We have witnessed a growing number of investigations into the acquisition process in a multilingual context, which has become recognized as an independent field, quantitatively and qualitatively different from second language acquisition. Scholars have started to differentiate between learners/speakers on the basis of the complexity of their linguistic background, with the numer of known languages being an additional variable. A growing body of studies into the acquisition of third language phonology demonstrates an inherent complexity of the field reflected, among others, in multidirectional dynamic cross-linguistic influence. As shown, multilingual learners have at their disposal a broadened phonetic repertoire, a raised level of metalinguistic awareness and enhanced perceptual sensitivity, which may facilitate the learning of subsequent phonological systems. Thus, this chapter aims to compare bilingual and trilingual phonetics and phonology by providing an overview of recent research into both subdomains, identifying their common features and, importantly, their points of departure for L3 phonology, with a view to providing new insights into the acquisition of speech.

Introduction

Differences in pronunciation between different regional, national and social accents of languages are immediately striking even to non-linguists. While a small number of these differences are purely incidental, for example the fact that the second vowel in tomato is pronounced /a/ in Southern Standard British English but /eɪ/ in Standard American English, most differences between accents of any language are systematic. This systematic variation can be analysed with reference to a large number of phonological domains and structures and different phonological and phonetic processes. On the segmental level (see Gut, Chapter 13, this volume, for variation on the suprasegmental level), these include:

1. the phoneme inventory (vowels and consonants),

2. the phonetic realization of vowels and consonants, and

3. the phonotactic distribution of phonemes.

In recent years, many research methods for exploring phonological and phonetic variation on the segmental level have become established. Moreover, a number of new methods have become available, some of which have not yet been applied to this area. In the following sections, both the widely used and some new methods will be presented and discussed. Section 2 is concerned with the analysis of vowels in terms of phoneme inventory (Section 2.1), their phonetic realization (Section 2.2) and their articulation (Section 2.3). Section 3 evaluates methods of studying consonants: the consonant inventory (Section 3.1), their phonetic realization (Section 3.2) and their articulation (Section 3.3).

Introduction

This chapter describes methods of analysing phonetic and phonological variation on the suprasegmental (or prosodic) level. The suprasegmental level comprises all phonological units and processes that are larger than individual speech sounds. Thus, the prosodic features that can be investigated to document language variation and change include:

• stress placement in words and utterances

• speech rhythm and

• intonation.

There are a number of widely accepted research methods for studying and describing phonological and phonetic variation on the suprasegmental level. In Section 2, the auditory (Section 2.1) and the acoustic method (Section 2.2) of analysing word stress are discussed. Section 3 presents methods of analysing variation in speech rhythm. The methods of studying variation in intonation are presented in Section 4. This comprises the auditory method (Section 4.1), the combined auditory-acoustic method (Section 4.2) and the acoustic analysis of pitch and intonation (Section 4.3).

Word stress

Stress is a property of syllables and refers to the relative prominence a syllable has. It is defined as an abstract phonological category that forms part of a speaker’s knowledge: it refers to the speaker’s mental representation of a property of a specific syllable of a word. While in intonation languages such as Mandarin and Igbo stress seems to play a minor role, in other languages words have specific stress patterns. In the case of fixed stress languages, all multisyllabic words of a language have stress on a particular syllable, for example on the last syllable in Turkish or on the penultimate syllable in Welsh. In languages with free lexical stress like English, all content words with two or more syllables have at least one stressed syllable. Which of the syllables is stressed can be at least in part predicted by a set of complex rules (see e.g. Roach 1991: chapters 10 and 11).