Phoneme Manner Types and Function and Content Words as Biomarkers for Different Types of Stutters in Speakers Who...

This study addressed whether or not manner of phonemes at the onset of function and content words is linked to different types of stutter.

Sixty-six spontaneous speech samples from 22 participants (three recordings per participant) were employed. These were annotated with types of stutter (dependent variable), word type (function versus content). Phoneme manner at word onset was available in the transcriptions. The first analysis looked at associations between types of stutter and types of words. The second analysis looked at individual onset phoneme manner influences on types of stutter separately for function and content words. Both analyses employed multinomial logistic regression, obtained predictive probabilities and reported permutation tests to assess statistical significance, and robustness. A separate set of data was used to replicate the findings.

In analysis one, whole-word repetitions (WWRs) were predominantly associated with function words whereas the other types of stutter (prolongations, part-word repetitions, and word breaks, all of which mainly occur at word-onsets) were associated with content words. In analysis two WWRs on function words were mainly associated with vowel onsets. Prolongations, part-word repetitions and word breaks on content words were associated with several consonant manners at onset. The main patterns were that prolongations occurred with continuant manners (primarily fricatives) whereas part-word repetitions and breaks occurred mainly with obstruent manners (primarily voiceless plosives and affricates).

WWR happen mainly on function words whereas the other types of stutter occur on initial parts of content words. The analysis protocol used in analysis two (target-and-type-maps) provides a framework for analysis of stuttering in languages that do not have separate function words and for distinguishing conditions that have some similarities to stuttering (e.g. word-finding difficulties).

What is already known on this subject Past work has shown that stutters tend to occur on different word types with children who stutter showing higher incidence on function words whereas adults who stutter have higher rates on content words. A number of other formal and usage properties of language affect stuttering rate. Examples of the factors that have such impacts include phonological properties at word onsets (formal) and word frequency (usage). What this paper adds to existing knowledge Manner (continuant versus obstruent) is one language factor not explored to date which could be linked to stuttering. Manner may improve understanding of the distribution of stuttering types particularly when stuttering continues into adulthood. Due to the relatively small number of function words and the fact that they use a limited number of manners on onsets, function and content words were analysed separately. Also, given that stutter types are distributed unevenly across word types, analyses were conducted on separate stutter types (whole-word repetitions, part-word repetitions, prolongations and breaks within words) on a group of children aged between eight years and teenage. Whole-word repetitions occurred predominantly on function words that started with vowels. Part-word repetitions and breaks occurred on content words that started with phonemes with obstruent manners. Prolongations occurred on content words that started with phonemes with continuant manners. The large proportion of whole-word repetitions on function words in the speech of some young children who stutter reported by other authors may suggest that the trajectory they follow may be one that facilitates cessation of stuttering. What are the potential or actual clinical implications of this work? The results suggest that a generative mechanism leads to symptoms of different types due to the properties of the language currently attempted. For example, whole-word repetitions happen on function words, prolongations arise when content words start with a continuant phoneme whereas part-word repetitions occur when content words start with an obstruent). The findings have several important clinical implications for assessment, including diagnosis, of stuttering. The work provides clinicians with a way to examine disfluency patterns across word types during assessments (perhaps using the type-and-target maps introduced in this paper). Examining stuttering patterns in this way may be particularly important when assessing spontaneous speech samples, since focusing solely on overall stuttering frequency risks overlooking the distinction of stutter types occurring on different word types. Applications with important potential clinical implications (to be tested) include stuttering cessation, assessment of stuttering patterns in monolingual, bilingual and multilingual children who speak languages other than English.