You are on page 1of 7

SU St.

Kliment Ohridski FACULTY OF CLASSICAL AND MODERN PHILOLOGY

RESUME Subject:

The Syllable

for the subject: Phonology

Student: Mila Marinova Fac. 25738 Group 3

Teacher: Vladimir Philipov

Sofia, 2012

Phonotactics is the aspect of phonology that deals with the number and arrangement of phonemes (be it vowels or consonants) in clusters within the syllable. Therefore, the syllable constitutes the domain of phonotactics. Phonotactics establishes the permissible onset, peak and coda clusters within the syllable and, for that purpose, a certain sonority scale must be devised to serve as a governing principle when classifying syllables. It is recommendable that the most economical set of features are used in it to eliminate redundancy and to enforce binary opposition, which means that phonological features such as oral stop, fricative, etc are entirely disregarded. Thus, the main binary opposition is between [-sonarant] and [+sonorant]. The most sonorous phoneme can be discovered by using 5 different binary features: it would be [+sonorant], [+continuant], [+lateral], [+consonantal] and [+high].

Onset Phonotactics
The possible onset clusters in English are examined through the use of a (positive) template and (negative) filters. Numerous filters may be utilized to further restrict the number of possible onset clusters, but only the most important ones will be mentioned in this summary. Template: Three X-positions are assumed as the default number in an onset (the first considered as an appendix). While it is true to say that X3 needs to be more sonorous than X2, it would not be an entirely sufficient restriction where English is concerned. X2 also has to be a [-sonorant] and X3 a [+sonorant] in order for us to have a possible onset (note that in Bulgarian that is not the case we can have kn as in kniga). The feature specifications of the onset template are as follows: X1 appendix /s/ X2 [-sonorant] X3 [+sonorant] Filters: 1.Only /s/ can occur with /m/ or /n/. Therefore, no nonstrident phonemes occur with /m, n/. 2. Any cluster that is forbidden without a preceding /s/ is also forbidden with a preceding /s/. 3. In single-X onsets /j/ is free to occur as it does in yield, you, yell, yarn, etc. However, when it clusters with coronal consonants, it is completely absent in GA: /sjut/, /zjus/, /nju/ become /sut/, /zus/, /nu/. It is important to note that usually it is necessary to have a peak and a coda to form a phonological unit (onset and peak do not), but cluster-/j/ is demonstrably both part of the onset and the peak. Therefore, in this case part of the onset forms the phonological unit.

Rhyme Phonotactics
Template: X2 is associated with either peak or coda X3-6 are optional X4-6 are [-sonorant, +coronal] X1-3 decrease in sonority left-to-right Filters:

1. VR (Vowel-r) is only possible in rhotic varieties of English and not, therefore, in RP. 2. In the template it is clearly stated that the phonotactic constraints of English do not tolerate rhymes in which X3 is more sonorous than X2. That is why we can have kiln /kln/ but kennel /knl/ has two syllables since the /nl/ cluster in the latter cannot be a rhyme. 3. /h/ does not occur in the rhyme. 4. // only occurs after lax vowels, that is, in the X2 position of the template. Hang, bunk are well formed but */ba/, */bk/ are not.

Vowel and consonants and their relationship within the peak. Monosyllabicity
In stressed monosyllables, the only segments that can occur in the peak are vowels. The possible peak configurations are: 1) 1 lax vowel // - 1 X-position 2) 1 tense vowel /i/ - 2 X-positions 3) 1 diphtong /a/ - 2 X-positions Since no segments other than vowels car occur as peaks in stressed syllables, and since all vowels are [-consonantal], we can posit the constraint on such syllables that peaks must be [consonantal]. It would be important then to examine the consonants that have the feature [consonantal] as potential syllable peaks, namely /j/ and /w/. // and /j/ are treated as realizations of the same phoneme, as are // and /w/. /j/ and /w/ are always associated with onsets of syllables (yell, well, quick); // and // occur in peaks but never in onsets (bit, foot). Their distribution is therefore complementary: where /j/ and /w/ can occur, // and // cannot, and vice versa. The phonotactic constraint which excludes /j/ and /w/ from the rhyme is to some extent invalid. Given the nondistinctness of /j/ and /w/ from // and // respectively, this constraint is not quite correct: the phonemes in question can occur in the peak, in the form of the lax monophtongs // and // or as second elements of diphthongs (as in /a/, /a/ etc.). In codas, neither pair can occur. In general terms, if the X2 position is [+consonantal] then it is a coda consonant and if it is [consonantal] then it is a vowel, part of the peak Despite the phonemic nondistinctness of /j/- //, etc, different symbols are employed for the sake of a more transparent transcription (at the cost of some redundancy). Unstressed syllables may be light (only one X-position in the rhyme), which is automatically the peak of such syllables. Moreover, this X need not be [-consonantal]: the second syllables in button / btn/, /rm/, little /ltl/ may contain peaks that are [+conosantal, +sonorant]. Note that the /n/ in button is still a consonant: although a syllable peak, it does not meet the phonetic requirements by not being [-consonantal].

Polysyllabicity

The claim that phonotactic constraints operate within the syllable and not the word is relevant in polysyllabic words. Any well-formed polysyllabic word should be a sequence of wellformed syllables. That is, there can be no polysyllabic word in English that cannot be broken up into well-formed syllables. The sonority-based phonetic theory identifies peaks, slopes and troughs in the sonority profiles of words, thereby predicting the overall number of syllables for a given word as well as the analysis of these syllables into onsets, peaks and codas. Therefore, we must establish the regularities that govern the placement of syllable boundaries in polysyllabic words. Let us take a look at the word aroma: it does not tell us where, precisely, a syllable boundary is located. One could argue whether the /r/ in aroma is the coda of the first syllable or the onset of the second. Similarly, does the /m/ belong to the second syllable or the third? In this case, it is most reliable to take into account the intuition of the speaker of the language. Speakers will generally agree with the following syllabification: a.ro.ma Other examples: a) ma.ri.na a.ro.ma pho.ne.mic co.di.fy b) al.ti.tude nigh.tin.gale a.gen.da stan.dard c) a.pri.cot al.ge.bra Hum.phrey ma.tron

Nonetheless, syllable boundaries are difficult to judge by intuition, one reason being that like all nonphonemic information they do not constitute a kind of knowledge of the language that speakers ever need to draw on consciously in the construction of utterances. In the examples in a), we have simple consonant-vowel sequences: (C)VCV(C), where each C constitutes a trough in sonority and each V a peak. A single consonant between vowels is a syllable onset rather than the coda of the preceding syllable. This generalization of CV being favoured over VC syllables is not only true for English; it is a universal fact about the languages of the world. In b) we have examples of VCCV sequences, where the first C is more sonorous than the second: in /lt/ (altitude), /l/ is more sonorous than /t/; in /ng/ (nightingale), the /g/ similarly constitutes the sonority trough. And again this trough is always part of the onset rather than being the coda of the preceding syllable. Note that in Humphrey, the three-consonant cluster /mfr/ has a syllabic boundary before the /f/ - again before consonant that is less sonorous than both its neighbours. It is possible to conclude then that syllable boundaries occur immediately before the consonant that constitutes a sonority trough. In other words: syllable boundaries are placed in such a way that each onset contains as many consonants as possible and that each coda contains as few consonants as possible. Exceptions: e.nig.ma, at.las, hem.lock speakers syllabify enigma as e.nig.ma simply because /gm/ cannot be a syllable onset. On the other hand, the /tr/ in ma.tron, for instance, is a possible syllable onset.

Accounting for all the facts discussed so far, the rule for the placement of syllable boundaries can be formulated like this: Within words, syllable boundaries are placed in such a way that onsets are maximal (Syllable-Boundary Rule). Here we clearly see that the syllable is the domain of phonotactics: all the words we have discussed so far consist of well-formed syllables. But before we look into this issue more systematically, let us consider some more examples: a.pple pe.trol e.pic me.tric pe.de.stal ca.me.ra la.bra.dor A.fri.ca ma.do.nna ru.be.lla con.fe.tti in.te.gri.ti

They violate the principles in the Syllable-Boundary Rule, which forbids stressed syllables to be light. /a/ in apple cannot be a stressed syllable because it does not have a branching rhyme, nor can /pe/ in petrol and pedestal, /d/ in madonna and so on. In order to comply with the syllabic template, the stressed syllable in apple must be /ap/, that in petrol /pet/ and so forth. The problem is, then, that the /p/ in apple must belong to the first syllable due to the branching-rhyme requirement and to the second syllable due to the Syllable-Boundary Rule. However, what appears to be a problem for our analysis is, rather, quite consistent with the facts of syllabification: there are good reasons to believe that the consonants in question do indeed belong to both the preceding and the following syllable. They are syllabified ambiguously (with both syllables); in short, they are ambisyllabic. In what contexts, then, are consonants ambisyllabic? They are ambisyllabic only where the principles of syllabification require them to belong both to the preceding and the following syllable. This is the case where the Syllable-Boundary Rule requires them to be part of an onset and the preceding syllable is stressed and would, without such a consonant, have a single-X rhyme. Therefore, a consonant is ambisyllabic if it is a part of a permissible onset cluster and if it immediately follows a stressed lax vowel. This brings us back to the claim that the syllable is the domain of phonotactic constraints and that all English words consist of strings of well-formed syllables and that no other words can exist. For instance, most people would agree that *umbna and *atctic are impossible words in English. They are impossible because they cannot be syllabified by means of the mechanisms that we have deplayed in this chapter. */mbl / is an impossible syllable and so is */bn/, leaving the /b/ stranded between two syllables, unable to go with either. In the latter case, */atk/ and */ktk/ are impossible syllables, so that, again, the /k/ cannot be associated with either the first or the second syllable. Note that /mbr/ would be a possible English word: unlike in */mbln/, the /b/ can here be part of the onset: */bn/ is an impossible onset while /br/ is possible. Similarly, a change from atctic to arctic produces a possible word, simply because the problematic /k/ can now be part of the first rhyme: /rk/ is a possible coda (in rhotic accents) while */tk/ is not. The following, then, is a condition on the structure of well-formed English words: Bring is a well-formed word because /br/ is a well-formed syllable onset and /i/ a well-formed rhyme. *Bning is impossible because */bn/ cannot be a syllable onset. In polysyllabic words, any consonant must be syllabifiable with either the rhyme of the preceding syllable or the onset of the next. All strings of phonemes that cannot be syllabified belong in the category of impossible words.

Segments and X-positions


Phonological representations do not merely consist of segments (that is, bundles of features), arranged in sequence like beads on a string, but that such representations are further structured. Strings of feature bundles make up but one tier of our representations: on another such tier, we have X-positions (which are again arranged in sequence, and associated with the elements of the fuature tier); and on yet another tier we have rather more complex structures in which X-positions are grouped into higher-order units. What is important for our understanding of these three tiers is primarily the way in which they are associated with one another. What are the principles governing the association of feature bundles with X-positions, and what are those that control the way in which syllables are erected on top of the X-tier? The only statement that we have made regarding the association of feature bundles and Xpositions up to this point has been the Vowel-Length Rule: 1) Associate a [-tense] vowel with one X-position. 2) Associate each element of a diphthong with one X-position. 3) Associate a [+tense] vowel with two X-positions. The details ho not need concern us again, nor does the fat that SSE is not covered by this rule. Two things are worth noting here: firstly, the correspondence between feature bundles and Xpositions is not necessarily one-to-one; and secondly, in the approach chosen here, the number of X-positions associated with any given vowel is entirely determined by the segmental features of the vowel. The X-tier contains no information that is not also present if encoded in a different way in the feature tier. It confirms our assumption that in the vowel system of our reference accents any differences in length are redundant: predictable from the differences in quality (for example, tenseness) that we treat as phonemic. Phonemic differences among vowels are expressed on the feature tier rather than on the X-tier. This is not to say, of course, that the X-tier cannot contain any phonemic information: all we have said is that in the case of these English vowel systems, it does not. Indeed, an alternative approach to English phonology might well suggest that the English vowel system (more precisely, the RP and GA ones) is bifurcated in such a way that the members of one subset are phonemically associated with XX and the members of the other with X, and that the tense/lax specification of any given vowel follows from its properties on the X-tier; an XX vowel would then be redundantly [+tense] and an X vowel [-tense]. In terms of phonemic transcription, this would imply that pairs of vowels should be transcribed as, for example, /i/ vs /i/ (so that the difference in length, here accounted for as XX vs X, is the underlying one). In affricates we see for the first time the association of two elements of the segmential tier with one X. Affricates are, in phonetic terms, stop-plus-fricative sequences which share the same voicing specification and the same place of articulation. In phonological terms, however, they function as single units primarily in phonotactic terms: for example, the onset template does not permit sequences of two [-sonorant] segments; yet /t/ and /d/ - the only affricates in English do occur in onsets: chin, gin. Invoking the X-tier, we can resolve this apparent contradiction: affricates consist of two feature bundles associated with a single Xposition.

Let us turn now, briefly, to the association of X-positions with syllable structures. Little needs to be said about this association, which is relatively straightforward. The elements of the Xtier constitute the bottom elements of the tree structures that represent syllables: X-positions are grouped into onsets, peaks and codas; and the latter two are in turn grouped into rhymes, which, together with onsets, form syllables. The X-tier, then, simply marks the number of slots that any given segment occupies in a syllable. However, we have come across two separate cases where an X-position was analyzed as being part of two higher-order units at the same time. Sample analyses are repeated in view and petrol. The /j/ in view forms part of both onset and peak. And in petrol, the /t/ belongs to both syllables: it constitutes the coda of the first syllable, thus making the syllable heavy, while at the same time being part of the onset of the second syllable. Note that this analysis does not imply that the /t/ in petrol is long: it is represented by one X-position only unlike the (hypothetical) long one in (/b/ - associated with two X-positions). For the kind of dual association of X-positions with syllable structures we use ambisyllabicity.

You might also like