An acoustc study of coarticulation in the speech of greek adults with normal hearing and hearing imprairment

Abstract

Research has shown that speech acquired in profound hearing loss presents differences on both the segmental and the suprasegmental levels compared with normal hearing speech. Recent studies focus on the dynamic aspects of hearing impaired speech, i.e., coarticulation as coproduction of gestures, but findings are variable. Although this issue has received a lot of attention in the English literature, phonetic research in Greek is still scant, this being the first study of coarticulation inGreek hearing impaired speech.The main aim of the present thesis is the acoustic exploration of (a) vowel-tovowel and consonant-to-vowel coarticulation in degree and/or temporal extent and (b) static characteristics such as vowel space, distribution and duration of the three point vowels, in the speech of Greek young adult male and female speakers with normal hearing (NH) and hearing impairment (HI). The aforementioned dynamic and static acoustic properties are investigated in relation to certain variables, i.e., vocalic and consonantal context, stress, syllable position, as well as speaker gender and intelligibility. The speech of nine subjects with profound HI, five male and four female, and five subjects with NH, two male and three female, was analyzed acoustically by measuring formant frequencies F1 and F2 at vowel onset, midpoint and offset of the Greek point vowels [i, a, u] in disyllables of the form [pV 1 CV 2 ] with consonants [p, t, s] stressed on the first or the second vowel. An additional experiment was conducted in order to rate the intelligibility of the speakers with HI. Three groups, i.e., very high, high and medium intelligibility, were formed on the basis of judgements made by 60 naïve listeners with NH who rated 101 words and 25 short sentences produced by the speakers with HI. This has been a long journey. Looking back, I have many people to thank for their valuable contribution to the completion of this research.

Intelligibility enhancement of casual speech based on clear speech properties

“…The "header sentences" are not evaluated. The scoring system is based on previous research on English intelligibility tests (Monsen, 1978(Monsen, , 1982Picheny et al, 1985a), supported also by researchers for Greek language (Sfakianaki, 2012). Each word is considered incorrect even if there is a mismatch in one phoneme e.g "fíki" instead of "thíki" ("seaweed" instead of "case").…”