The current study lays the groundwork for a model of Mandarin tones based on both native speakers' and non-native speakers' perception and production. It demonstrates that there is variability in non-native speakers' tone productions and that there are differences in the perceptual boundaries in native speakers and non-native speakers. There are four experiments in this study. Experiment 1 utilizes native speakers' production data from a published speech database to explore the features of tone production by native speakers. Inter-speaker normalization is used to analyze the data. Experiment 2 synthesizes 81 tones that are carried by four sentences to measure perception by native and non-native speakers. The intra-speaker and inter-speaker normalization is used to investigate the perceptual space of T1, T2, T3, and T4. The researcher also explores the salient features distinguish native speakers' and non-native speakers' perception of the four principal tones. Experiment 3 uses both synthesized tones and natural tones that are carried by sentences to explore how pitch values of tones create overlapping areas in the perceptual map. Experiment 4 examines tone production by non-native speakers to identify the differences between native speakers' perception and non-native speakers' production; and the differences between non-native speakers' perception and their production of tones. The results of the perception and production experiments with native speakers show the perceptual boundaries and tonal categories in the perceptual space and the production space. The difference of native speakers' perception and production shows us the perceptual cue for perception. Meanwhile, the similarities of native speakers' perception and production reveal the acoustic cues, including register and contour, for tone perception and production. The results of the perception experiments with non-native speakers indicate that there are no clear boundaries, and that tone overlap in the perceptual space. Register plays an important role in the perception of tones by non-native speakers. The results of non-native speaker production also show overlapping tones in the acoustic space. The non-native speaker production appears to be determined by the contour of the tones in contrast both the contour and register determine the tonal categories of native speaker.

<p>To achieve a robust system the variation seen for different speaking styles must be handled. An investigation of standard automatic speech recognition techniques for different speaking styles showed that lexical modelling using general-purpose variants gave small improvements, but the errors differed compared with using only one canonical pronunciation per word. Modelling the variation using the acoustic models (using context dependency and/or speaker dependent adaptation) gave a significant improvement, but the resulting performance for non-native and spontaneous speech was still far from read speech.</p><p>In this dissertation a complete data-driven approach to rule-based lexicon adaptation is presented, where the effect of the acoustic models is incorporated in the rule pruning metric. Reference and alternative transcriptions were aligned by dynamic programming, but with a data-driven method to derive the phone-to-phone substitution costs. The costs were based on the statistical co-occurrence of phones, association strength. Rules for pronunciation variation were derived from this alignment. The rules were pruned using a new metric based on acoustic log likelihood. Well trained acoustic models are capable of modelling much of the variation seen, and using the acoustic log likelihood to assess the pronunciation rules prevents the lexical modelling from adding variation already accounted for as shown for direct pronunciation variation modelling.</p><p>For the non-native task data-driven pronunciation modelling by learning pronunciation rules gave a significant performance gain. Acoustic log likelihood rule pruning performed better than rule probability pruning.</p><p>For spontaneous dictation the pronunciation variation experiments did not improve the performance. The answer to how to better model the variation for spontaneous speech seems to lie neither in the acoustical nor the lexical modelling. The main differences between read and spontaneous speech are the grammar used and disfluencies like restarts and long pauses. The language model may thus be the best starting point for more research to achieve better performance for this speaking style.</p>

To achieve a robust system the variation seen for different speaking styles must be handled. An investigation of standard automatic speech recognition techniques for different speaking styles showed that lexical modelling using general-purpose variants gave small improvements, but the errors differed compared with using only one canonical pronunciation per word. Modelling the variation using the acoustic models (using context dependency and/or speaker dependent adaptation) gave a significant improvement, but the resulting performance for non-native and spontaneous speech was still far from read speech. In this dissertation a complete data-driven approach to rule-based lexicon adaptation is presented, where the effect of the acoustic models is incorporated in the rule pruning metric. Reference and alternative transcriptions were aligned by dynamic programming, but with a data-driven method to derive the phone-to-phone substitution costs. The costs were based on the statistical co-occurrence of phones, association strength. Rules for pronunciation variation were derived from this alignment. The rules were pruned using a new metric based on acoustic log likelihood. Well trained acoustic models are capable of modelling much of the variation seen, and using the acoustic log likelihood to assess the pronunciation rules prevents the lexical modelling from adding variation already accounted for as shown for direct pronunciation variation modelling. For the non-native task data-driven pronunciation modelling by learning pronunciation rules gave a significant performance gain. Acoustic log likelihood rule pruning performed better than rule probability pruning. For spontaneous dictation the pronunciation variation experiments did not improve the performance. The answer to how to better model the variation for spontaneous speech seems to lie neither in the acoustical nor the lexical modelling. The main differences between read and spontaneous speech are the grammar used and disfluencies like restarts and long pauses. The language model may thus be the best starting point for more research to achieve better performance for this speaking style.

We refactor the traditional two-step approach of automatic speech recognition for spoken language translation. Instead of conventional graphemes, we use phonemes as an intermediate speech representation. Starting with the acoustic model, we revise the cross-lingual transfer and propose a coarse-to-fine method providing further speed-up and performance boost. Further, we review the translation model. We experiment with source and target encoding, boosting the robustness by utilizing the fine-tuning and transfer across ASR and SLT. We empirically document that this conventional setup with an alternative representation not only performs well on standard test sets but also provides robust transcripts and translations on challenging (e.g., non-native) test sets. Notably, our ASR system outperforms commercial ASR systems. 1

<p>The present dissertation explores the effect of exposure to non-native speech via foreign film on non-native speech production and perception. In order to explore potential effects, two main experiments were developed, which examined French production and perception by monolingual native speakers of English before and after exposure to French film. Across both experiments, two variables were selected for observation: high rounded vowels (/y/ and /u/) and consonant voicing (VOT). The production component of the dissertation investigated whether exposure to French film aided in the ability of monolingual American English speakers (n=74) to shadow French words containing high rounded vowels, /y/ and /u/, as tested through acoustic analyses and native French listener perceptual judgements (n=221). Perception of high rounded vowels and consonant voicing were examined using a perceptual assimilation task with category goodness ratings and a binary forced-choice voicing identification task, respectively.</p> <p>With regard to the role of foreign film in non-native speech production, results indicated that a single session of exposure to French film had a small but significant effect on shadowing of French /y/, which was also perceptible to native French listeners. Shadowing of /u/, however, was not significantly affected by exposure. Additionally, while the acoustic analysis of VOT did not reveal any significant effects of film, native French listeners perceived post-film exposure productions to be significantly more target-like than pre-film exposure productions. This finding suggests that although VOT was seemingly unaffected by foreign film exposure, participants may have adjusted alternative acoustic correlates of voicing and that these modifications were perceptible to native listeners. In general, these results suggest that while potential effects of film are present, they are highly dependent on the variable being observed.</p> <p>Results from the perceptual portion of the dissertation do not provide evidence that film exposure was effective at influencing non-native speech perception for either high rounded vowels or consonant voicing. However, it is suggested that this could be due to the difficulty of the tasks chosen rather than the effectiveness of foreign film.</p> <p>Taken together, the present dissertation provides evidence that exposure to non-native speech via foreign film can affect some aspects of non-native speech learning. It is hypothesized that further sessions may compound these initial benefits, especially in those who are already learning a second language.</p>

No description available.

Les technologies de reconnaissance automatique de la parole sont désormais intégrées dans de nombreux systèmes. La performance des systèmes de reconnaissance vocale pour les locuteurs non natifs continue cependant à souffrir de taux d'erreur élevés, en raison de la différence entre la parole non native et les modèles entraînés. La réalisation d'enregistrements en grande quantité de parole non native est généralement une tâche très difficile et peu réaliste pour représenter toutes les origines des locuteurs. Ce travail de thèse porte sur l'amélioration des modèles acoustiques multilingues pour la transcription phonétique de la parole de type « réunion multilingue ». Traiter ce type de parole constitue plusieurs défis : 1) il peut exister de la conversation entre des locuteurs natifs et non natifs ; 2) il y a non seulement de la parole non native d'une langue, mais de plusieurs langues parlées par des locuteurs venant de différentes origines ; 3) il est difficile de collecter suffisamment de données pour amorcer les systèmes de transcription. Pour répondre à ces défis, nous proposons un processus d'adaptation de modèles acoustiques multilingues que nous appelons « adaptation autonome ». Dans l'adaptation autonome, nous étudions plusieurs approches pour adapter les modèles acoustiques multilingues de manière non supervisée (les langues parlées et les origines des locuteurs ne sont pas connues à l'avance) et qui n'utilise aucune donnée supplémentaire lors du processus d'adaptation. Les approches étudiées sont décomposées selon deux modules. Le premier module qui s'appelle « l'observateur de langues » consiste à récupérer les caractéristiques linguistiques (les langues parlées et les origines des locuteurs) des segments à décoder. Le deuxième module consiste à adapter le modèle acoustique multilingue en fonction des connaissances fournies par l'observateur de langue. Pour évaluer l'utilité de l'adaptation autonome d'un modèle acoustique multilingue, nous utilisons les données de test, qui sont extraites de réunions multilingues, contenant de la parole native et non native de trois langues : l'anglais (EN), le français (FR) et le vietnamien (VN). Selon les résultats d'expérimentation, l'adaptation autonome donne des résultats prometteurs pour les paroles non natives mais dégradent très légèrement les performances sur de la parole native. Afin d'améliorer la performance globale des systèmes de transcription pour toutes les paroles natives et non natives, nous étudions plusieurs approches de détection de parole non native et proposons de cascader un tel détecteur avec notre processus d'adaptation autonome. Les résultats obtenus ainsi, sont les meilleurs parmi toutes les expériences réalisées sur notre corpus de réunions multilingues. / Automatic speech recognition technologies are now integrated into many systems. The performance of speech recognition systems for non-native speakers, however, continues to suffer high error rates, due to the difference between native and non-speech models trained. The making of recordings in large quantities of non-native speech is typically a very difficult and impractical to represent all the origins of the speakers. This thesis focuses on improving multilingual acoustic models for automatic phonetic transcription of speech such as “multilingual meeting”. There are several challenges in “multilingual meeting” speech: 1) there can be a conversation between native and non native speakers ; 2) there is not only one spoken language but several languages spoken by speakers from different origins ; 3) it is difficult to collect sufficient data to bootstrapping transcription systems. To meet these challenges, we propose a process of adaptation of multilingual acoustic models is called "autonomous adaptation". In autonomous adaptation, we studied several approaches for adapting multilingual acoustic models in unsupervised way (spoken languages and the origins of the speakers are not known in advance) and no additional data is used during the adaptation process. The approaches studied are decomposed into two modules. The first module called "the language observer" is to recover the linguistic information (spoken languages and the origins of the speakers) of the segments to be decoded. The second module is to adapt the multilingual acoustic model based on knowledge provided by the language observer. To evaluate the usefulness of autonomous adaptation of multilingual acoustic model, we use the test data, which are extracted from multilingual meeting corpus, containing the native and nonnative speech of three languages: English (EN), French (FR) and Vietnamese (VN). According to the experiment results, the autonomous adaptation shows promising results for non native speech but very slightly degrade performance on native speech. To improve the overall performance of transcription systems for all native and non native speech, we study several approaches for detecting non native speech and propose such a detector cascading with our self-adaptation process (autonomous adaptation). The results thus are the best among all experiments done on our corpus of multilingual meetings.

The purpose of this work is to measure the degree of intelligibility of native and non-native English speech as well as analyse the major sources of intelligibility failure when speakers of these varieties of English interact. British and American English (henceforth BrE and AmE) and Cameroon English (hereafter CamE) are used as a case study with focus on segmental and supra segmental features. The study was motivated by a number of concerns, several of which are more prominent: First, it was motivated by the trepidation scholars like Gimson (1965, 1980); Prator (1968); etc. nursed that the unprecedented spread of English across the globe and the emergence of non-native varieties would cause English to disintegrate into mutually unintelligible languages, in the way Romance languages devolved from their Latin ancestor. The second motivation was that previous researchers (Bansal 1969, Tiffen 1974) on intelligibility have often concentrated their efforts on the traditional approach, which sees intelligibility from a one-sided perspective. To them, the non-native varieties of English are deficient and not different varieties from the native varieties. They were seen as substandard, incorrect, and unintelligible and thus needed remediation at all costs. The native varieties were seen as prestigious, correct, intelligible and the sole norm that must be emulated by non-native English speakers. In this way any interaction between a native speaker and a non-native speaker should be characterised by the non-native speaker making all the efforts to be understood as well as to understand the native English-speaking partner. This explains in large part why these researchers concentrated on measuring the intelligibility of non-native speech to native speakers and never vice versa. It was as if it was treasonable to measure the intelligibility of native speech to non-native speakers. Even if some researchers managed to do this, the comments that followed such data still showed that the aim was not to test the intelligibility of native speakers but to find out how efficient the non-native speakers were in understanding the native speaker. Another aim could also be to reinforce the teaching of the native norm, which was seen as “correct” against non-native features, which were seen as “incorrect”, to measuring intelligibility. While accepting that these studies reflected the conventional wisdom of the time, this study aims to move the debate forward by looking at intelligibility from a two-sided perspective. It sees communication between speakers of different varieties as a game of give and take, where both participants “tune in” to make the process successful rather than one participant being obliged to make all the efforts because s/he speaks a new English variety. That explains why we are testing not only the intelligibility of non-native speakers to native speakers but also native speakers to non-native speakers. / Gegenstand der vorgelegten Promotionsarbeit ist die Untersuchung der gegenseitigen Verständlichkeit von muttersprachlichem und nicht-muttersprachlichem Englisch. Im besonderen werden die Hauptquellen und Ursachen des Scheiterns von Verständlichkeit in einer empirischen Studie bestimmt, klassifiziert und analysiert. Die Untersuchung wird exemplarisch anhand des Kamerunischen Englisch einerseits und des Britischen und Amerikanischen Englisch anderseits vorgenommen. Motiviert ist diese Arbeit vor allem durch folgende Punkte. Erstens bedarf es der Auseinandersetzung mit den durch eine Reihe von Autoren geäußerten Befürchtungen (z.B. Gimson 1965, 1980 und Prator 1968), daß die Herausbildung und Entwicklung neuer Varianten des Englischen letztlich zu einer Auflösung des Englischen in gegenseitig nicht mehr verständliche Sprachen führt, ein Prozeß, wie er sich historisch bei der Entstehung der romanischen Sprachen aus dem Lateinischen vollzog. Derartige Befürchtungen werden genährt durch die bisher ohnegleichen fortschreitende Verbreitung des Englischen über den gesamten Globus. Hier ergibt sich die dringende Notwendigkeit vergleichender Studien zur gegenseitigen Verständlichkeit zwischen den bestehenden Varianten. Zweitens folgen die meisten zu diesem Thema vorliegenden Untersuchungen im wesentlichen einer traditionellen, überkommenen Grundperspektive: die nicht-muttersprachlichen Varianten des Englischen werden als „abweichend“ bzw. sogar „defizitär“ aus Sicht der muttersprachlichen betrachtet, nicht aber als eigenständige Sprachformen (z.B. Bansal 1969; Tiffen 1974). Dies führt nach Auffassung des Autors zu einer einseitigen Betrachtung und Bewertung. Im besonderen wird in der sprachlichen Interaktion bei einem solchen Zugang die Last zu verstehen und für den Kommunikationspartner verständlich zu sein einseitig dem nicht-muttersprachlichen Sprecher übertragen. Auf diesem Hintergrund untersuchen die vorliegenden Studien anderer Autoren primär die Verständlichkeit nicht-muttersprachlicher Sprachformen für den muttersprachlichen Sprecher, nicht jedoch die umgekehrte Konstellation. Wenn die umgekehrte Perspektive überhaupt berücksichtigt wird, so zeigen die Kommentare, daß nicht die Verständlichkeit muttersprachlicher Sprecher für den Nichtmuttersprachler eigentliches Ziel und Gegenstand der Untersuchung war, sondern vielmehr die Frage, wie effizient sich Nichtmuttersprachler beim Verstehen muttersprachlicher Äußerungen zeigten. Des weiteren stehen diese Studien oft im Kontext des Bestrebens, im institutionalisierten Spracherwerb die muttersprachlichen Normen gegen die nicht-muttersprachlichen Merkmale durchzusetzen, die als „nicht korrekt“ angesehen werden. Diese Positionen, der zugrundeliegende Zugang und die einseitige Ausrichtung bedürfen einer kritischen Auseinandersetzung.

Cette thèse s'inscrit dans le cadre d'un travail descriptif en phonétique acoustique, avec comme objet d'étude les productions des voyelles du luxembourgeois dans la parole native et non native. L'intérêt est de concilier la variation du luxembourgeois, une langue principalement parlée, composée de nombreuses variétés régionales, évoluant dans un contexte multilingue, et son apprentissage dans le cadre de l'enseignement des langues étrangères au Grand-Duché de Luxembourg. Comme nous partons du fait que l'apprentissage d'une langue implique la connaissance des traits contrastifs des sons, nous nous intéressons aux productions de locuteurs dont la langue maternelle possède des traits différents de ceux du luxembourgeois, comme le français, afin de voir si ces traits sont reproduits dans la parole non native. Les productions vocaliques de locuteurs francophones sont étudiées en comparaison aux productions de locuteurs natifs de la région située autour de la capitale du Grand-Duché de Luxembourg, dont la variété sert de référence à l'enseignement du luxembourgeois en tant que langue étrangère. Le but de l'analyse est :- d'étendre les descriptions sur les propriétés acoustiques des voyelles produites dans une variété régionale du Grand-Duché de Luxembourg,- de relever les difficultés de productions de locuteurs francophones qui apprennent le luxembourgeois,- d'interpréter les résultats dans le cadre de l'enseignement du luxembourgeois en tant que langue étrangère. Une partie importante du travail empirique a été consacrée à la collecte des données et la création d'un corpus obtenu à travers des enregistrements de 10 locuteurs luxembourgophones et de 10 locuteurs francophones. Le corpus de compose de 12h30 de parole lue et spontanée, incluant de la parole native et non native du luxembourgeois, ainsi que de la parole native du français. Ce corpus constitue un premier corpus sur la parole native et non native du luxembourgeois et permet de faire divers analyses comparatives. Dans notre étude, nous avons fait des analyses acoustiques sur les données de la parole lue. La méthodologie utilisée a permis d'effectuer des comparaisons entre les données de la parole native et non native du luxembourgeois ainsi qu'entre les données de la L1 et la L2 des francophones. Les résultats ont apporté des informations tant sur les productions natives que sur les productions non natives des voyelles. Ils ont montré que, d'un côté, les productions vocaliques varient en fonction des locuteurs, même si ceux-ci parlent la même variété régionale et, d'un autre côté, les locuteurs francophones apprenant le luxembourgeois en niveau B1/B2 ont des difficultés à produire les contrastes en luxembourgeois, à savoir :- la durée des voyelles longues [i:], [e:], [a:], [o:], [u:] et brèves [i], [e], [ɑ], [ɔ], [u],- le timbre de la voyelle longue [a:] et des voyelles brèves [æ] et [ɑ],- le timbre du début des diphtongues [æi], [æu], [ɑi], [ɑu].Ces résultats, ainsi que les descriptions approfondies sur les voyelles dans la parole native, enrichissent non seulement les connaissances sur le luxembourgeois, mais aussi sur la variété servant de référence au luxembourgeois en tant que langue étrangère. En outre, ils ouvrent des perspectives d'étude sur le luxembourgeois en problématisant l'instauration de règles pour ce type d'enseignement, malgré l'absence d'un enseignement suivi de la langue dans les écoles et l'évolution des variétés régionales sur un territoire géographique concentré. / This thesis is part of a descriptive work in acoustic phonetics, with the aim of studying the productions of Luxembourgish vowels in native and non-native speech. Its objective is to conciliate the variation of Luxembourgish, mainly a spoken language, composed of many regional varieties, evolving in a multilingual context, and the learning of Luxembourgish as a foreign language in the Grand-Duchy of Luxembourg. As we assume the fact that language learning implies knowledge of sound contrast in speech, we investigate the productions of speakers whose mother tongues have different features than Luxembourgish, such as French, to see whether if the contrast are reproduced in non-native speech. Productions of French speakers are compared to those of native speakers from the region around the capital city of the Grand-Duchy of Luxembourg, whose variety serves as a reference to the teaching of Luxembourgish as a foreign language. The purpose of the study is the following :- to extend the descriptions on the acoustic properties of vowels produced in a regional variety of the Grand-Duchy of Luxembourg,- to highlight the specific difficulties of productions by French learners of Luxembourgish,- to interpret the results regarding the teaching of Luxembourgish as a foreign language.Fieldwork and the creation of a corpus through recordings of 10 Luxembourg speakers and 10 French speakers are an important part of the empirical work. We obtained a corpus of 12 hours and a half of spoken and spontaneous speech, including native speech and not native of Luxembourgish and also native speech of French. This corpus represents a first corpus containing native and non-native speech of Luxembourgish and enables to conduct different comparative studies. In our thesis, we did comparative analyses of the data in read speech. The methodology we used made it possible to compare data of native and non-native speech and also data of the L1 and L2 of French speakers.The results gave information about native and non-native productions of vowels. They showed that, on the one hand, vowel productions vary among speakers, even if these speak the same regional variety and, on the other hand, French speakers who learn Luxembourgish at B1/B2 level have difficulties producing contrasts in Luxembourgish. This concerns :- the quantity of the long vowels [i:], [e:], [a:], [o:], [u:] and short vowels [i], [e], [ɑ], [ɔ], [u],- the quality of the long vowel [a:] and the two short vowels [æ] et [ɑ],- the quality of the beginning of the diphthongs [æi], [æu], [ɑi], [ɑu].These results as well as thorough descriptions of the vowels in native speech, extend knowledge not only of Luxembourgish, but also of the variety which serves as the reference for Luxembourgish as a foreign language. In addition, they open up prospects for studying Luxembourgish by problematizing the introduction of rules for this type of education, despite the absence of language instruction in schools and the evolution of regional varieties in a concentrated geographical area.

The purpose of this work is to measure the degree of intelligibility of native and non-native English speech as well as analyse the major sources of intelligibility failure when speakers of these varieties of English interact. British and American English (henceforth BrE and AmE) and Cameroon English (hereafter CamE) are used as a case study with focus on segmental and supra segmental features. The study was motivated by a number of concerns, several of which are more prominent: First, it was motivated by the trepidation scholars like Gimson (1965, 1980); Prator (1968); etc. nursed that the unprecedented spread of English across the globe and the emergence of non-native varieties would cause English to disintegrate into mutually unintelligible languages, in the way Romance languages devolved from their Latin ancestor. The second motivation was that previous researchers (Bansal 1969, Tiffen 1974) on intelligibility have often concentrated their efforts on the traditional approach, which sees intelligibility from a one-sided perspective. To them, the non-native varieties of English are deficient and not different varieties from the native varieties. They were seen as substandard, incorrect, and unintelligible and thus needed remediation at all costs. The native varieties were seen as prestigious, correct, intelligible and the sole norm that must be emulated by non-native English speakers. In this way any interaction between a native speaker and a non-native speaker should be characterised by the non-native speaker making all the efforts to be understood as well as to understand the native English-speaking partner. This explains in large part why these researchers concentrated on measuring the intelligibility of non-native speech to native speakers and never vice versa. It was as if it was treasonable to measure the intelligibility of native speech to non-native speakers. Even if some researchers managed to do this, the comments that followed such data still showed that the aim was not to test the intelligibility of native speakers but to find out how efficient the non-native speakers were in understanding the native speaker. Another aim could also be to reinforce the teaching of the native norm, which was seen as “correct” against non-native features, which were seen as “incorrect”, to measuring intelligibility. While accepting that these studies reflected the conventional wisdom of the time, this study aims to move the debate forward by looking at intelligibility from a two-sided perspective. It sees communication between speakers of different varieties as a game of give and take, where both participants “tune in” to make the process successful rather than one participant being obliged to make all the efforts because s/he speaks a new English variety. That explains why we are testing not only the intelligibility of non-native speakers to native speakers but also native speakers to non-native speakers. / Gegenstand der vorgelegten Promotionsarbeit ist die Untersuchung der gegenseitigen Verständlichkeit von muttersprachlichem und nicht-muttersprachlichem Englisch. Im besonderen werden die Hauptquellen und Ursachen des Scheiterns von Verständlichkeit in einer empirischen Studie bestimmt, klassifiziert und analysiert. Die Untersuchung wird exemplarisch anhand des Kamerunischen Englisch einerseits und des Britischen und Amerikanischen Englisch anderseits vorgenommen. Motiviert ist diese Arbeit vor allem durch folgende Punkte. Erstens bedarf es der Auseinandersetzung mit den durch eine Reihe von Autoren geäußerten Befürchtungen (z.B. Gimson 1965, 1980 und Prator 1968), daß die Herausbildung und Entwicklung neuer Varianten des Englischen letztlich zu einer Auflösung des Englischen in gegenseitig nicht mehr verständliche Sprachen führt, ein Prozeß, wie er sich historisch bei der Entstehung der romanischen Sprachen aus dem Lateinischen vollzog. Derartige Befürchtungen werden genährt durch die bisher ohnegleichen fortschreitende Verbreitung des Englischen über den gesamten Globus. Hier ergibt sich die dringende Notwendigkeit vergleichender Studien zur gegenseitigen Verständlichkeit zwischen den bestehenden Varianten. Zweitens folgen die meisten zu diesem Thema vorliegenden Untersuchungen im wesentlichen einer traditionellen, überkommenen Grundperspektive: die nicht-muttersprachlichen Varianten des Englischen werden als „abweichend“ bzw. sogar „defizitär“ aus Sicht der muttersprachlichen betrachtet, nicht aber als eigenständige Sprachformen (z.B. Bansal 1969; Tiffen 1974). Dies führt nach Auffassung des Autors zu einer einseitigen Betrachtung und Bewertung. Im besonderen wird in der sprachlichen Interaktion bei einem solchen Zugang die Last zu verstehen und für den Kommunikationspartner verständlich zu sein einseitig dem nicht-muttersprachlichen Sprecher übertragen. Auf diesem Hintergrund untersuchen die vorliegenden Studien anderer Autoren primär die Verständlichkeit nicht-muttersprachlicher Sprachformen für den muttersprachlichen Sprecher, nicht jedoch die umgekehrte Konstellation. Wenn die umgekehrte Perspektive überhaupt berücksichtigt wird, so zeigen die Kommentare, daß nicht die Verständlichkeit muttersprachlicher Sprecher für den Nichtmuttersprachler eigentliches Ziel und Gegenstand der Untersuchung war, sondern vielmehr die Frage, wie effizient sich Nichtmuttersprachler beim Verstehen muttersprachlicher Äußerungen zeigten. Des weiteren stehen diese Studien oft im Kontext des Bestrebens, im institutionalisierten Spracherwerb die muttersprachlichen Normen gegen die nicht-muttersprachlichen Merkmale durchzusetzen, die als „nicht korrekt“ angesehen werden. Diese Positionen, der zugrundeliegende Zugang und die einseitige Ausrichtung bedürfen einer kritischen Auseinandersetzung.