SLA Research and FLT Vanpatten
SLA Research and FLT Vanpatten
SLA Research and FLT Vanpatten
(1) canonical word order (the learner keeps elements in a sentence in XYZ order
without alteration), (2) adverb preposing (the learner begins to move adverbs to the
beginnings of sentences), (3) particle shift (the learner can split a past participle from its
auxiliary), and (4) subject-verb inversion. Peinemann's work indicates that learners
cannot skip stages and that the order of stages cannot be altered. (The stages are
accounted for by particular speech-processing constraints that change over time, but that
topic is beyond the scope of this essay.) Extending his work to English as a second
language, Pienemann has found that the stages involved in acquiring negation are also
immutable.
Of course, some learners go through a stage so quickly that they appear, to the instructor
(or researcher), who is not observing them twenty-four hours a day, to have skipped a
stage. There are, however, no published accounts of learners whose acquisition of
syntactic rules involved a reordering or absence of stages.
Finding 2. Explicit grammar instruction generally results in temporary gains unless the
learner is psycholinguistically ready for the instruction.
Researchers investigating whether explicit instruction has a positive effect on grammar
acquisition report two consistent tendencies. First, gains in the use of a grammatical
feature learned through instruction dwindle over time. That is, the learner's
improvement appears to be temporary. Studies have examined the effect of instruction
both on syntax (e.g., Ellis, Classroom ; White, Learnability; VanPatten,
Acquisition) and on the acquisition of morphemes and functors (e.g., Lightbown,
Exploring; Pica; Terrell et al). In addition, accuracy in using a feature or rule may
only manifest itself when there is a focus on form for example, in a discrete-point test
or in slow, monitored speech or writing (e.g., Schumann; Krashen; Kadia).
Second, what is teachable seems limited to what is learnable. Pienemann has
demonstrated that in the acquisition of German word-order rules, outlined in finding 1
above, teaching a structure from stage 4 does no good if the learner is at stage 2
(Learnability; Constraints). However, the learner at stage 2 who is taught a rule from
stage 3 may move to stage 3 more quickly. In short, what can be effectively taught must
coincide with what the learner would naturally acquire next without explicit instruction.
Pienemann has also shown that teaching a rule at a stage far too advanced for the learner
can sometimes cause the learner to retreat to a previous stage. The research does not
address such things as verb morphology and nonsyntactic aspects of language; to my
knowledge, no research exists on psycholinguistic readiness and the acquisition of
morphemes and functors.
Finding 3. Correcting errors in learner output has a negligible effect on the developing
system of most language learners.
Of all the findings on the effects of instruction, the one least likely to be accepted is that
error correction does little good; language teachers simply don't want to believe this.
Many instructors, including those who have written detailed descriptions of correction
techniques, have apparently not examined the relevant research. The constraints on the
effect of explicit instruction outlined in findings 1 and 2 suggest that the value of error
correction is widely overestimated.
In several studies, researchers have concluded that direct correction of learners' errors
has no significant effect. These studies examined both oral output (e.g., Schumann;
Plann; Dvorak; Holley and King) and written output (e.g., Hendrickson; Cohen and
Robbins; Semke). Holley and King found that learners sometimes improved their
accuracy if they were given enough time to rephrase their utterances. Dvorak reports
that learners in classrooms in which overt correction was absent performed just as well
as, if not better than, those students who had overt correction all semester long. Studies
of learners writing in English as a second language have shown that error correction
(both systematic and unsystematic) does not result in increased accuracy of grammar in
compositions over time. Semke, who investigated grammatical accuracy in dialogue
journals rather than in compositions, found that learners who received personal
responses rather that corrective feedback from their instructors actually outperformed
correction groups in measurements of general language proficiency at the end of the
treatment period.
In one contradictory study, Lalande found that a particular type of systematic error
correction helped second-year university students of German increase the accuracy of
their compositions. I have discussed these results elsewhere (Juries), arguing that one
interpretation of increased variation around statistical means might lead proponents of
error correction to be less sanguine about Lalande' findings. It is quite possible that only
certain students in his error-correction group benefited from correction (what
researchers call outliers) but that there were enough of them to cause a statistically
significant difference between the error-correction group and the other subjects in his
study. Moreover, the correction process that he researched was so focused on form (it
used a detailed coding scheme and charts in which students and instructor monitored
progress) that it may have pushed learners to monitor their output more but not
necessarily to acquire more. 1 A causative relation between monitoring and acquisition
is tenuous at best.
Finding 4. Classroom learners tend to acquire more language and have greater
accuracy in performance than their nonclassroom counterparts.
Researchers who cite an effect for instruction often couch their claims in terms like
those in finding 4 (e.g., Long; Ellis, Instructed ). In general, classroom learners acquire
more grammar and perform more accurately over time than nonclassroom learners do.
Studies that support this viewpoint include Pica's investigation of morpheme acquisition
in ESL, Pavesi's research on the acquisition of relative clauses in ESL, Ellis's work on
the acquisition of German word order (Naturalistic Acquisition), and others. At first
glance, the research seems quite convincing: all the studies just cited, for example,
reveal that classroom learners are more grammatically accurate in production than
nonclassroom learners are and that classroom learners sometimes use a wider range of
grammatical devices (e.g., more types of relative clauses) than do their non-classroom
counterparts. Before we take this as evidence that contradicts findings 13 above,
however, one thing should be made clear: no study comparing classroom and
nonclassroom learners has ever been able to pinpoint explicit instruction and error
correction as the specific cause of the observed differences. In other words, classroom
activity involves much more than grammar instruction; in fact, Pavesi attributes her
findings not to explicit instruction but rather to the different kinds of language that
classroom and nonclassroom learners are exposed to (i.e., planned and formal discourse
vs. conversational discourse). She argues that classroom learners have easier and more
consistent access to input that advances the learners' developing grammatical system
than do nonclassroom learners, who are less likely to encounter this type of input in the
environments in which they live and work. Elsewhere I argue that Pica's findings (and
probably most of the findings on the effect of instruction and the acquisition of noun
and verb final morphemes in English) may be due to a sociolinguistic and dialectal
transfer of the first-language phonological system (Juries). Further, classroomnonclassroom comparisons are questionable because they normally cannot control for
differences in affective and socioeconomic factors that are known to correlate with the
level of proficiency attained. Since classroom learners often self-select language study
and differ from nonclassroom learners in educational background, the two groups are
unlikely to have the same motivation.
Thus, while the focus on grammar that exists in many language classes might seem to
account for the differences observed in the performances of classroom and
nonclassroom learners, a causal connection has not been adequately demonstrated.
Because findings 13 on route of development, accuracy, and error correction all
indicate that instruction does not have a long-term effect, we should be cautious in
attributing the superior performance of classroom learners to a focus on grammar during
instruction.
Many instructors, particularly those who believe they see improved performance in their
students, do not easily accept the limited influence of explicit instruction and feedback.
Current research and theory, however, suggest that the limited kind of data a language
acquirer can use also limits the effectiveness of instruction.
Finding 5. For successful language acquisition, learners require access to
comprehensible and meaningful input.
Second-language learners cannot learn from a steady diet of grammatical instruction
and practice. They need input that has two basic characteristics. First, it must be
comprehensible, so that learners can understand the sentences they see or hear. Second,
input must encode some referential meaning to which learners can respond. Thus, much
corrective feedback, most (if not all) pattern practice, and many explanations of
grammatical concepts are processed not as input for acquisition but as knowledge about
the language (hence the general findings cited previously). In other words, learners need
to hear and see language that is used to communicate messages. Comprehensible and
meaningful inputthough necessary for successful acquisitionis not sufficient to
ensure it. (For the role of input, see Hatch; Krashen; Gass and Madden; Ellis, Instructed
and Understanding ; White, Universal Grammar. )
In a recent colloquium held at Concordia University in Montreal, Schwartz argued that
explicit instruction and error correction have no effect on language acquisition.
According to Schwartz, language acquisition takes place in what Fodor calls the
language module of the brain. For acquisition to occur, the language module needs
primary linguistic data, that is, utterances in the language. In Schwartz's
conceptualization, explicit instruction and error correction do not supply primary
linguistic data and therefore do not get processed as data for acquisition. Schwartz
suggest that explicit instruction and error correction can result in learned linguistic
knowledge that resides in the brain apart from the language module. This type of
knowledge is, in effect, knowledge about the language.
At the same colloquium, I attempted to redefine the notion of explicit instruction and
reported the findings of recent research that tested this new definition (see VanPatten
and Cadierno). Because comprehensible and meaningful input is a necessary ingredient
for second-language acquisition, we can sketch the three distinct sets of processes
involved in language acquisition as shown in figure 1.
The first set of processes converts input to intake, where intake is the subset of the input
that is comprehended and attended to. From intake the learner must still develop an
acquired system; that is, not all intake is automatically fed into the acquired system. The
processes in the second set promote the accommodation of intake and the restructuring
of the developing linguistic system. They include the kinds of processes discussed by
White, Universal Grammar , and McLaughlin. Finally, since output studies do not show
that the learner's language is a direct reflection of acquired competence, a third set of
processes must be posited to account for certain aspects of language production.
Traditional foreign language grammar instruction usually focuses on the manipulation
of the learner's output. That is, teachers explain a grammatical concept and then have
learners practice producing a given structure or form (see fig. 2). Because
comprehensible and meaningful input plays an important role in second-language
acquisition, the value of instruction that stresses the practice of grammar in output is
questionable. In figure 1, the input data for the developing system flow into the system.
In other words, the arrows go from left to right and not from right to left. Instead of
manipulating learners' output to effect change in the developing system, instruction
might seek to change the way learners perceive and process input. This scheme is
depicted in figure 3. Theoretically, altering input processing should have a significant
influence on changing the internatized knowledge. 2
In a 1991 study (VanPatten and Cadierno), we found that students benefit more from
instruction focused on input than they do from instruction focused on output. We taught
different groups of students about Spanish direct objects and object pronouns in two
different ways: with traditional instruction, based on a popular first-year university-level
textbook, and with processing instruction, in which we taught learners to perceive and
process object pronouns and word order in a way different from that documented in
second-language-acquisition research (LoCoco; VanPatten; Comprehension).
Traditional instruction involves explanation followed by practice with output;
processing instruction involves explanation followed by attention to input for meaning.
In short, the students in the processing group produced no output. We found that
processing instruction had a positive effect on the acquisition of object pronouns and
their availability for use in output, even though the subjects never once produced on
object pronoun. We also discovered that students given processing instruction improved
their ability to make correct form-meaning connections when listening, whereas
students in the traditional-instruction group showed no gains in understanding what
object pronouns referred to when listening. The results further suggest that processing
instruction affects the learners' developing system, whereas traditional instruction
affects the learned linguistic system described by Schwartz.
In summary, theoretical arguments that explain the limited effect of traditional grammar
instruction also provide new avenues for exploring the role of explicit instruction. With
further research on the effect of a focus on input, we may find a more appropriate role
for the explicit teaching of grammar in communicative language teaching.
At the present time, second-language research cannot offer a detailed answer to the
question, posed in part 1 of this essay, of what grammar can most profitably be taught
and when. However, research does support those approaches to language instruction that
move away from lockstep grammatical syllabi and a heavy emphasis on accuracy. Many
professionals in this field have colleagues whose traditional approaches to language
teaching place great emphasis on explicit grammar teaching and practice. Secondlanguage-acquisition research can serve as a useful tool for engaging these colleagues in
discussions that may lead them to try different methodologies or approaches to the
language curriculum.
Knowledge about second-language-acquisition findings can also help teachers in
training form appropriate attitudes about language learning (see Lightbown, Great
Expectations). Often the student teacher's approach to instruction is, Oh my God if I
don't teach the grammar right and correct everything then the students won't learn and it
will be my fault. I find that discussing second-language-acquisition research with
student teachers helps them understand that, ultimately, learners are in control of
acquisition and that teachers can only facilitate the process. While still concerned about
pushing learners along a path to better performance, these student teachers look for
novel and engaging ways to get input into the classroom, to get learners to attend to the
input, and to make judicious use of explicit grammar instruction and practice.
Second-language research is still a young field. Although significant advances have
been made in theory construction and research methodology, many questions remain. As
these questions are answered new ones will undoubtedly spring up. In time,
practitioners and classroom researchers will be able to use the research to formulate
specific techniques and methods, but the current research serves to inform, not to
prescribe. As such, it is a valuable reminder that teaching cannot be equated with
learning.
The author is Associate Professor of Spanish at the University of Illinois, Urbana. This
paper was prepared by the author in conjunction with a course he taught at the 1989
Summer Linguistic Institute cosponsored by the Modern Language Association and the
Linguistic Society of America at the University of Arizona.
Notes
Krashen has pointed out to me that the gains made by Lalande's correction group,
while significantly different statistically from those made by the other groups, were so
small that they raise the practical question of whether or not such correction is
worthwhile on an ongoing basis.
2
I am not suggesting here that practice in using language may not be important. See
Sharwood-Smith.
Works Cited
Figure 1:
Processes in second-language acquisition
Figure 2:
Traditional grammar instruction
Figure 3:
Processing grammar instruction