پیش بینی لکنت زبان از پیچیدگی آواشناختی در آلمان

This study investigated how phonetic complexity affects stuttering rate in German and how this changes developmentally. Phonetic difficulty was assessed using Jakielski’s index [Motor Organization in the Acquisition of Consonant Clusters, Dissertation/Ph.D. Thesis, University of Texas Austin, 1998] of phonetic complexity (IPC) in which words are scored on eight different characteristics. Stuttering rate was not related to IPC score for German function words, as previously shown for Spanish and English. Significant correlations between stuttering rate and IPC score were found for content words for children over the age of six and adults. It was also found that German content words have a higher mean IPC sum compared to their English counterparts. There was a bigger difference in IPC score between fluent and stuttered words in German than in English. Factor 5 (word shape) influenced stuttering rates in both German age groups. This has also been found for Spanish but does not apply to English. Educational objectives: The reader will be able to: (1) describe a method to measure phonetic complexity and how this affects stuttering rates for words of different grammatical classes; (2) explain why this method is suitable for different languages and age groups; (3) detect which phonetic characteristics have most impact on different age groups in English and German; (4) assess possible theoretical reasons for these findings.

Stuttered events do not occur at random points in utterances. In particular, their position is constrained in part by the linguistic properties of the segments that make up the utterance. There is some variation across ages both in terms of the type of stuttering events that occur and where the different types of stuttering events are positioned relative to the linguistic units that give rise to them. The description of stuttered events, developed in work on adults, needs to be refined to reveal the relation of these events to the points where linguistic difficulty within an utterance is high and how this varies over age groups. Johnson and co-workers’ (1959) list for characterising stuttered events in adult speech is used as a starting point. The events on this list are: (1) interjections of sounds, syllables, words and phrases; (2) word repetitions; (3) phrase repetitions; (4) part-word repetitions; (5) prolonged sounds; (6) broken words; (7) revisions; and (8) incomplete phrases. It is difficult to specify what linguistic characteristic led to event types 7 and 8 and how many words these events affect, so they are often not included in stuttering assessments. Howell, Au-Yeung, Sackin, Glenn, and Rustin (1997) developed a parser for stuttered speech to remove them, leaving events of types 1–6. Howell (2004) has advocated that the first three categories should be grouped together (all involve hesitation or repetition of whole words, which are termed generically ‘stalling disfluencies’). Howell (2004) also suggests that the remaining three categories should be grouped together as they involve breakdown within a word (‘within-word stutterings’). One thing associated with the change in stuttering events over ages, is that whereas within-word stutterings occur most often on content words, stalling disfluencies occur on or around the phonetically simpler function words (Howell, Au-Yeung, & Sackin, 1999). Content words are nouns, main verbs, adjectives and adverbs that constitute an open class of words that expands as new words are added to a speaker’s lexicon (see Hartmann & Stork, 1972; Quirk, Greenbaum, Leech, & Svartvik, 1985, for basic definitions). Function words are the remaining words (articles, pronouns, prepositions, conjunctions and auxiliary verbs) that are a closed class of words that is not added to once the grammar of a language has been established. A second observation, made by Bloodstein (2002), was that function words that are repeated by children, are often produced fluently as in the utterance ‘his … his … his … strawberry.’ These two facts suggest that it is unlikely that there is anything inherently wrong with preparation of the function word, so repetition or hesitation around them is not determined by difficulty in preparing these words for output. Howell, 2002 and Howell, 2004 suggests stalling delays the attempt at a subsequent content word (content words would be difficult to prepare as they are more likely to include phonetic structures with complex characteristics). Stuttering on content words, unlike stalling that involves whole function words, affects the initial parts of these words alone. This is consistent with the view that the content words are not completely prepared. Howell, 2002 and Howell, 2004 EXPLAN theory maintains that the change from stalling to producing within-word stuttering, reflects a change from delaying before a content word to attempting the content word before the speaker is ready. EXPLAN regards the content word as the locus of difficulty at all ages though such difficulty precipitates different stuttering event types at different ages (stalling at an early age, within-word stuttering in adulthood). Howell’s taxonomy that separates stalling and within-word stutterings differentiates what Wingate (2002) considers to be true signs of stuttering (the within-word stutterings) from those events he would not consider to be characteristics of stuttering (the stalling disfluencies). Wingate’s point of view reflects the fact that he has worked almost exclusively with adults who stutter who show a preponderance of within-word stutterings (conversely stallings predominate in the speech of children who stutter). Wingate (2002) argues that metrical influences specify the locus of complexity in utterances that have within-word stutterings. In particular, he suggests that the stuttering problem arises at the onset–rhyme transition, and the difficulty at this point is compounded when the word carries stress (Wingate, 1984, Wingate, 1988 and Wingate, 2002). Howell, 2002 and Howell, 2004, in contrast, has focussed on how phonetic structure within syllables and words determines how difficult a word is. As indicated earlier, phonetic complexity in English and many other languages depends on word type to some extent. Consequently, these word classes should be examined separately to establish any relation to stuttering (Howell & Au-Yeung, 2002; Howell et al., 1999; Howell, Au-Yeung, & Sackin, 2000). Phonetic influences are not ruled out by Wingate, and metrical influences are not ruled out by Howell, as factors leading to within-word stutterings even though these authors place the emphasis on different factors. Thus Wingate (2002) when commenting on the notion of whether particular phones cause difficulties for speakers who stutter, indicates that such influences can be subsumed under syllable constituency at word onset. Wingate’s notion would include phonetic variables like whether a word includes a consonant string at onset and the manner of the consonants in these strings (both of which are factors that Howell et al., 2000, have examined). Wingate (2002) also emphasised that his studies that indicate that stress is an important determinant of stuttering, used somewhat contrived utterances involving English samples in which function words were stressed. It is unusual for function words to be stressed in English (whereas English content words are frequently stressed). Use of such artificial material, may prohibit generalisation of the results to more typical speech. Howell (submitted), on the other hand, has conducted a study to determine whether metrical, as well as phonetic, factors affect stuttering in natural (spontaneously produced) material. English was not used as it is difficult using this language to dissociate the influence of syllabic and metrical factors, given that content words tend to weight highly on indexes of phonetic complexity and stress is also carried almost exclusively on these word types. Spanish, on the other hand, has stressed function words, so stress can be dissociated to some extent from phonetic factors associated with lexical word class. In an analysis of Spanish speech, Howell (submitted) found both phonetic and metrical factors are important, and independent, determinants of stuttering in adults who stutter: The importance of phonetic factors was indicated by the fact that non-stressed content words had higher stuttering rates than non-stressed function words. The importance of stress was indicated by the fact that stressed function words had higher stuttering rates than non-stressed function words. Though phonetic or metrical difficulty can precipitate within-word stuttering, on content words, these sources of difficulty do not account for why stalling occurs on function words. Wingate’s position where he dismisses the latter class of events as characteristics of stuttering avoids the issue of specifying whether and how linguistic properties lead to stalling and why stalling changes to within-word stuttering with age. This stance is reasonable if the speech of adults who stutter alone is considered, as within-word stuttering predominates in these speakers (Wingate, 2002). However, this is more problematic for those researching into childhood stuttering as children who stutter show high incidences of stalling (Conture, 1990). Understanding the change from stalling to within-word stuttering over age groups might provide important clues about why stuttering persists in some speakers. The current study examined whether phonetic properties determine which words are stuttered by German speakers who stutter. Metrical factors are regarded as affecting stuttering on content words in the same way as they do in English. They are not examined in this study. They are treated as a random-effect factor within each word class that is not linked to words with particular types of phonetic difficulty. Ages of the speakers examined ranged from 2 years to adulthood, and function and content words were analysed separately for all age groups. In the following, (1) details of different phonetic metrics that have been used recently and justification for the one selected for this study are given, (2) reasons why an analysis of German is informative are presented, and (3) the patterns expected for German speakers who stutter of different ages are outlined. There have been two recent attempts to quantify the phonetic difficulty of words for English that can be used to investigate how well these indices of complexity predict stuttering. The first index was derived by Throneburg, Yairi, and Paden (1994). One factor they included was whether consonants appeared early or late in development. They selected nine consonants that Sander (1972) showed are acquired late in development (the late emerging consonants, LEC). They also assessed whether a word had a consonant cluster (CS) and whether words had more than one syllable (MS). All factors were examined for all words (function and content words together) and CS and LEC factors were scored when they appeared at any position in a word. Throneburg et al. (1994) reported that for all words these three characteristics had no effect on stuttering rates of pre-school children who stutter. However, Howell et al. (2000) analysed function and content words separately, given that stuttering may have different roles for each of these word classes. They also investigated Throneburg et al.’s (1994) factors when they appeared in initial position in an utterance (the position that leads to the majority of instances of stuttering). They reported, for adults, that when phonetically complex material occurred in initial position in content words (but not function words), there was an increased incidence of stuttering compared with the function words that were phonetically simpler. The second metric to specify the level of phonetic difficulty of words was developed by Jakielski (1998). Her index of phonetic complexity (IPC) was motivated by MacNeillage and Davis’ (1990) frame/content hypothesis about how speech is acquired. ‘Frames’ are the rhythmic oscillations of the mandible in early babbling. When the infant develops control over the articulators the ‘content’ emerges, that is more variation in production of segments as children change manner, height and place of their articulations. Jakielski (1998) developed measures for eight phonetic factors (indicated in Table 1) using MacNeilage and Davis’ framework and specified how these can be combined to yield a metric of difficulty. Phonetic properties that occur in early development are deemed to be easy and receive no score for difficulty (zero) whereas those properties that do not occur in the babbling stage are considered difficult and whenever one of these phonetic attributes occurs, it is given a difficulty score of one point. An overall IPC score is then calculated by adding up the scores on the eight separate factors. The attributes that score a point and those that do not for the eight individual factors are given in Table 1.