Abstract
This paper examines rhythm speech metrics in Modern Standard Arabic using the ALGASD and West Point corpora. While ALGASD includes only native (L1) speakers, West Point includes recordings of (L1) and non-native (L2) speakers. To reveal differences in rhythm metrics, measures are calculated for L1 speakers from both corpora. The rhythm metrics computed for each sentence are three interval measures (V, Delta V, and Delta C), two time-normalized indices (VarcoV, VarcoC), two pairwise variability indices (nPVI- V, rPVI- C) and compensation and control indices (CCI- V, CCI- C). A second experiment assessed patterns in rhythm metric properties of L2 speech. Comparisons are made between L1 speakers and L2 speakers who have English as their first language. The effect of gender on rhythm metrics is also tested for both L1 and L2 Arabic speakers. A comparison of vowel lengthening between L1 and L2 speech was also investigated. Results show significant differences between native and non-native in terms of speech rhythm either in consonants and in vowels durations.