Segmental Contributions to Speech Intelligibility in Nonconcatenative vs. Concatenative Languages

Abstract

This study investigated the contributions of segments (consonants vs. vowels) to speech intelligibility in Arabic and English. In these two languages, consonants and vowels play crucially different grammatical roles. Arabic is a nonconcatenative language that primarily assigns lexical information to consonants and morphosyntactic information to vowels, while English is a concatenative language that does not assign distinct roles to either class of segments. On this basis, we hypothesized that consonants and vowels would play very different roles in the intelligibility of the two languages. Five laboratory experiments were conducted, three on Arabic and two on English. Participants listened to words and sentences in which either all consonants or all vowels were replaced with silence and were asked to indicate what they heard. Unlike previous studies, all stimuli were carefully controlled for ratio of consonants to vowels. Results showed that in Arabic, consonants made a greater contribution than vowels to speech intelligibility, both in isolated words and in complete sentences. Furthermore, in consonant-only conditions, stimuli containing more consonants were more intelligible than those containing more vowels, displaying a clear effect of segmental ratio. In English, by contrast, the consonants and vowels made roughly equivalent contributions to speech intelligibility, and segmental ratio played a negligible role. These two disparate findings suggest that segmental contributions are crucially modulated by language-specific factors. That is, the different contributions of consonants and vowels to speech intelligibility are not solely determined by their distinct acoustic cues, but also by the grammatical role they play.