–37, 46, 58, 83–84, 88–89, 130, 138–141, 142, 144,
146, 151, 157–159, 167, 174–176, 189, 195
variations, in language acquisition 39–40, 41
first-language 39
limited effect 39–40
social factors 40

399
verb phrases
in Minimal Trees Hypothesis 77
verbal forms, formation/order of acquisition 6, 21–22, 29, 128–129 see also past
tenses
vocabulary (mental lexicon) xiv
influence of working memory 179, 185–186

weak transfer see Valueless Features Hypothesis

Williams, John 157, 167
Wode, Henning 10, 22
working memory 171–172, 178–180, 188
correlation with L2 outcomes 180
defined 178
relationship with comprehension 180
relationship with syntactic knowledge 179–180
relationship with vocabulary acquisition 179, 180, 185–186
tests 178–179
wug(s) see “this is a wug” experiment
Wulff, Stefanie 83

Young-Scholten, Martha 77–80

Yuan, Boping 74–77

400