15th International Congress of Phonetic Sciences (ICPhS-15)

Barcelona, Spain
August 3-9, 2003

[b]-[d]-[g] as a Universal Triangle as Acoustically Optimal as [i]-[a]-[u]

Christian Abry

ICP-CNRS, France

In the world language databases, [b]-[d]-[g] are the most prevalent places for consonants, as are the point vowels [i]-[a]-[u]. Jakobson, Fant & Halle conceived acoustically these places in parallel with the vowels, as a triangular binary representation, until Chomsky and Halle switched to articulatory features. Lindblom's and our endeavor rests on an acoustic space for vowel systems computational prediction. At ICPhS 79, Ohala put forward the false prediction of dispersion theory for a putative 7-consonants systems: ɗ, k', ts, ɬ, m, r, |. But in search of consonant-place-space (CPS), we have to leave aside manner, as it is done in predicting oral but not nasal vowel space together. This means that the proposed ɗ, k', ts, ɬ, m, r, | inventory is reduced to ubiquitous labial, coronal and velar places. It is noticeable that Haskins' patterns could have been represented as a triangle of CV transitions - typically toward [a] - in a F2-F3 plane, where, from the "hub" locus of [g] (F2=F3), F3 is rising while F2 is falling, F2-F3 rising both for [b], and falling both for [d]. What could then be the differentiation process, the genesis of the workspaces for consonants and vowels in this common framework of acoustic coordinates? In canonical babbling, F1 is the audio movement corresponding to the carrier of speech, the mandible, with labial [bababa] or coronal [dadada] "frames". When independence of carried articulators (lip and tongue) from the jaw becomes settled, F2-F3 stream carries the information on places of contact. Thus F2-F3 plane is basically orthogonal to F1. When coarticulation emerges, after one year, the [a] vowel can now be produced during the closure phase, before opening, i.e. vocalic F1 is coproduced. Up to 4 years the mastering of the control of the whole vocal tract for [i] and [u] vowels will be in progress, if, like almost all languages, the mother tongue gets them. The differentiation process is thus comparable within the two streams finally overlapping: F1-F2 mainly for universal [i]-[a]-[u] vowels, and F2-F3 for universal [b]-[d]-[g] consonants. So the Dispersion Theory proves efficient for structuring both acoustic planes: this is an answer to the challenge issued by Ohala, an answer which does not call ab initio to his joined proposal of compositionality (Maximum Use of Available Features). Additional third dimensions may be used, F3 for vowels (e.g. to contrast [y] vs. [i] in French), or F1 for consonants (contrasting pharyngealized vs. plain segments in Arabic). In these cases our Dispersion-Focalization Theory is better at issue.

Full Paper

Bibliographic reference.  Abry, Christian (2003): "[b]-[d]-[g] as a universal triangle as acoustically optimal as [i]-[a]-[u]", In ICPhS-15, 727-730.