For some speakers of English, realization of the fricative in the consonant cluster /stɹ/may be undergoing a sound change in progress. Previous reports suggest that /s/ is being realized as the postalveolar fricative [ʃ], with the approximant /ɹ/ as a possible trigger. Research in sociolinguistics also indicates that there may be various intermediate stages of the change, falling somewhere between a typical /ʃ/ and /s/. However, information about the precise phonetic characteristics of these novel variants is still sparse, and existing sociolinguistic studies have not compared speakers’ realization of /stɹ/ to their typical /s/ and /ʃ/ productions. This paper uses the acoustic measurement of spectral peak to compare ten English speakers’ productions of the onsets /ʃ/, /ʃɹ/, /stɹ/, and /s/. The findings suggest that the majority of tokens produced by these speakers fall within their normal range for /ʃ/, with only limited evidence of intermediate forms. It is proposed that spectral peak measurement seems to be a stable method for use in future research categorizing variants of /stɹ/. However, this should include comparison to subjects’ normal /ʃ/ productions.

This paper presents some findings from a case study of repair sequences in conversations between a dysarthric speaker, Chris, and her interactional partners. It adopts the methodology of interactional phonetics, where turn design, sequence organization, and variation in phonetic parameters are analysed in unison. The analysis focused on the use of segmental and prosodic variation found during attempts by Chris to repair a previously identified trouble source. The results indicate that trouble sources were extremely common in the recorded conversations, but that repair attempts were almost always communicatively successful. Analysis of the fragments revealed that repair sequences are often collaborative achievements, with the participant's conversational partners signalling the specific trouble source within a turn, or providing feedback about repair attempts. It was also observed that successful repair attempts were not always simple repetitions of the trouble source, but varied in other linguistic areas. It is suggested that intelligibility repairs should be studied using both experimental and qualitative methods.

The importance of the use of narrow phonetic transcription in transcribing a variety of speech disorders is emphasized. This point is illustrated with clinical data from the authors' own research. The examples used are the transcription of a severely disfluent client, a child with progressive hearing loss, repair sequences in dysarthric speakers, a child with idiosyncratic velar articulations, and an adult with progressive speech degeneration. The use of the extended International Phonetic Alphabet (IPA) for the transcription of disordered speech and the Voice Quality Symbols (VoQS) system for transcribing voice quality is introduced and illustrated in the examples provided. The need for training in narrow transcription is discussed as part of a combination of impressionistic and instrumental description techniques.

Conversational repair is a commonly studied aspect of both normal and disordered human interaction. It has the potential to reveal a great deal about how speakers deal with problems in naturally occurring talk-in-interaction. However, the extent to which repair has been defined, classified, and then compared to such phenomena as repetitions, slips of the tongue, pauses, and fillers has not always been altogether consistent. This paper discusses the use of the term repair in a variety of disciplines and proposes a taxonomy of repair types. It is suggested that this will facilitate the quantification of repair in corpora of disordered conversation, allowing for comparisons of frequency counts across speakers, disorders, and time points.

Eight children aged 4;1-8;1 and their primary caregivers participated in a study designed to evaluate their use of the onset cluster /str-/ in both read and conversational speech. The cluster is currently undergoing a reported sound change in many varieties of English, with the initial /s/ being retracted to [ʃ]. The study compared the initial fricative of the cluster in both the children and their mothers. Acoustic analysis was carried out in order to categorize tokens as either [s] or [ʃ] using spectral peak analysis. Results found that children as young as 5;1 were starting to exhibit the usage patterns of their mothers. The distribution of the novel variant suggests that the children may be learning this form via a process of lexical diffusion rather than by rule. Implications for the study of dialect acquisition and phonological acquisition in general are discussed.