| Foreign language teachers traditionally rely on their auditory-perceptual judgments to identify pronunciation problems, to decide on learning strategies, and to evaluate progress in pronunciation over time and training. They typically compare the perceived stimulus to their own internal standards or scales. However, their perceptual ratings are contingent upon factors such as the tasks, the influence of grammatical or lexical errors, the specificity of the rating scales, the teacher/learner length of acquaintance, and the teachers' levels of fluency and training. All these factors raise the question of judgment reliability.;The waveform of utterances can be measured. Waveform components comply with the laws of physics and can, therefore, be quantified. In acoustic phonetics, the analyst isolates specific cues to perform acoustic analyses on selected features of speech samples, e.g., vowels, consonants, and stress patterns. Such analyses include measures of amplitude (dB), fundamental frequency (F o) or formant frequency variations measured in Hertz (Hz), and duration. However, they require that the measurements and the protocol be strictly controlled. In addition, the selected cues may not necessarily be salient or significant correlates of an 'accent,' thus raising the question of validity.;This research measures the outcome of formal pronunciation training by comparing acoustic measurements to auditory-perceptual ratings. The speech samples consist of segmental and suprasegmental audio-recordings produced by forty-five female subjects, aged 18--45, at three dates approximately three months apart. There were fifteen native speakers of French to provide the normative baseline, twenty adult American students learning French, and ten experienced non-native speakers of French.;In the acoustic analyses, fifteen dependent variables are analyzed as potential carriers of foreign-accented pronunciation, using the Kay Elemetrics Computerized Speech Lab (CSL). The perceptual study involves six French native speakers who teach French but are not trained in teaching pronunciation. In each study, the speech stimuli are evaluated along three temporal and pedagogical parameters: (1) pre-, posttest, (2) longitudinal study, and (3) computerized audio-visual training.;The results show that the difference between Formant 2 and Formant 1 frequency of [y] and [u], total duration of utterance, voice onset time (VOT) of syllable initial [p], and syllable duration are carriers of an English accent in French and correlate with perceived levels of foreign accentedness. |
