Vowels and Syllables
Recall that “phonetic” refers (roughly) to measurable, un-abstracted data, while “phonologic” refers to the abstract structure of such data (as defined by their value in a given language).
In this chapter we will consider the kinds of vowel sounds that occur in the world’s languages. But before we do this we should try to define what we mean by a vowel. In many linguistic descriptions sounds are classified as either vowels or consonants. The original intuition behind this classification was that vowels are sounds that may be pronounced alone, but the sounds called consonants must be accompanied by a vowel. The phonetic basis of the distinction between vowels and consonants is not straightforward. An important contribution on this topic was made by Pike (1943) who began by splitting segments in another way. He first of all made a distinction between vocoids and contoids, with a vocoid being defined as a central resonant oral. He then went on to define a vowel as a syllabic vocoid. In practice this is very similar to the definition given by Chomsky and Halle (1968) in the latter part of The Sound Pattern of English. Their definition is that a vowel is a segment with the features [+syllabic,−consonantal], with [−consonantal] sounds beind defined as those that do not have a central obstruction of the oral tract. In many ways this is functionally equivalent to the later practice of autosegmental phonologists in defining a vowel as a [−consonantal] segment attached to a V slot. Whichever definition is used it is equivalent to saying that a vowel is defined by features that ensure that there are no major strictures in the vocal tract; and that it is syllabic.
We know what we mean by there being no obstructions in the vocal tract, but what, from a phonetic point of view, do we mean by syllabic? There is no phonetic parameter that can be used to define syllabicity in articulatory, or physiological terms. When Pike proposed his definition, he suggested that syllables correspond to the valleys between peaks in strictural degree. Thus the English word pit forms a single syllable because there is a more open degree of stricture between two regions of greater stricture. [cf. sonority hierarchy.] However, the word split on this definition would have two syllables, as s has lower stricture than p. This conflicts with the intuitive sense of what constitutes a syllable, and with the behavior of s in other positions. An earlier proposal was that of Stetson (1951), who claimed that each syllable is associated with the particular kind of respiratory activity that Stetson called a ‘chest pulse’. We now know that syllables are not necessarily associated with a chest pulse (Ladefoged 1967), but phoneticians have not been able to suggest an alternative definition of the physiological properties of a syllable. The best that we can do is to suggest that syllables are ‘necessary units in the organization and production of utterances’ (Ladefoged 1982). This is a neurophysiological, or cognitive view of the syllable, making the syllable a phonological rather than a phonetic unit. Syllables are identifiable as the primary elements over which the rhythmic patterns of language can be observed, or the primary domain over which sequential constraints apply, or coarticulatory adjustments can be made. Vowels are defined by the physiological characteristic of their having no obstruction in the vocal tract, and by their function within a phonologically defined syllable. At the end of the chapter we will consdier semivowels, which we will take to be sounds that are like vowels in that they have no obstructions in the vocal tract, but unlike vowels in that they are not syllabic.
—Ladefoged & Maddieson’s The Sounds of the World’s Languages is full of parts like this, where a single precise paragraph is enough to clarify my doubts.
A summary of the section on vowel features:
- The traditional dimensions of Backness (front/back) and Height (high/low) were originally thought to be defined by the position of the highest point of the tongue, or the “mean position of the tongue in the saggital plane”; but modern study with x-ray movies shows that they’re not at all correlated.
- Stevens & House (1995) and Fant (1960) proposed this definition: The position of the point of maximum constriction of the vocal tract, plus the cross-sectional area of the vocal tract at that point. This gets closer, but still doesn’t correspond to the traditional vowel map.
- The two characterizations above can be combined: The dimensions would be measured by the distance of the highest point of the tongue from the roof of the mouth—notice the height of the soft palate changes with the “height” of vowels. This starts to look like the traditional Backness/Height maps, but even then, it’s not yet a complete correspondence. (E.g. [ɔ] would be much “lower” than [æ], and [ɪ] more “back” than [i]).
- A very close correlation can be found with acoustic data, more than with any articulatory mechanism proposed so far. A plot is suggested based on the first two formants: on one coordinate, F1, and on the other, F2−F1 (and the scale on F1 is doubled, so as to represent the human judgement/perception). The results are pretty much the same as the standard vowel map (notorious difference: [ɔ] significantly lower than [ɛ], but still higher than [æ]—which is at the same height as [ɑ]). In particular, the cardinal vowels end up in the expected corners.
- So Height and Backness can be derived from formants, but formants represent more data than this (e.g. roundness, nasality, ATR…). In other words, vowels with “the same” Height and Backness can be distinguished in formant values. So articulatory parameters are relevant after all, if in a secondary way.
See pages 283–286 for useful plots and diagrams.