Chapter IV

Acoustic D.F.’s and the
New Postulates

4.1 Chomsky and Halle’s inventory of D.F.’s grew out of an earlier hypothesis which was first illustrated in Jakobson, Fant, and Halle’s Preliminaries to Speech Analysis (hereafter PSA). Generativists discarded these primes, which are based on acoustic data, as the emphasis of linguistic study shifted from perception of language to its production and, in this case, the production of sound in language. J&W recently (1979) defended the acoustic postulates and steadfastly insisted that communication is achieved by sound and not by its production, the study of which they regard as (83) “. . . a crude metrical attitude [opposed to their] . . . sane, relational, topological treatment.” They maintain that sufficiently distinctive spectral configurations of the sounds of a language can be produced using articulatory gestures that are quite different from the ones linguists have considered to be standard for a particular language. They go so far as to assert (96-7) that a tongue is not even necessary to produce intelligible speech. Certainly J&W cannot be claiming that tongueless speech, regardless of its intelligibility, which certainly must depend on both the good will and close attention that listeners would have when trying to understand the speech of a tongueless person, sounds anything like normal speech. Regardless of the preferability of acoustic primes over articulatory ones, both the JFH inventory and the one I have proposed are theories of the speech sounds of language; both have a distinct generative capacity and can be compared vis a vis their power as a generative mechanism, their adequacy as descriptive tools, and their simplicity as a theory of human speech—which is the object of this chapter.

4.2.0 The features JFH proposed in PSA are [1 vocalic, 2 consonantal, 3 interrupted, 4 checked, 5 strident, 6 voiced, 7 compact, 8 grave, 9 flat, 10 sharp, 11 tense, 12 nasal]. They contrast with their negations i.e. [1 -vocalic, 2 -consonantal, 4 -checked, 6 -voiced] or with other features: [3 interrupted-continuant, 7 compact-diffuse, 9 flat-plain, 10 sharp-plain, 11 tense-lax, 12 nasal-oral]. For the purposes of this paper, I shall consider the original 12 to contrast with their negations. In the following paragraphs I present the definitions of the acoustic primes.

4.2.1 [vocalic] “Vowels have no obstructive barrier along the median line of the mouth cavity ... (PSA, 19) Phonemes possessing the vocalic feature have a single periodic (‘voice’) source whose onset is not abrupt ... the vocalic formants have small dampening ... [expressed] in the narrow bandwidth of the formants ... the lower formants have greater intensity.” (PSA, 18-19). In PSA, the authors did not consider the possibility of the phonemic status of surd vowels which contrast with voiced vowels in languages such as Comanche, Ute, Mayan Chontal, Gulla, Teso, and Bagirmi (J&W, 135). This admission obliges J&W to resort to the syntagmatic definition of vowels as (85) “... the most usual carriers of syllable nuclei ...”

4.2.2 [consonantal] “... consonants have a barrier sufficient to produce either complete occlusion or a turbulent noise sound. (PSA, 19-20) Phonemes possessing the consonantal feature are acoustically characterized by the presence of zeros that affect the entire spectrum.” (19) This appears to mean that the sound spectra of consonants are less evenly textured than those of vowels. The authors, (JFH, 19) and (J&W passim) claim that the class of sounds known as liquids, r’s and laterals, are both vocalic and consonantal although their intensity is lower than that of the vowels. Glides (h and ?), they claim, are neither vocalic nor consonantal. They lack the harmonic source of vowels and zeros in the spectra. Sounds commonly represented as y and w are not considered glides, but non-syllabic vowels, (PSA, 20).

4.2.3 [Interrupted] “The abrupt onset distinguishes the interrupted consonants (stops) from the continuant consonants (constrictives) . .. Stops . . . [have] a sharp wave front preceded by a period of complete silence, . . .” (21). JFH consider liquids to be of 2 sorts: [continuant] such as l, t, etc. or [interrupted] such as the flap. They consider the English /r/ to be a type of schwa (22).

4.2.4 [Checked] “In spectrograms . . . checked phonemes are marked by a sharper termination . . . The air stream is checked by the compression or closure of the glottis.”

4.2.4 [Strident] “Sounds that have irregular waveforms are called strident . . . oscillograms show a distinctly higher periodicity and uniformity in mellow constrictives [i.e., -interrupted] such as /Ө/ in comparison with /s/ and other strident constrictives. Strident phonemes are primarily characterized by a noise which is due to turbulence at the point of articulation ... a supplementary barrier that offers greater resistence to the air stream is necessary ... [for the production of] stridents.” (23-4)

4.2.5 [Voiced] “The most striking manifestation of ‘voicing’ is the appearance of a strong low component which is represented by the voice bar along the base of the spectrogram .. . Voiced phonemes are emitted with constant periodic vibrations of the vocal bands . . .”(26)

4.2.6 [Compact] “Compact phonemes are characterized by the relative predominance of one centrally located formant region (or formant). They are opposed to diffuse phonemes in which one or more non-central formants or formant regions predominate ... consonants articulated against the hard or soft palate (velars and palatals) are more compact than the consonants articulated in the front part of the mouth ... open vowels are the most compact, while close vowels are the most diffuse.” (27)

4.2.7 [Grave] “... this feature means the predominance of one side of the significant part of the spectrum over the other. When the lower side of the spectrum predominates, the phoneme is labeled grave; when the upper side predominates, we term the phoneme acute ... gravity characterizes labial consonants as against dentals, as well as velars vs. palatals and, similarly, back vowels articulated with a retraction of the tongue vs. front vowels with advanced tongue.” (30).

4.2.8 [Flat] “Flattening manifests itself by a downward shift of a set of formants or even of all the formants in the spectrum . . . Flattening is chiefly generated by a reduction of the lip orifice (rounding) . .. Instead of the front orifice of the mouth cavity, the pharyngeal tract, in its turn, may be contracted with a similar effect of flattening.” (31)

4.2.9 [Sharp] “This feature manifests itself in a slight rise of the second formant and, to some degree, also of the higher formants ... the oral cavity is reduced by raising a part of the tongue against the palate.” (31)

4.2.10 [Tense] “... tense phonemes display a longer sound interval and a larger energy ... tense phonemes are articulated with greater distinctness and pressure than the corresponding lax phonemes.” (37-8)

4.2.11 [Nasal] “The spectrogram of nasal phonemes shows a higher density than that of the corresponding oral phonemes . . . between the first and second vowel formants there appears in the nasal vowels an additional formant with concommitant weakening in the intensity of the former two . . . the nasal consonants add a ... murmur . . . [which] possesses two constant and clear formants, one at about 300 cps. and the other at about 2500 cps.” (39-40)

4.3.1 Contrasting the JFH features with the ones advocated here (see Appendix C) establishes the following as equivalents:

JFH Brakel

The features [DT 1°, DT 2°, suction, glottalic, velaric] are not commensurate with anything proposed by JFH. DT 1° and 2° are prosodic features, features that are superimposed on the properties of a sound, either from an articulatory or acoustic point of view, when the sound is analyzed as being a long or complex segment instead of a sequence of segments. Suction, glottalic, and velaric are air stream features not taken into consideration by JFH, but certainly important in the characterization of distinctive sound used in the languages of the world. The remaining features to be contrasted are:

JFH Brakel

That is, where commensurate, my system has two more features than the acoustic set, and either my system can be trimmed by two, or the JFH system is lacking in two.

4.3.2 JFH proposed to distinguish laterals and r sounds (liquids) in the following manner:

There are several contradictions in these specifications. 1) Notwithstanding the claim of (PSA, 19) that laterals and r sounds have both the consonantal and vocalic features, the major means of distinguishing between them is to use the feature [vocalic] to identify laterals as opposed to r sounds and the other consonants which are [-vocalic]. JFH also assert (26) that despite the high dampening of the formants of fricative laterals and r’s, both retain acoustic traits of liquids, they are liquids with superimposed stridency. Their specification is inconsistent with this statement. 2) JFH proposed to distinguish /r/ and /1/ (the flap and the alveolar lateral) by the feature [interrupted]. In a language such as Portuguese /r/ and /1/ contrast with /d/ intervocalically: mora ‘he dwells’, mola ‘spring’, moda ‘style’ and with /i/ and /u/ post-consonantally, prazer ‘pleasure’, placa ‘license plate’, piano ‘piano’, pueril ‘puerile’. The contrast of these segments can only be maintained by the following specifications in JFH D.F.’s.

I have placed a circle around the + for [vocalic] as a specification for /r/. In order to contrast /r/ with /d/, /r/ must be specified as such, but such a specification [+interrupted, +vocalic] makes it a peculiar sound indeed, given that Jakobson considers vocalic sounds to be syllable nuclei. If one were to specify /r/ as [-interrupted], on the other hand, it would no longer contrast with /1/. The flap /r/ should be classified as I have recommended, as a contoid which is negatively specified for every distinctive parameter other than [raised], making it the least marked of contoids. This leaves [vocalic] as the sole distinctive feature of laterals according to the JFH approach, since they are both consonantal and vocalic. I suggest, then, that [-lateral] be accepted as an apt substitute for [vocalic]. All non-contoids will be [ lateral] whereas a subset of the contoids, the laterals, will receieve a positive mark for laterality, their unique trait. This has not reduced the number of different features, it has just replaced one for another.

4.3.3 The PSA features [grave], [flat], and [compact] (along with [sharp] which corresponds to my [palatal]) reflect different points of articulation as well as different configurations of sound spectra. They correspond to my features [labial], [dorsal], [high], and [raised] and distinguish a total of 16 apertures and points of articulation as opposed to the 29 distinguished by my system.¹ A comparison of how the two systems characterize vocoids will be revealing.

Brakel System—Vocoid Distinction

PSA System—Vocoid distinction

The problem with the PSA specifications is that they do not contrast the high vowels i, u with e, o nor the high lax vocoids, I, U with E, O—unless one considers the latter to be [+compact], i.e. low. In addition, they do not contrast the central vocoids, ɨ, ə , and â from one another in any convincing manner. About the lack of a mark to distinguish e from i, etc., J&W say the following (131): “It must be remembered . .. that . . . the geometric mean /e/ ... is noncompact in relation to the compact /ä/ and nondiffuse in relation to the diffuse /i/ ...” Earlier, they have claimed “.. . we are, actually, clearly faced in these cases with the bifurcation of the binary opposition diffuse ~ compact and nondiffuse ~ diffuse.” They have, in other words, either raised diffuse to the level of distinctive feature or they have abrogated the strictly binary nature of their feature system. In order to maintain strict binarity, it is necessary to add another prime to their inventory in order to distinguish high and mid vowels. Their alternative is to abrogate the principle of binarity which they adamantly insist upon as correct and the only way to describe communication (cf. J&W 23-4, 81-2, 146, 173-4). This makes the two systems the equivalent of one as far as their generative capacity of point of articulation and degree of aperture is concerned.

4.3.4 The PSA features contain the feature [strident] which is used to contrast sounds such as θ from s, x from X, the former being [-strident] (mellow) and the later [+strident]. Unfortunately, PSA has no means of distinguishing contrasts between and f or and v the latter of which occurs in Continental Portuguese: Cabo [kábu] ‘handle’ cavo [kávu] ‘I dig’;² since both &and vare non-strident. Other languages that maintain contrasts between &and v are Béni (Wescott, 1965), Mbembe (Barnwell, 1969), Etskato (Laver, 1969) and Ewe (Berry, 1951). I have distinguished these phonemes by considering f, v, Ө, , š and x to be slit fricatives and , s, s, and X to be non-slit. That is, the fissure at the point of articulation is not preponderantly horizontal. The // of some languages may be an approximant, i.e. a labial contoid which may have audible local friction only when there is no laryngeal impedence. In this case it would be specified as [+contoid, +labial, -friction], in the case that there was audible friction when voiced, it would be positively specified for [fricative] but not for [slit]. It may be concluded that there is a need to distinguish, within the contoids known as fricatives, two subsets, one characterized by slit articulation, the other not. Non-slit fricative articulation within the oral cavity produces the stridency noted by JFH etc., whereas labial fricative articulation produces a mellow fricative, regardless of whether the articulation is labio-dental [slit] or bilabial [-slit].

4.4 It has been shown that in order for the PSA features to distinguish contrasts linguists know exist it is necessary to add two features, [raised] and [slit], to the inventory. This makes the system JFH propose the equivalent of the one proposed here. Fudge (1967, 4) has said “It is thus dangerous and misleading to say that either articulatory or auditory feature ARE the phonological elements, unless they correlate so closely that no facts of language are obscured by treating them as if they were the same: . . .’’ Concerning the powers of specification of the revised acoustic primes and those I suggest, they appear to be near equals. A further test of their equivalence will come in the following chapter, where strength hierarchies of segmental types will be contrasted.

Footnotes

¹The articulatory primes that I have proposed, [palatal, labial, dorsal, raised, high] do not have a generative power of 2⁵ or 32 points of articulation because no segment can be [+high, -raised], so three units must be subtracted giving it a generative power of 29 points of articulation.

²A phonological analysis of some dialects of Portuguese would identify [] as an allophone of /b/. In some dialects, however, the non-occlusives [, , ] are preponderant—they only occur as occlusives after a homorganic nasal. In other words, these phonemes may be identified as fricatives rather than occlusives since their occlusive realization is definitely a minority.