How do listeners segment the continuous speech input into words? This paper addresses this important question by reviewing current views of lexical segmentation. After providing some background to the more general problem of spoken word recognition, we examine the different cues - including phonetic, phonological, prosodic, and lexical cues - that have been shown to play a role in speech segmentation. We then analyze how these cues are used to locate alignment points in the continuous speech signal. These points serve to define what parts of the signal are matched with which representations in the mental lexicon. Several solutions to the alignment problem are discussed in light of existing experimental evidence.

More often than not, oral data differ from written data both from a frequential and distributional point of view. They lead to a sharpening of the description as they supply us with construction characteristics or contexts that do not exist in writing.Dislocations and the form 'il y a', known to be lavishly used in the oral language, provide us with often predictable examples regarding their distributional characteristics as well as their function in textual dynamics.Although 'certains' as a subject is not much used in spoken language, its offers a variety of distribution facts which are also clearly divided and actually related to the kind of observed corpus.The following article therefore aims at presenting us with three case studies that are representative of the relationship between data and analysis.

In learning the phonetics of a foreign language an adult student takes in new sounds still sustaining the definitive influence of the phonological categorization of his or her mother tongue. In Russia phonetic correction mainly makes use of such methods as visual display as well as tactile and motive drills aimed at developing articulation. By making pronunciation subordinate to myoneural control that procedure virtually disregards the prosodic component of the speech. Teaching foreign language prosody as presented in this article is based on the fact that a student cannot only hear but also see the appropriate audio signal. The use of the visual channel enhances work on the prosody of a language to be studied both at the stage of its perception and at the stage of production.

This article describes ideas of the project "Collection of Currently Spoken German". This project is presently planned at the Institute of Phonetics at Munich University. The aim of the project is to collect aspects of currently spoken German by means of a static network of recording stations distributed all over the German speaking area. The RVG1 (Regional Variants of German) corpus serves as a prototype for regionally covered speech data. It can be seen as a first small database of regionally covered recordings of German representing the most common dialectal regions or at least all those regions which could be important for categorizing regional variants into broader classes. RVG1 contains read numbers, phonetically rich sentences and computer commands as well as spontaneous speech. Some features of this corpus will be introduced and discussed.

The PAC project (The Phonology of Contemporary English: usage, varieties, structure) aims at giving a better picture of spoken English in its unity and its geographical, social and stylistic diversity. Based on Labovian methods, the project seeks to describe both rhotic and non rhotic accents of English, from traditional standards to more recent postcolonial varieties. This large corpus enables researchers to analyse and compare intervarietal features such as rhoticity as well as more specific phenomena such as vocalic length in Australian English or variable rhoticity in New Zealand English. Today LVTI, a collaborative project aiming at an interdisciplinary sociolinguistic survey of great urban centres such as Manchester and Toulouse is being set up following the PAC/PFC classical protocol.

This article presents the structure of the Corpus International Ecologique de la Langue Française, an extensive corpus of spoken French that will soon be available on the Internet, from both an epistemological and empirical perspective. Explanations are given with regard to the ideas that guided the data collection (ecological approach, comparability of the different areas of the Francophonie and communication situations) and to the choices made ("communicative spaces" and "activity types") with a view to relevant analyses in various research fields (variation, interaction, multimodality, French in contact, oral syntax) and an attempt is made to fill existing gaps in the current corpus. The article further addresses the issue of building up a network of experts, problems that had to be solved during fieldwork in the different areas and questions concerning standardisation, archiving and publication of the collected data (audio and video recordings, transcriptions, metadata), whereupon several examples are presented for comparative analyses.

In recent times, in the academic field related to the training of L2 pronunciation teachers, the already existing gap between fundamental research and the application of its results in the classroom has widened. In some degree, this has been a consequence of that training being focused on methodological aspects more than on the intrinsic knowledge of the subject to be taught. In this article, on the basis of two concrete examples, the need for keeping pronunciation teachers permanently informed about the findings of the basic research in phonetics / phonology is defended as a means to achieve a very fruitful interaction between the two sides, theoretical ad applied, of the discipline.

The description of intonation is particularly difficult for spontaneous speech, due to the extreme variability of data. This forced many phoneticians to describe intonation through the use of sophisticated statistic tools, whereas phonologists convinced themselves to use reductive transcription tools such as ToBI. We show here that intonation models built for read speech data can be used successfully for the analysis of spontaneous speech, which after all uses the same linguistic code as read speech production. The experimental data presented illustrate the diversity of speaker strategies to orally structure their sentences, still using the same prosodic rules specific to French as in read speech conditions.

Theoretical work in syntax is based mainly on written language. This means that theories of first language acquisition such as the Chomskyan one are based on a model of syntax relating to written language. But children acquire spontaneous spoken language first and only later learn written language. In spite of recent intensive research on spontaneous spoken language even analysts who distinguish planned and unplanned language and different dimensions of syntax and lexis fail to recognise the extent of the syntactic differences between formal written language and spontaneous unplanned spoken language. Ong proposed the label 'magnavocabulary' for the store of lexical items to be found in written texts composed in any language with a long written history. 'Magnasyntax' is proposed as a label for the parallel syntactic phenomenon; constructions from written texts are recorded in grammars and can be used/revived by writers and analysed by grammarians. Much syntactic theory focuses on magnasyntax. This focus on magnasyntax and the failure to distinguish constructions from different periods of a given language and from different genres is illustrated in two pieces of work on English syntax, Van der Auwera's discussion of whether 'that' introducing relative clauses is a complementiser or pronoun and Quirk and Greenbaum's discussion of IT clefts.