Phonotactic constraints as a filter of the articulatory-acoustic signal

Abstract

The present study investigates the working and status of phonotactic constraints in the modeling of speech production and speech perception. Phonotactic constraints act as a filter of the acoustic signal and the subsequent articulatory gestures. The adaptation of foreign borrowings and foreign language learning provide many examples of the transfer of the phonotactic constraints of L 1 to the perception and production of L 2, e.g., English speakers' pronunciation [“nou” sei] for the Spanish sequence [“no”se] “I don't know,” and [ɛl ei “trei⃗” bɛl] for the French sequence [ɛl ɛ “trɛ” bɛl] “she is very beautiful,” shows how the English syllable structure constraint that prevents short vowels from occurring in stressed open syllables is imposed on these languages. When this syllable structure constraint is not present, though, as in the words [ɛl] “elle,” [bɛl] “belle,” English speakers pronounce a short vowel. To check whether higher-order phonotactic constraints act as a filter of the acoustic signal, the same physical signals were placed in different syllabic structures by digital manipulation with an ILS program. The stimuli were presented to speakers of different languages to check whether they perceived the same stimuli differently depending on the phonotactic constraints of their language. The results show that phonetic and phonological identification is overridden by phonotactic constraints, which advocates for a top-down processing of the signal.