Volume 46, Issue 4

We report three listening experiments focussing on the phonetic correlates of perceived foreignness and perceived strength of foreign accent in Swedish. In the first experiment, it is shown that accentedness can be reliably measured on the basis of a relatively short speech sample and that a statistical count of deviating pronunciations correlates significantly with subjective judgements of accentedness by linguistically naive listeners. The second experiment shows a significant correlation between number of deviant features and perceived accent strength and a significant correlation between judgements of degree of deviation and foreignness. Some features are associated with impressions of stronger accents than others, and some with perception of ‘Finnish’ accent. The third experiment establishes which combinations of deviant phonetic characteristics most strongly tend to give the impression of Finnish accent.

Results are presented from two experiments which examine the application of instrumental phonetic methods to the needs of the teacher of a foreign language. First, the use of a traditional psychophonetic test procedure as a diagnostic tool is demonstrated. Second, the effect of reading, imitation, and prompted recall on production is examined, and the implications for teaching and pronunciation monitoring discussed.

This paper describes the production of word-final, utterance-medial /t, d/ in American English by native speakers and by three groups of nonnative speakers of English. Twenty-four nonnative subjects, whose native languages include Arabic, Korean, Spanish, and Thai, were rated by native English speakers as having no accent, a mild accent, or a medium to heavy accent in English. Measurement of five tokens each of /t/ and /d/ for each speaker and for 6 native English speakers were compared. Neither the Americans nor any of the nonnative groups significantly differentiated /t/ from /d/ by closure duration. As strength of accent rating increased, mean closure duration tended to increase. Also, as strength of accent rating increased, implementation of manner of stop consonant production (including flapping and aspirated versus nonaspirated release) tended to change. Results suggest that both duration and the ability to produce a number of durationally different release options in English may affect the degree of perceived foreignness.

This paper reports the results of an inquiry into the question of category versus continuum in intonation. Variants of the English rise-fall-rise pattern were used to study whether tonal alignment is a categorical or gradient distinction. LPC resynthesis was used to construct a set of stimuli in which the alignment of the F₀ rise-fall varied in small steps. Subjects heard the stimuli in randomized order and imitated what they heard. The position of the F₀ peak relative to the onset of the stressed vowel was measured in each response. Systematic deviations between the peak placement in the stimuli and those in the responses revealed the existence of two categories. We conclude that tonal alignment functions as a binary distinction in English intonation.

The perceptual moment of occurrence of a syllable, its P-center, has frequently been examined by instructing subjects to adjust a series of speech sounds until they sounded isochronous. The present two experiments examined the effect of changing the instructions. In addition to the overall isochrony instructions, we asked subjects to align pairs of syllables so that the syllable onsets, vowel onsets, or syllable offsets sounded isochronous. In the first experiment, 3 of 4 subjects showed no difference among the first three instruction sets, and the changes introduced by the fourth went in the wrong direction. All subjects found it impossible to make alignments with respect to offsets. In the second experiment, vowel durations of two versions of some stimuli differed by 100 ms, to enhance the difference in syllable rhyme durations. Two subjects received the same instruction sets as in Experiment 1, and again found alignment with respect to offsets impossible. These subjects showed differences among the other instruction sets, although the direction and magnitude of the differences indicated that they had not succeeded in changing their timing criteria. The results indicate that P-center alignments are the syllable timing judgments that subjects most naturally make, and they may, indeed, be the only isochrony judgments that subjects can make reliably.

In three experiments, it is investigated to what extent naive listeners agree in ranking the stress of syllables in Dutch words. Four factors influencing the scores are isolated: the position of the primary stress, the awareness of rhythmic patterns, a tendency to assign prominence to the initial syllable and a segmental factor which disallows prominence on syllables with schwa. As these factors oppose or strengthen each other, they explain fluctuations in the subjects’ agreement.

This paper is concerned with the dependence of segment perception on microprosodic perturbations of a macroprosodic intonation contour. The test paradigm involves tokens of the German word pair ‘leiten’/’leiden’, where the vowel and closure durations are simultaneously manipulated. A falling versus a flattish fundamental frequency contour during the diphthong is shown to bias phoneme judgements towards lenis. As regards variation of F₀ in the syllabic nasal (falling vs. flattish vs. rising) both falling and rising contours cause an increase in fortis judgements. Furthermore, variations of F₀ in the vowel and in the nasal do not interact. In view of the necessity to use LPC synthetic speech as a tool in F₀ manipulation, the potential artefact of LPC synthesis in these experiments on F₀-conditioned segment perception is also investigated. The use of a synthetic instead of a natural vowel quality shifts the phoneme boundary towards more lenis, though this effect is weak.