Wladyslaw Cichocki

Teaching Languages to Future Health Professionals: A Needs Assessment Study.

This article presents the results of a needs assessment study carried out with university students who were preparing to work as health professionals. Questionnaire data gauged the students’ expectations for language courses intended specifically for their areas of specialization. Analysis follows a multidimensional approach and examines profiles of the student body, student aspirations, and course contents. The results reveal a number of distinct patterns, which are discussed in comparison with current trends and practice in second language acquisition. They constitute a potential and valuable source of information for designing curricula for health professionals based on learners’ needs.

Geographic Variation in Acadian French /r /: What Can Correspondence Analysis Contribute Toward Explanation?

This article presents elements of a Correspondence Analysis (CA) approach to the measurement of linguistic distances in dialectology. It argues that both linguistic and spatial factors are part of an explanation of geolinguistic variation, and it shows how the exploratory and graphical properties of CA can contribute to such an explanation. The application is a study of the different realizations of the phoneme /r/ in Acadian French, a dialect spoken in Canada. Data are from the Atlas linguistique du vocabulaire maritime acadien and include over 5,000 tokens from eighteen localities. Using chi-square distances, the analysis results in a two-dimensional space that arranges the localities along continua. Linguistic interpretation of this space, based on those features of /r/ that are identified as accounting for the structuring of these continua, suggests a hierarchy of phonological processes-including alternation between apical and dorsal articulations (in French words), and replacement of the retroflex rhotic found in English-origin words by apical and dorsal variants. Two external spatial factors, local concentration of francophone speakers and spheres of activity, are shown to correlate with the linguistic distances among localities.

Native and non-native class discrimination using speech rhythm- and auditory-based cues

HighlightsNew models of speech rhythm and auditory knowledge are proposed.Use both duration and intensity metrics to classify native and non-native accents.Perform accent classification by logistic regression (LR) and with baseline systems.LR-based approach provides the best classification of native/non-native Arabic speech.Combination of auditoryc indicative features and rhythm metrics provides the best classification. In recent years, the use of rhythm-based features in speech processing systems has received growing interest. This approach uses a wide array of rhythm metrics that have been developed to capture speech timing differences between and within languages. However, the reliability of rhythm metrics is being increasingly called into question. In this paper, we propose two modifications to this approach. First, we describe a model that is based on auditory cues that simulate the external, middle and inner parts of the ear. We evaluate this model by performing experiments to discriminate between native and non-native Arabic speech. Data are from the West Point Arabic Speech Corpus; testing is done on standard classifiers based on Gaussian Mixture Models (GMMs), Support Vector Machines (SVMs) and a hybrid GMM/SVM. Results show that the auditory-based model consistently outperforms a traditional rhythm-metric approach that includes both duration- and intensity-based metrics. Second, we propose a framework that combines the rhythm metrics and the auditory-based cues in the context of a Logistic Regression (LR) method that can optimize feature combination. Further results show that the proposed LR-based method improves performance over the standard classifiers in the discrimination between the native and non-native Arabic speech.

Investigating Emphatic Consonants in Foreign Accented Arabic

This paper investigates the four emphatic consonants of Arabic from the point of view of automatic speech recognition. Comparisons of the recognition error rates for these phonemes and for their non-emphatic counterparts are analyzed in five experiments that involve different combinations of native and non-native Arabic speakers. In addition, the target consonants are described in time-frequency domain analyses. All experiments used the Hidden Markov Model toolkit (HTK) and the Language Data Consortium (LDC) WestPoint Modern Standard Arabic (MSA) database. Results confirm that emphatic consonants are a major source of difficulty for ASR. While the recognition rate for certain emphatic consonants such as /D/ can drop below 15% when uttered by non-native speakers, there are advantages to including non-native speakers in ASR. Regional differences in the pronunciation of MSA by native Arabic speakers require the attention of Arabic ASR research.

An Application of Dual Scaling in Dialectometry

Dialectometry is the application of numerical classification techniques to the analysis of dialect data. Among the goals of this kind of study is the search for methodologies which calculate linguistic distances among dialects or which extract patterns of correlations that are present between localities and the linguistic variants found in the localities. Because they provide &dquo;objective&dquo; procedures which are designed specifically to carry out this kind of investigation, multivariate statistical techniques are receiving increasing attention among researchers working in this area. The multivariate approach is especially useful when the data set under study is large. Several multivariate techniques are used in dialectometrical analysis. In multidimensional scaling, distances or similarity measures among the localities are calculated and represented as a configuration of points in a twoor three(or greater) dimensional space (e.g. Embleton 1987). Cluster analysis places the localities at the bottom of a hierarchical tree; the structure of the branching nodes on the tree indicates the relative degree of relatedness among the localities (e.g. Linn and Regal 1985). Correspondence analysis permits the relations among the localities and among linguistic elements used in describing the localities to be represented together in a multidimensional space (e.g. Philps 1985, Flikeid and Cichocki 1988). Researchers, then, can select the technique which is best suited to the particular research questions that they

Variation of rhythm metrics in regional varieties of Acadian French

This preliminary investigation studies variation of rhythm metrics in dialects of Acadian French spoken in New Brunswick (Canada). The aim is to determine whether regional and social factors are significant sources of this variation. Data are recordings of 140 speakers who represent five geographic regions, both genders and two age groups (20-30 and 40-55 years of age). Sound files were segmented manually; eight interval-based metrics were calculated. Mean metric scores indicate that regional varieties of Acadian French pattern with other dialects of French; these scores are similar to those of other so-called syllable-timed languages. Regional differences are found for several metrics (%V, Delta C, VarcoC, nPVI-C, rPVI-C), although the five geographic regions are not all clearly distinguished by these metrics. A major pattern that emerges is regional variation in high-vowel devoicing and/or deletion. Analyses also show that social factors are significant sources of interspeaker variability: gender (Varco...