English Prosody of Advanced Learners: A Contrastive Interlanguage Analysis
The present study investigated the production of prosody of 135 learners of English with the aim of identifying the acoustic properties of f0 in advanced Czech, German, and Spanish Learner English (n=45 each). Their performance was compared to that of monolingual native speakers of British (n=45) and American English (n=45). The main focus of the ... study concerns prosodic features of declarative utterances elicited from the second language (L2) and native spontaneous English speech (dialogic and monologic), as well as the influence of interviewer questions on some of the declarative utterances. Thus, it is essential to examine entrainment at the phonetic level within spontaneous interactions between interlocutors from the different first languages (L1s) and dialect backgrounds in the case of the native speakers. Within the framework of the autosegmental-metrical approach (Beckman and Pierrehumbert 1986), production of prosody was examined in different tonal events: pitch accents, boundary tones, and their tone heights (f0 measures). In order to achieve a phonetic and phonological description of these tonal events, it is important to investigate the domain of intonational phrasing (i.e. intonation phrases (IPs) and intermediate phrases (ips)) and the patterning of these tones into tunes. Additionally, the study seeks to clarify the relationship between prosodic form and socio-pragmatic meaning, necessitating the analysis of form-function correlations. The present study is one of the first to discuss the findings within a theoretical framework specifically designed for intonation, i.e. the L2 Intonational Learning Theory (LILt, cf. Mennen 2015). The findings are discussed in light of all extralinguistic variables known about all native and non-native speakers (n=225 in total), while the different interlanguages are compared to and with each other and to the two native varieties of English, thus, applying a Contrastive Interlanguage Analysis (Granger 1996, 2015). Results reveal a marked contrast between native and non-native speech production of nearly all prosodic events. Compared to the native speakers of English, most of the L2 speakers tend to produce longer ips in seconds, a slower speech rate within intonation units (IUs), more high edge tones, a narrower f0 span for high-low tunes, and a wider and higher f0 for high edge tunes. The only native-like prosodic features are tone frequency, the f0 level of high-low tunes, and IU frequency (for Spanish English learners only). In order to better explain these results, a multifactorial regression analysis of learning and recording context variables was performed, e.g. speaking style, age, gender, stays abroad, pragmatic functions (continuation, listing, etc.), speech acts (statements vs. answers), etc.. Main effects that explain the variation in L2 prosody are found in speaker gender and certain fluency variables, while minor effects were found for proficiency level, age, speaking style, and stay abroad. The results seem to support some predictions made by the LILt model (Mennen 2015). While the learners were found to differ on nearly all intonational dimensions of the LILt model and all prosodic variables investigated, the analyses of individual differences by learner-variety revealed specific trends. For instance, while the Spanish English speakers deviate mostly on the realizational level, i.e. speech rate per IU and f0 range in the different tunes, Czech and German English learners differ the most on the distributional and functional level with the choice of tonal categories (same trend), the variety of pragmatic functions chosen for these tonal categories (similar trends), and IU frequency (opposite trend). Based on these results, mostly proficiency-related explanations were given rather than L1-based influences, since all learners exhibited similarly deviant prosody. In sum, this study presents a quantitative mixed-method analysis of f0 in spontaneous native and non-native speech by identifying patterns of variation and universals in the phonological and phonetic form of f0, as conditioned by extralinguistic and contextual factors. Such an analysis informs our understanding of prosodic variation in native and non-native speech, as well as which factors can predict such variation. The study has implications for theoretical and applied linguistics. The results of the present study can fruitfully inform areas of relevance, such as theoretical second language acquisition processes, language assessment, language teaching, development of computer-based language teaching materials, and speech technology.