ICPhS XVII Regular Session Hong Kong, 17-21 August 2011 1458 INTERACTION OF VARIABLES IN THE CIVILI VOWEL DURATION Hugues Steve Ndinga-Koumba-Binza Centre for Text Technology, North-West University, Potchefstroom, South Africa 22602569@nwu.ac.za ABSTRACT This paper investigates the interaction between phonetic environments and syntactical positions in the determination of vowel duration in Civili. The paper will mainly report on the statistical analysis that followed a measurement procedure of vowel duration from a specific set of speech data. Keywords: Civili, vowel duration, variables interaction 1. INTRODUCTION The nature of vowel duration is an issue in the phonological system of Civili (cf. [9, 10]), a developing Bantu language spoken in Gabon and neighboring countries, although a number of studies have established the existence of short and long vowels on the basis of minimal pairs occurring in the language [2, 4, 5, 7, 8]. Table 1 below gives the set of Civili vowels in their respective phonological representations. Table 1: Civili Vowels: phonological representations. Front Central Back Sh Lg Sh Lg Sh Lg 1 st degree /i/ /i:/ /u/ /u:/ 2 nd degree /e/ /e:/ /o/ /o:/ 3 rd degree /a/ /a:/ However, the determination of the vowel duration characteristics was mostly based on unverified data with impressionistic analytic procedures rising out the complexity of the phenomenon [9]. Moreover, the analysis in [10] failed to investigate thoroughly the interaction between the variables that include phonetic environments (or contexts) as well as three syntactical positions (or positions) of the word that contains the concerned vowel. This paper addresses the interaction of these variables at the segmental level. It is herewith hypothesized that the interaction between these variables has definite influence on the Civili vowel duration. This paper adopts a common method of an analysis comprising a measurement procedure of natural duration and a statistical analysis of measurement results. The paper will mainly report on the statistical analysis. 2. VARIABLES: CONTEXTS AND POSITIONS In this paper, the term “context” is mostly used to refer to the phonetic environment and the term “position” to the syntactic position in the sentence or phrase. This study is particularly concerned with the following factors: (i) The position of the word to which the studied vowel belongs, i.e., a. the word in isolation b. the word as object in a phrase or sentence c. the word as subject in a phrase or sentence (ii) The context or environment of the vowel, i.e. the immediate or direct phonological context consisting of preceding adjacent segments and/or following adjacent segments. It should be noted that at the syntactical level, Civili is a Subject-Verb-Object language with the specifications known for many Bantu languages (cf. [2, 7]). According to certain views of the phonology of Civili, variation in duration could occur in the following environments: (i) /C_NC/ (a vowel is long when it is followed by a nasal cluster). (ii) /CG_C/ (a vowel is long when it follows a consonant-glide sequence). In fact, a number of phonological descriptions and analyses (cf. [1, 3, 6, 13] to name a few) have stated that any vowel is lengthened or is long when (i) it precedes a nasal segment, leading to the following formalism: (1) V [+long] / __N (ii) it follows a consonant-glide segment, thus leading to the following formalism: (2) V [+long] / CG__ These claims apply to Civili, as could be seen from most of the previous studies by various