An empirical investigation of pronunciation problems of young learners of English as a foreign language - identification and remedial strategies

This thesis describes a novel pronunciation training programme devised for young learners of English as a foreign language in schools in Germany. The intervention is grounded in current theories of second language, employs valid and established pronunciation task formats, and targets a core set of significant pronunciation problem areas typical of young learners. This core set was identified from research literature, curricular requirements as well as data from questionnaires completed by 245 English language teachers in the German state of Hesse. The English pronunciation intervention was implemented over five months within the language education setting of English L2 learners (ages 10 to 12) in two schools in Frankfurt/Main. Quasi-experimental classroom research was conducted using a control group design to examine the effects of the pronunciation intervention. To obtain within-subject measurements, the data were collected at three points: prior to the intervention (pre-test), immediately after the intervention (post-test) and six months later (follow-up). There were two types of analyses of the pronunciation data. First there was auditory evaluation: all test utterances by the students were assessed auditorily using a three-point scale to rate their performance. Secondly, acoustic analysis of the same data was conducted using the computer software programme Praat. The acoustic parameters of interest included fundamental and formant frequencies, vowel and fricative durations, spectral peaks and spectral rate of change, to assess changes in pronunciation. The main aim of the analyses was to measure the impacts, if any, of the pronunciation intervention. The results revealed that maturation processes took place for both the intervention and control groups, as predicted. Overall, the auditory analysis presented mixed results but showed some training effects for the fricatives /ð/ and /θ/ and the affricate /dʒ/. Most salient of all, there were clear frequency effects. The acoustic data analysis indicated more systematic effects of the intervention on acoustic parameters associated with precision and stability. The scatter plots and Euclidean distances computed from the acoustic data revealed more precision and stability in the production of the vowels /ɑː, ɔː, ə, æ/, the diphthongs /ɪə; eɪ/ and the approximant /w/ in the intervention group. In addition, the centre of gravity values indicated that the training led to a more native-like production of /θ/ and /ð/.