Lings_P2_M32: Generative Phonology

31. 1 Introduction
In this module, we are concerned with presenting the model of the theory of
Generative Phonology presented in Chomsky & Halle’s 1968 book, Sound Pattern of
English (also known as SPE). We propose to discuss the following topics as
introduction to the theory of Generative Phonology: its origin as a critique of
American Structuralist Phonology, its main tenets, the illustration of the tenets with
phonological data, the notion of alternations predicted by rule, a preliminary
discussion of interaction among phonological rules. In the end we offer a brief
discussion of the Halle’s 1962 paper.
31.2 The Main Premises of Generative Phonology
The model of the theory of generative phonology presented in SPE was practiced for
well-over two decades, with its first formal presentation in Morris Halle’s 1962
paper, ‘Phonology in Generative Grammar’. The main premises of the theory (see
Kenstowicz
http://web.mit.edu/linguistics/people/faculty/kenstowicz/generative_phonology.p
df) are the following:
(1)
a. Phonology is part of a generative grammar and phonological knowledge
is part of linguistic knowledge in that phonological structure reflects the
linguistic competence of the individual native speaker;
b. The competence of the individual native speaker lies in the ability to
compute the phonetic representation on the basis of which to interpret
the syntactic structure of a sentence in a language;
c. Phonological competence as part of an individual speaker’s linguistic
competence can be investigated following the methods of normal science.

These premises of the theory of generative phonology have proved very fertile in
influencing the growth of knowledge of sound structure in diverse fields such as
language acquisition, language change, language loss and language cognition. The
theory has grown because of its scientific emphasis on refutation and modification,

1
the two main characteristic features of normal science. As we will see in the course
on Advanced Phonology, Paper 5), the premises continue to hold grounds,, even
though many of the tenets of theory (discussed in the following section) have come
to be revised.

31.3 Criticism of American Structuralist Phonology

1. Criticism of Bllomfieldian and support of Sapir’s model of phonology

http://en.wikipedia.org/wiki/Phoneme

Biuniqueness[edit]
Biuniqueness is a requirement of classic structuralist phonemics. It means that a given phone,
wherever it occurs, must unambiguously be assigned to one and only one phoneme. In other words,
the mapping between phones and phonemes is required to be many-to-one rather than many-to-many.
The notion of biuniqueness was controversial among some pre-generative linguists and was
prominently challenged by Morris Halle and Noam Chomsky in the late 1950s and early 1960s.

An example of the problems arising from the biuniqueness requirement is provided by the
phenomenon of flapping in North American English. This may cause either /t/ or /d/ (in the appropriate
environments) to be realized with the phone [ɾ] (an alveolar flap). For example, the same flap sound
may be heard in the wordshitting and bidding, although it is clearly intended to realize the
phoneme /t/ in the first word and /d/ in the second. This appears to contradict biuniqueness.

For further discussion of such cases, see the next section.

Neutralization and archiphonemes[edit]

Phonemes that are contrastive in certain environments may not be contrastive in all environments. In
the environments where they do not contrast, the contrast is said to be neutralized. In these positions
it may become less clear which phoneme a given phone represents. Some phonologists prefer not to
specify a unique phoneme in such cases, since to do so would mean providing redundant or even
arbitrary information – instead they use the technique of underspecification. Anarchiphoneme is an
object sometimes used to represent an underspecified phoneme.

An example of neutralization is provided by the Russian vowels /a/ and /o/. These phonemes are
contrasting in stressed syllables, but in unstressed syllables the contrast is lost, since both
are reduced to the same sound, usually [ə] (for details, see Vowel reduction in Russian). In order to

2
assign such an instance of [ə] to one of the phonemes /a/ and /o/, it is necessary to
consider morphological factors (such as which of the vowels occurs in other forms of the words, or
which inflectionalpattern is followed). In some cases even this may not provide an unambiguous
answer. A description using the approach of underspecification would not attempt to assign [ə] to a
specific phoneme in some or all of these cases, although it might be assigned to an archiphoneme,
written something like |A|, which reflects the two neutralized phonemes in this position.

A somewhat different example is found in English, with the three nasal phonemes /m, n, ŋ/. In word-
final position these all contrast, as shown by the minimal tripletsum /sʌm/, sun /sʌn/, sung /sʌŋ/.
However, before a stop such as /p, t, k/ (provided there is no morpheme boundary between them),
only one of the nasals is possible in any given position: /m/ before /p/, /n/ before /t/ or /d/,
and /ŋ/ before /k/, as in limp, lint, link ( /lɪmp/, /lɪnt/, /lɪŋk/). The nasals are therefore not contrastive in
these environments, and according to some theorists this makes it inappropriate to assign the nasal
phones heard here to any one of the phonemes (even though, in this case, the phonetic evidence is
unambiguous). Instead they may analyze these phones as belonging to a single archiphoneme, written
something like |N|, and state the underlying representations of limp, lint, link to be |lɪNp|, |lɪNt|, |lɪNk|.

This latter type of analysis is often associated with Nikolai Trubetzkoy of the Prague school.
Archiphonemes are often notated with a capital letter within pipes, as with the examples |A| and |N|
given above. Other ways the second of these might be notated include |m-n-ŋ|, {m, n, ŋ}, or |n*|.

Another example from English, but this time involving complete phonetic convergence as in the
Russian example, is the flapping of /t/ and /d/ in some American English (described above
under Biuniqueness). Here the words betting and bedding might both be pronounced [ˈbɛɾɪŋ], and if a
speaker applies such flapping consistently, it would be necessary to look for morphological evidence
(the pronunciation of the related forms bet and bed, for example) in order to determine which phoneme
the flap represents. As in the previous examples, some theorists would prefer not to make such a
determination, and simply assign the flap in both cases to a single archiphoneme, written (for example)
|D|.

For a special kind of neutralization proposed in generative phonology, see absolute neutralization.

The Main Tenets of Generative Phonology

Goldsmith & Lacs point out four tenets of Generative Phonology

(http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.170.604&rep=rep1&type=pdf

The tenets are summarized as follows:

(2)

a. First, the goal of the theory of generative phonology, in accordance with

the theory of generative grammar, was to develop an explicit phonology
that generates all and only the surface forms of a language.

3
 b. Second, the phonological representations in the explicit phonologies
“were linear sequences of matrices of feature values.”

c. Third, in order to generate the forms of a language, phonology “should

employ derivational means (that is, sequential, processual analyses)…”

d. Fourth, a high priority of the theory was the discovery and empirical
establishing of explicit rule ordering which was consistent with the data.

A sub-tenet could be added to the fourth tenet above: The ordered rules could
apply cyclically or non-cyclically.

We will try to understand the working of these tenets.

2. Explanation of each tenet with examples

Two levels of representation. Counter position- Null Hypothesis (Kenstowicz &
Kisseberth)

Alternants

Phonologically Conditioned

Morphologically Conditioned

From SIL

(73) Noun Verb

house house [U], [\]
advice advise [U], [\]
breath breathe [6], [&]
mouth mouth [6], [&]
bath bathe [6], [&]
strife strive [H], [X]
belief believe [H], [X]
grief grieve [H], [X]
thief thieve [H], [X]
You should recognize that the feature [voice] is again consistently changing in these pairs. The
voiceless fricative in the noun is replaced by its voiced counterpart in the verb. If the noun is considered
basic (simply an assumption at this time), then one might propose a rule such as the following that operates
in the derivation of the verb form:
(74) Make the final consonant of the stem [+voice].
But again, it is not just any consonant that participates in this change. There are no noun-verb pairs in
English such as stick–*stig, or pipe–*pibe. And it is not just a haphazard set of consonants, but a natural
class. The following chart includes the consonants of English and the feature specifications for two
important features. (The definitions of these features are not important at this point; but see Appendix A.)
(75)
R D V_ F M I V5 F< 5 < H X 6 & U_ \

4
 sonorant – – – – – – – – – – – – – – – –
continuant – – – – – – – – + + + + + + + +
O P N_ . L Y J
sonorant + + + + + + –
continuant – – – + + + +
Of these two features, neither one by itself selects exactly the set of consonants that participate in the
noun-verb alternations. The feature [+continuant] comes very close. The combination [-sonorant,
+continuant] comes even closer; it specifies the natural class of “fricatives”. If we really wanted to specify
a
smaller set, “fricatives that are pronounced in the front of the mouth”, we would have to add another
feature. This illustrates the point that we made above that the fewer the features specified, the larger the set,
while the more the features specified, the smaller the set.
Key Concepts
features [voice] natural class
evaluation of simplicity
42
6.1 Try it for yourself (features and natural classes)
Assume the following inventory of sounds:
R__V__M__D__F__I__H__U__5__X__K__#__W__
Which of the above sounds would be included in the natural classes that each of the following features or
feature combinations define:44
1. [+voice]
2. [+labial] (sounds made with the lips as active or passive articulators)
3. [-voice], [+labial]

Lexically Conditioned

3. Accounting for alternations (Carr, Jensen)

Three types of alternations and three types of sound substitutions (Silverman)
Distributional restrictions (allophones) and conditions on alternation (Peng)
The instrumental role of “the concept of underlying representation [in]…
uncovering the patterns of alternation” (Peng) foam ~ phone (Silverman)

More on nasals in English. Nasals in Hindi

Interaction among different alternations, “creating complex pattern interactions”

(Peng)
4. Discussion of Halle’s 1962 paper