Prediction of vowel systems using a deductive approach

Abstract

A deductive approach is developed to predict vocalic systems. First, vowels are proposed from an efficient and simple use of an acoustic tube. Then, a maximum acoustic dispersion criterion is applied to classify the obtained vowels.

Explaining vowel systems: dispersion theory vs natural selection

“…As has been noted in the literature, Quantal Theory (1.i) makes several incorrect predictions. First of all, the fourth most common vowel, [ε], is not stable as predicted by the theory (Carré 1996). Second, Livijn's (2000) survey of 28 differently-sized inventories showed no evidence for acoustically-favored hotspots, contrary to the core prediction of Quantal Theory.…”

Brownian dynamics for the vowel sounds of human language

“…The interactions between particles are modeled mathematically as physical forces, but are derived based on our understanding of the language-learning process. Our work builds on an already substantial collection of models of vowel system evolution [13,14,[18][19][20][21][22][23][24]27,30], so it is important to make clear what we are adding. Because the mechanisms which control the evolution of sound inventories are not fully understood, existing models have invoked a considerable range of mechanisms in order to explain observations.…”

“…Vowel systems, like most elements of languages, evolve over time and may therefore be viewed as dynamical systems coupled to human social dynamics, and also to geography and social networks [15][16][17]. Cross-linguistic similarities suggest that their internal dynamics may play a particularly powerful role, and numerous models have been proposed [13,14,[18][19][20][21][22][23][24][25]. In the early work of Liljencrants and Lindblom [13], vowels were modeled as electrical charges, based on the principle of maximal contrast [26], yielding a single idealized vowel system for each cardinality (number of vowels).…”

“…Focalization theory, which adds an attractive interaction to vowel dynamics, hypothesizes the convergence of formants [18,27], increasing acoustic salience and "perceptual value." Other models involve iterative construction driven by contrast maximization [20], or agent-based imitation games between speech synthesizers [14,21]. One inspiration for our work is exemplar dynamics [22][23][24][25]28,29], where agents store in memory a large number of exemplars for each sound, organized into sound categories.…”

Empirical approaches for investigating the origins of structure in speech

“…The constructions produced by their model resemble natural syllable sets regarding places of articulation, vowel contrasts and phonemic coding (syllable starting and ending points can be shared between different syllables). Carré (1996) also model the emergence of vowel systems as the result of efficient use of an articulatory space. They apply the rule of maximum acoustic contrast to select vowel systems of different sizes.…”