Generative Phonology

 

Karim Nazari Bagha,

Islamic Azad University – Astara Branch.

 

Phonetics is the study of speech sounds. "Phonetics gives a purely acoustic or physiological description of sounds" (Bloomfield, 1958: 137). But phonology (or phonemics) is the study of sound systems: that is, it studies how speech sounds structure and function in languages. Phonology studies the distinctive features in a given language [see below].

A phonetic study tells us how sounds are made in a language: it is concerned with the physical properties of speech sounds: but phonology tells us how these sounds are used to convey the meaning.

Phonetics tells us that /p/, /t/, /k/, in English are aspirated word initially, as in ‘pie’ [p^haɪ], ‘tie’ [t^haɪ], ‘kite’ [k^haɪt] respectively; and that they are unaspirated afer the segment /s/, e.g. ‘spy’ [spaɪ], ‘stay’ [steɪ] and ‘sky’ [skaɪ]. Phonology, on the other hand, will tell us that aspiration is not distinctive in English: whether you use aspiration with /p,t,k/ in the word initial positions or not, it will not make a change in meaning. But the phonology of the language Thai will tell us that aspiration is distinctive in that language. In their words for ‘split’ [p^haa] and ‘forest’ [paa], both [p^h] and [p] occur word initially; so they are two phonemes. The aspiration is distinctive: it makes a change in meaning.

To give an example from English: The minimal pairs ‘Sue’[1] /su/ and ‘zoo’ /zu/ show us that /s/ and /z/ represent two contrasting phonemes in English. /s/ is [-voice] or ‘voiceless’, whereas /z/ is [+voice] or ‘voiced’. These two words differ in meaning just because of voice. So we can conclude that voice (the vibration of the vocal cords) is significant or distinctive for consonants in English. With this feature a;one, /p,b/, /k,g/ and other pairs are distinguishable in English.

Phonology is the study of these distinctive features of a particular language and generative phonology is part of a comprehensive theory of language known as transformational generative grammar.

A. Distinctive and Redundant Features

Let us begin with an example. The sound /p/ in the morpheme {pea} has the following phonetic make-up:

	[p	i]
1) Vocalic	-	+
2) Consonantal	+	-
3) Continuant	-	+
4) Nasal	-	-
5) Abrupt release	+	-
6) Lateral	-	-
7) Voice	-	+
8) Tense	+	+
9) Aspiration	+	-
10) Strident	-	-
11) Anterior	+	-
12) Coronal	-	-
13) High	-	+
14) Low	-	-
15) Back	-	-
16) Round	-	-

The above chart is the phonetic representation of the morpheme {pea}, i.e., it shows how this morpheme is actually pronounced by the speaker. But there are certain regularities among these features that we have overlooked. Some of the values can be predicted on the basis of values for other features. That is to say, the above matrix contains some amount of redundancy. The above matrix contains both distinctive and predictable features. The predictable features could be eliminated from the matrix by a set of phonological rules. [see below].

First, let us assume that the phonemic or underlying representation of the morpheme {pea} has the following feature make-up: [-vocalic, +consonantal, +abrupt release, -voice, +anterior, -coronal].

The speaker learns the word ‘pea’ with only these six distinctive features for the first sound /p/. He internalizes these six distinctive features. The remaining ten features are predictable: they can be eliminated by phonological rules.

Then, we will argue that the above six feature make-up is the right assumption:

1) We know that /p/ (and /t/, /k/ and /č/) have the feature [+aspiration] in word-initial position. Since /p/ in {pea} is used word-initially, we know that it is [+aspiration]. So we will eliminate this feature from the matrix.

2) If /p/ is [+abrupt release], then it cannot be a fricative; so the feature [strident] is not necessary on the matrix.

3) [-voice] shows that the segment is not a nasal (because nasals are [+voice] in English). So the feature [nasal] is removed from the matrix.4) [+anterior] means that the segment is not /k/; therefore, features [back] and [round] are not needed.

5) We know that [-voice] stops are [+tense]; therefore, [+tense] is eliminated from the matrix.

6) [+abrupt release] shows that there was an obstruction on the way of the segment; therefore, [continuant] is not necessary on the matrix.

7) Since liquids have the feature matrix [+vocalic, + consonantal], and since /p/ is [-vocalic, +consonantal], therefore this segment is not a liquid. So the feature [lateral] can be omitted from the matrix.

8) In English, all [-high] consonants are [-back]; therefore, [back] is omitted from the above phonetic representation.

In this way, we have eliminated ten redundant features from the above phonetic representation. This we have done through phonological rules. This is what the speakers do to minimize the number of features in the phonemic (or mental) representation for each morpheme. The speaker is left with six distinctive features for the /p/ sound in the morpheme {pea}. These six distinctive features distinguish this morpheme from all other morphemes in the language. Those ten redundant features are not memorized by the speaker. In his actual pronunciation of the word ‘pea’, he will produce 16 features: six distinctive and ten redundant features.

From the above discussion it becomes obvious that speakers have three different kinds of knowledge about their sound system:

1) The phonetic representation of the lexical items like ‘pea’. Each morpheme has a set of phonetic features. /p/ in {pea} has 16 such features.

2) For each morpheme, the speakers know which features are necessary and distinctive in order to distinguish this morpheme from all other morphemes. This knowledge is referred to as the phonemic representation of the morpheme. This knowledge is stored in the mind: it is a mental representation. It is part of the speaker's linguistic competence.

3) The speakers know a set of phonological rules as well. These rules relate the phonemic and phonetic representation levels. These rules eliminate the redundant features from the phonetic matrix for each morpheme. In this way the speaker's burden becomes less. He has to memorize only a few sets of the distinctive features for each morpheme.

Here are more examples for the notion of distinctive and redundant (predictable) features:

(1) English vowels are oral. Therefore, the feature [nasal] is redundant for English vowels. In French, however, there are two sets of vowels: oral vowels and nasal vowels as the following examples show [^- shows nasalization]:

 

1.	Fin	/fε/	(masculine)	fine	/finə/	(feminine)	‘fine’
2.	Vain	/vε/	(masc)	vaine	/vεnə/	(fem)	‘vain’
3.	Un	/ɔ/	(masc)	une	/ünə/	(fem)	‘one’
4.	Bon	/bɔ/	(masc)	bonne	/bɔnə /	(fem)	‘good’
5.	Don	/dɔ/	(masc)(gift)	donner	/dɔnə /	(fem)	‘give’

 

(2) Say the English words ‘God, got’. Repeat them several times. You will notice that /o/ before /d/ (a voiced sound) is longer than /o/ before /t/ (which is voiceless). This difference in length is predictable in English and consequently not distinctive.

B. Redundancy Rules

You may recall, from our previous discussion, that for each morpheme:

(1) There is a phonetic, surface representation containing information of how the word is actually pronounced; and

(2) an underlying, deep, phonemic representation containing only the distinctive features.

In the above matrix for /p/ in ‘pea’ we saw that there were ten redundant features which we eliminated from the phonetic representation through rules. Rules of this sort are called redundancy rules.

Now, we will try to present some redundancy rules more formally:

(1) If: #           [-continuant, -voice]

Then:             [+aspiration]

In this rule, the boundary symbol # shows that this sound segment occurs at the beginning of the word. This rule is read: ‘If a voiceless stop (p,t,k,č) appears word initially, it has aspiration’. Here are two more examples:

(2) If: #           [-consonantal, +high, -back]

Then:             [round]

‘If a segment is a high front vowel (i, I), it is not round’.

(3) If: #           [-consonantal, +high, +back]

Then:             [+round]

‘If a segment is a high back vowel (u, U), it is +round’.

Since rules (2) and (3) deal with the relationships between the features [round] and [back], it is necessary to employ variables. So instead of the rules (2) and (3), we will have rule (4) in our grammar.

(4) If: #           [-consonantal, +high, α back]

Then:             [α round]

If we have [α] as [+], we will get rule (3); if we have [α] as [-], we will get rule (2). Phonologists will tell us that rule (4) is more economical because it employs only four items, whereas rules (2) and (3) both employ eight items. A rule with fewer features specified is a simpler rule than a rule with more features specified.

C. Phonological Rules

As we saw before, a morpheme is the smallest unit which carries meaning. In generative phonology, it is assumed that each morpheme has one base form. This underlying base form is called its systematic phonemic representation. There are rules in grammar which convert these underlying representations to phonetic representations (or surface forms).

In most cases there is no great difference between the two levels of representation for morphemes. The morpheme meaning ‘pen’ is always produced /pεn/. Its phonemic and phonetic representations are the same.

But the phonemic and phonetic representations for the plural morpheme in English are different. We will assume that the underlying deep structure or the phonemic representation for the plural morpheme has the following structure: [-voice, +strident, +anterior, +coronal].

When we wish to pronounce the plural form ‘bags’; we will carry out the following procedures:

 1) The phonemic representation:

            bæg [-voice, +strident, +anterior, +coronal] (s).

 2) A redundancy rule converts [-voice, +strident, +anterior, +coronal]

into the /s/; so we will get: [bægs]. Remember [s] is voiceless.

 3) A rule, called voicing rule, changes [-voice] segment /s/ to [+voice] segment /z/ after the [+voice] segment /g/. This rule can be presented formally as:

[-voice] → [+voice] / [+voice] [+strident, +anterior, +coronal].

This rule reads: ‘wherever you have a voiceless sound after a voiced sound, change it to the voiced sound’.

This is an example of phonological rules. It is also an example of assimilation.

 

The Function of the phonological Rules

Rules of this kind, or phonological rules, can perform the following functions:

1) Changing segments: In English /s,z,t,d/ are changed to [š,ž,č,j] respectively before /j/; that is, alveolar consonants became alveopalatal before the palatal glide /j/. This process is called palatalization. This can be shown by the following rule:

 

            [s,z,t,d] → [š,ž,č, ] /-j/

 

Examples:      We miss you                           [we mišju]     

                        They please you                     [ðeɪpližju]

                        I got you                                [əigɔčju]

           

                        Would you                             [wu ju]

                        Immediate                              [ɪmijɪət]

2) Deletion: Phonological rules can delete segments. In some /r-less/ dialects of English, word final /r/ is dropped before a consonant, but not before a vowel:

His father is here. /hɪz faðə ɪz hɪə/.

Certain English morphemes ending in a vowel, drop the vowel before a suffix beginning with a vowel: Mexico, Mexican [= Mexico + an]. The /t/ of the root {-crat} is lost before derivational {-cy}, as in ‘democracy’.

3) Insertion: Phonological rules can insert segments. Spanish inserts [e] before word-initial /s C/ sequence: (where C stands for any consonant). Thus, the phonemic representation /sculla/ becomes phonetically as [esculla].

English inserts schwa /ə/ between [+strident, +coronal]: segments: (that is, s,z,š,ž,č, ) and the plural morpheme {s} . Thus, ‘class’ /klæs/ becomes [klæsəz]; and

            Buzz               /bʌz/                becomes                      [bʌzəz]

            Dish                /dɪš/                becomes                      [dɪšəz]

            Bench             /bεnč/              becomes                      [bεnčəz]

            Judge              / ʌ /                  becomes                      [ ʌ əz]

 

This is shown formally by the following rule:

Ø → [ə] / [+strident, +coronal] [-voice, +strident, +anterior, +coronal].

This rule states: ‘Insert a schwa between [s,z, š, ž, č, ) and the plural morpheme {s}’.

English adds/g/ before inflectional {-er} and {-est} as in /lɔŋ-lɔŋər- lɔŋəst/-longer-longest. These are other examples for insertion.

4) Interchanging odr metathesis: in some occasions, phonological rules interchange segments. For example, sometimes a speaker might pronounce /æsk/ as [æks]. This rule has converted the phonemic representation /sk/ to the phonetic representation [ks].

 

References

 

1.                  Anderson, S.R. (1974). The Organization of Phonology. New York: Academic Press.

2.                  Bloomfield, L. (1933/1961). Language. New York: Holt, Rinehart and Winston.

3.                  Falk, J.S. (1978). Linguistics and Language. New York: John Wiley and Sons.

4.                  Farrokhpey, M. (1990). An Introduction to Linguistics Vol. I: English Phonetics, Phonology, and Morphology. Tehran: University of Az-Zahra.

5.                  Farrokhpey, M. (2000). Linguistics and Language. Tehran: SAMT.

 

Поступила в редакцию 02.02.2010 г.