Comparison of Exchange Patterns of Stuttering In Spanish and English Monolingual Speakers and a Bilingual Spanish-English Speaker
P. Howell, L. Ruffle, A. Fernandez-Zuniga, R. Gutierrez, A.H. Fernandez, M.L. O’Brien, M. Tarasco, I. Vallejo Gomez & J. Au-Yeung
University College London, 26 Bedford Way, London WC1H OAP
Samples of the spontaneous speech of monolingual Spanish speakers who stutter of different ages and a Spanish-English bilingual speaker who stutters were analyzed. The patterns of stuttering in these samples were examined to establish how they compare to those reported to occur for English. The characteristics examined for the monolingual speakers were the difference in stuttering rate on function and content words over ages (Howell, Au-Yeung & Sackin’s 1999 exchange relations). Young Spanish speakers showed a higher rate of function than content words whereas older speakers showed less of a difference (as reported previously for English). Stuttering rates on function and content words in the two languages of a bilingual speaker were examined. The speaker showed a more adult-like pattern in his dominant language (Spanish) but a more child- like pattern in his second language (English). The results are discussed in terms of the implications for theory, diagnosis and treatment of the disorder.
To start to establish what patterns might show differences between languages, first note that the most apparent feature of stuttered speech is that relatively fluent stretches of speech are interspersed with episodes where fluency fails. The types of disfluency at points where fluency fails can be divided into stalling and non-stalling disfluencies (Howell & Au-Yeung, 2002). Stalling disfluencies involve some interruption to the timing of speech, which is manifest as whole word repetition (either a single word or short phrase), or pauses (filled or silent). Non-stalling disfluencies involve breaks within words such as prolongation (e.g. “sssister”), repetition of part of the word (e.g. “s.s.sister”) and complete stops, often around a syllable juncture, within the word (e.g. “di-nosaur”). One fact to noteis that neither class of disfluency (stalling nor non-stalling) is exclusive to speakers who stutter. However, non-stalling disfluencies are relatively rare in fluent speakers whereas stalling disfluencies occur frequently (Howell et al., 1999). Thus, fluent speakers often produce utterances like “and, and, I fell, in the, in the par “. A second fact is that the two types of disfluency usually occur on different types of word: Stallings occur on or near function words and non-stalling on the first part of Content words. (Function words comprise words such as articles, prepositions and conjunctions that have a grammatical role, whereas content words comprise words that have a semantic role, such as nouns and verbs.)
Non-stalling and stalling disfluencies have both been described as part of the pattern of stuttering for English. Other researchers consider that only, what has been called here, non- stalling disfluencies should be regarded as true stuttering (Wingate, 2002) and, as this class occurs infrequently in people who do not stutter (see above), they are a good indicator of the disorder. One reason for considering stalling disfluencies are stutters is that most, if not all, children who present with, and are subsequently treated at clinic for, the disorder show a high proportion of such disfluencies (Bloodstein & Gantwerk, 1967; Bloodstein & Grossman, 1981). It would be foolhardy to accept that these disfluencies have nothing to do with stuttering, given that the professionals categorize children showing these patterns as stuttering. Besides that, out own EXPLAN theory and data obtained from tests to assess it indicate that stalling and non-stalling disfluencies arise out of the same determinants (Au-Yeung et al., 1998; Au-Yeung et al., 2003; Howell, 2002; Howell & Au- Yeung, 2002; Howell et al., 1999). To explain this point more fully, some background to our theory is needed.
Howell et al. (1999) showed that children who stutter produce more stalling, than non-stalling disfluencies whereas adults who stutter do the opposite. The change from stalling disfluencies to non-stalling disfluencies with age for people who stutter (Howell et al., 1999) appears to be a sign associated with persistence of the disorder. Stalling disfluencies do not cause fluent speakers a problem at any age, so children who stutter who show this pattern may have a high chance of recovery. Conversely, the change to non-stalling disfluencies may explain why adults persist in the disorder.
An immediate question that arises is if stuttering is a response to difficulty in speech, and stalling is a characteristic of stuttering, is why does stalling happen on the simpler function words in children who stutter? Howell et al.’s (2000) answer was that stuttering is always a result of difficulty in speech and these points of difficulty occur on Content words in English. Thus, in an utterance like “I split it”, the point where difficulty is focussed is the content word (“split”), whatever the age of the speaker. This word is difficult as it starts with a consonant string, so would require more time to prepare for production than the surrounding function words. A young speaker who stutters will stall on the word prior to “split” to gain time and say, for example, “I, I, split it” or “I [pause] split it. Though these disfluencies arise because of the problematic content word, stalling occurs prior to the difficult word. One prediction that has been confirmed recently for speakers who stutter (Howell & Ladd, in preparation), is that when stalling disfluencies occur, the content word that follows is phonetically more difficult than a content word spoken fluently (i.e. not preceded by stalling disfluencies). An implication of the observation that stalling occurs in normal fluent speech development is that this way of dealing with the difficulty coming up is not problematic. Intervention is not called for in children suspected of stuttering if they are producing stalling responses.
The exchange data show that adults who stutter tend not to stall, but to attempt the content word before they are completely ready to produce it. They are then disfluent on the difficult content word itself. Adults who stutter are effectively changing the way they respond to the difficulty compared with fluent speakers and young speakers who stutter. When the adult speakers who stutter attempt the content word, the disfluency can only involve the section that is prepared (the first part), so this results in prolongation (sssplit) repetition (s.s.split) or a break on the initial part of the content word (s-plit). The changeover from stalling to non-stalling disfluencies is complete around age 12 (Howell et al., 1999). The non-stalling disfluencies are problematic insofar as they tend to be associated with forms of stuttering that persists into adulthood and, as such, are potential signs that the problem is worsening. A child who is being monitored who shows this pattern of disfluency may require therapy. The interpretation of developmental changes that has just been outlined is the essence of Howell and Au-Yeung’s (2002) EXPLAN theory as it applies to monolingual English speakers.
Does this pattern occur in Spanish where properties of function and content words differ from English (stress and phonetic)? This is tested in Experiment 1, and shows the exchange pattern occurs for Spanish as well as English. Experiment 2 looks at whether the disparity in stuttering on function and content words can co-occur in each language of a bilingual. Then implications for theory, diagnosis and treatment are drawn in the final discussion.
- Experiment 1
Forty-six monolingual Spanish speakers from various parts of Spain participated in the study. Their speech problem was confirmed by a speech pathologist who also confirmed that they did not have any other language disorder. The participants ranged in age from 3 to 68 years and there were 10 females and 36 males. The participants were divided into five age groups (Gl-G5) in the following ranges: Gl:3-5 year (n = 7), G2:6-9 years (n = 11), G3:l0-ll years (n = 10), G4:l2-l6 years (n = 9) and G5:20-68 years (n = 9). The division into age groups is similar to that used in Howell et al. (1999).
Spontaneous conversational speech samples between the participant and his/her therapist were obtained. The recordings were usually conducted in a clinic although there were six cases where the recording had to be made in the participant’s home. Five of these were in the oldest age group (the remaining one was in the 10-11 year age group, G3). The average length of sample per speaker was 667 words. The speech was transcribed by a native Spanish speaker with nine years experience in transcription. The convention for the transcriptions was broad phonetic in fluent regions and narrow phonetic in the disfluent regions (see Kadi-Hanifi & Howell, 1992). The accuracy of the transcriber has been assessed by comparing her transcriptions with those of another trained transcriber on seven randomly selected recordings. Agreement on word type was 95.10% (Cohen’s Kappa = 0.90). Agreement on fluency was 96.65% (Kappa = 0.93) and for PW segmentation 97.10%. All scores represent excellent levels of agreement.
Segmentation into phonological words (PWS) based on lexical status
Only function words before a content word can be used for stalling. The position of function words relative to content words was determined by segmenting the speech into PW units. PWs were introduced by Selkirk (1984) and are referred to both as phonological, and as prosodic, words (Selkirk used the latter in her 1984 work). PWs consist of a single content word and optional function words that precede (that can be used for stalling) and follow (that cannot be used for stalling) the content word. Her semantic rules were extended for segmenting speech as described by Au-Yeung et al. (1998) and Howell et al. (1999).
Results are presented in Figure 1. Over all age groups 11.5 % stutters occurred on initial function words 4.9 % on content words and 0.14% on post-content word function words. The discrepancy between stuttering rate on pre- and post-content word function words supports the idea of the stalling role of the function words that precede the content word. The function words after the content word cannot fill this role, so disfluency rate is low. Very little (2.54%) stuttering occurred on an initial function word and the content word that follows it in its PW. This suggests that function (stalling) and content (non-stalling) word stuttering are complementary. One explanation of this complementarity is that stalling disfluency prevents non-stalling disfluency and that if a speaker does not stall, then they produce stuttering on content words (i.e. the essence of the EXPLAN account).
The next issue examined is the relationship between function word and content word stuttering rates over age groups. As Howell et al., (1999) reported for English, there is a drop of function word stuttering rate and an increase of content word stuttering rate over age groups. The changes are gradual in Spanish (the same applies to English). One feature of note for the Spanish data is that stuttering rate on function words is higher than the rate on content words for all age groups. The stuttering rate on function words is higher than that found for English (Howell et al., 1999) and German (Dworzynski et al., submitted). The differences might be due to different phonetic complexity between Spanish, on the one hand, and English and German on the other. It is important to note, however, that though there is this difference between languages, an exchange relation still occurred for Spanish.
The results on Spanish are qualitatively similar to those of English. 1) Thus, when there is stuttering on a function word, the function word is in the initial position in a PW where it can have a delaying role. 2) Also, such stalling disfluency seems to prevent stuttering on the following content word, and if there is no stalling content word disfluency ensues. 3) The alternatives in 2) (stalling and non-stalling) tend to be used predominantly by young and old speakers respectively (i.e. an exchange relation). The main difference between English and Spanish monolinguals is that Spanish speakers maintain a higher rate of function word disfluency through to adulthood.
- Experiment 2
For current purposes, the interesting cases are those where the speaker does not have the same ability in the second language as the first (not equivalent mastery). Total mastery relative to peers is assumed in the speaker’s first language. Fluency assessments in the second language are then based on function and content word stuttering rates and these are compared with the same measures in the speaker’s first language. What would be expected to happen in a bilingual speaker who is disfluent to different extents in two languages? EXPLAN theory makes some rudimentary predictions what stuttering patterns may and may not occur in such cases. The assumption is that bilingual speakers will be like monolingual peers Who stutter in their dominant language, but more like younger children in the language they are less fluent in. An adult bilingual speaker more developed in the first, than the second, language would be likely to have non-stalling disfluencies in the first language and the second language could either have stalling disfluencies or no disfluencies at all. EXPLAN predicts that stalling disfluencies in the first language and non-stalling disfluencies in the second language would be unlikely to occur. The literature is examined next to see whether reports about bilingual speakers who stutter correspond with these predictions (later data on an imbalanced Spanish-English bilingual speaker is reported).
There is no data that uses function and content word stuttering patterns to check these predictions, though there are suggestions in the literature that most cases fit with non-stalling in the dominant language and something more akin to stalling in the non-dorninant language. The majority of studies report for most speakers that one language is affected more than another (though this is of limited value for current purposes if there is no indication whether this applies to stalling and/or non-stalling disfluencies). Related to this, a problem is that it is not always possible to determine what events were counted as stutterings to allow division of stutterings into the classes of disfluency distinguished above. Studies by Jayaram (1983), Bernstein Ratner & Benitez (1985) and Nwokah (1988) all showed that stuttering was observed to be more frequent in the dominant language. Only one study shows a pattern where the frequency of stuttering was higher in the less proficient language (an English - Afrikaans speaking adult, J ankelowitz & Bortz, 1996). In this case, the speaker was slightly more proficient in English than Afrikaans (94 and 90%, as assessed by cloze tests in each language). Inspection of their Figure 1 indicates that the higher disfluency rate in Afrikaans mainly occurred on interjections, phrase and word repetitions, all of which would be classified as stallings. Though the frequency of stuttering is higher in Afrikaans which is the less used language for this speaker, the forms of the stutterings in Afrikaans that give this imbalance are of the simpler pattern. This is consistent with the idea that the less advanced a second language is, the more likely it will show the function word-stalling pattern seen early in development. A study by Shenker et al., (1998) looked at disfluency patterns of an English - French speaking infant who mainly used English in the home. More word repetitions were observed in French but more part-word repetitions in English which is in line with the above analysis of the Jankelowitz and Bortz (1996) study which was interpreted as showing that stallings are more frequent in the less-used language. The studies have not examined stuttering on function and content words which is the best documented change over ages (Bloodstein & Gantwerk, 1967; Bloodstein & Grossman, 1981) and is :he basis of exchange plots (Au-Yeung et al., 2003; Dworzynski et al., submitted; Howell et al., 1999). An analysis of the speech patterns of a bilingual speaker who stutters is reported below. Spanish is more developed than English in this speaker. Stutterings are characterized as occurring on function or on content words (as in Experiment 1).
Participant The speaker was a Spanish-English bilingual speaker who lived in Spain. His father is Spanish and mother English (mother conversed with the boy in English). He also spent large amounts of time in England with family. His English was fluent, though not as fluent as his Spanish. At the time of recording he was 11 years 9 months.
Procedures for analysis. These were the same as for Experiment 1.
Table 1 shows the results for conversations and monologues in Spanish and English. The rows along the left indicate total number of words in each sample, the number of function words, the number of content words, percentage of function words stuttered, percentage of content words stuttered and the signed difference for each material type and language of function stuttering percentage - content stuttering percentage.
Table 1 Spanish English Conversation Monologue Conversation Monologue
There are three main features apparent in this table. 1) The absolute stuttering rates are higher overall in Spanish. Stuttering rates for all words were 29.5% (conversation) 38.5% (monologue) in Spanish and 19.7% (conversation) 18.4% (monologue) in English. This may indicate that the speaker has more of a problem in Spanish than English though (as later features show), the pattern of these stutterings also needs examining. 2) The differences in percentages of words stuttered across function and content words was higher for monologues than in conversation for both languages. 3) Most importantly, for both types of material, there is less of a difference between function and content word stuttering in Spanish than for corresponding material in English.
Finding 3) shows that this speaker produced a higher proportion of content, than function, word disfluencies in Spanish than English for both types of material. The higher rate of content word disfluencies in Spanish is consistent with the hypothesis that this speaker shows more non-stalling disfluencies in Spanish. To put it another way, this indicates this speaker’s problem is more severe in Spanish than English (more of an imbalance in favor of function words in English). Rephrased in terms of the earlier discussion of how function/content word patterns change over ages, the speaker appears to be producing more of an adult form of stuttering in Spanish than in English, though it is probably more appropriate to refer to the developmental patterns, in the case of bilingual speakers, in terms of fluency in a language acquired early versus late.
3.3 Discussion How is it possible for a speaker to maintain two different stuttering patterns, one in each language, and what does it tell us about the developmental age changes that have been observed in function/content word disfluencies? The data of the Spanish-English bilingual speaker show that different ways of dealing with disfluencies can be used by the same speaker when using different languages. Whatever structural or functional problem gives rise to stuttering, it must be adaptable to which language the speaker uses and his/her fluency in that language. EXPLAN offers an account of stuttering that would just need modifying to refer to stuttering patterns being determined by whether a language is acquired early or later by a speaker, rather than using the speaker’s chronological age as a criterion (the latter would still apply in monolingual speakers). It is possible that linguistic or motor accounts of stuttering could be adapted to account for why different patterns can predominate in each language of a bilingual speaker. Linguistic differences could, for instance, be attributed to language-specific processing though this seems unlikely to occur. There might be different ways of organizing motor output in languages that a speaker is more fluent in (i.e. the speaker’s first language) that affect language output adversely and ways of organizing output in languages that a speaker is less fluent in (i.e. the speaker’s second language) that assist language output (Klein et a1.,1994).
From a diagnostic point of View, therapists ought to monitor the distribution and types of disfluencies in each language. According to EXPLAN, the language that shows more non-stalling disfluencies is the problematic one. If neither language are showing non-stallings, then speech may need to be monitored in case the speaker changes to this pattern, but no intervention is called for.
The advice is often given to bilingual speakers to stop using the less fluent language until the problem in the first language goes away (van Borsel et al., 2001). There is no evidence for this View and, according to EXPLAN, this might be misleading advice. The complementarity between function (stalling) and content (advancing) stuttering types has been noted and also, the fact that stalling does not seem to be problematic (fluent speakers do it with impunity). Taking these two observations together, encouraging use of stalling may help promote fluency as discussed by Howell and Sackin (2001). In the case of bilinguals who stutter, experience using stalling disfluencies in the second language might generalize to the first language and promote fluency in it.
This work was supported by the Wellcome Trust.
Au-Yeung, J., Howell, P., & Pilgrim, L. (1998). Phonological words and stuttering on function words. Journal of Speech, Language and Hearing Research, 41, l019-l030.
Au-Yeung, J ., Vallejo Gomez, 1. & Howell, P. (2003) Exchange of disfluency from function words to content words with age in Spanish speakers who stutter. Journal of Speech, Language and Hearing Research, 46, 754-765.
Bernstein Ratner, N ., & Benitez, M. (1985). Linguistic analysis of a bilingual stutterer. Journal of Fluency Disorders, 10, 211-219.
Bloodstein, 0., & Gantwerk, B.F. (1967). Grammatical function in relation to stuttering in young children. Journal of Speech and Hearing Research, 10, 786-789.
Bloodstein, 0., & Grossman, M. (1981). Early stutterings: Some aspects of their form and distribution. Journal of Speech and Hearing Research, 24, 298-302.
Dworynski, K., Howell, P., Au-Yeung, J. & Rommel, D. (submitted). Stuttering on function and content words across age groups of German speakers who stutter. Journal of Multilingual Communcation Disorders.
Howell, P. (2002). The EXPLAN theory of fluency control applied to the Treatment of Stuttering by Altered Feedback and Operant Procedures. In E. Fava (Ed.), Current issues in linguistic theory series: Pathology and therapy of speech disorders (pp.95-118). Amsterdam: John Benjamins. 422 Theory, research and therapy in fluency disorders
Howell, P. & Au-Yeung, J. (2002). The EXPLAN theory of fluency control and the diagnosis of stuttering. Pathology and therapy of speech disorders. In Current Issues in Linguistic Theory series. Volume Editor E. Fava. Amsterdam: John Benjamins. Pp 75-94.
Howell, P., Au-Yeung, J. & Sackin, S. (1999). Exchange of stuttering from function words to content words with age. Journal of Speech, Language and Hearing Research, 42, 345-354.
Howell, P., Au-Yeung, J ., & Sackin. S. (2000). Internal structure of content words leading to lifespan differences in phonological difficulty in stuttering. Journal of Fluency Disorders, 25, 1-20.
Howell, P., & Saekin, S. (2001). Function word repetitions emerge when speakers are operantly conditioned to reduce frequency of silent pauses. Journal of Psycholinguistic Research, 30, 457-474.
Howell, P., & Ladd, J. (in preparation). Phonetic difficulty and stuttering.
Jankelowitz, D.L., & Bortz, M.A. (1996). The interaction of bilingualism and stuttering in an adult. Journal of Communication Disorders, 29, 223-234.
Jayaram, M. (1983). Phonetic influences on stuttering in monolingual and bilingual stutterers. Journal of Communication Disorders, 16, 287-297.
Klein, D., Zatorre, R. J ., Milner, B., Meyer, E., & Evans, A. C. (1994). Left putaminal activation when speaking a second language: Evidence from PET. Neuroreport, 5, 2295-2297.
Nwokah, E. E. (1988). The imbalance of stuttering behavior in bilingual speakers. Journal of Communication Disorders, 13, 357-373.
Selkirk, E. (1984). Phonology and syntax: The relation between sound and structure. Cambridge, MA: MIT Press.
Shenker, R. C., Conte, A., Gingras, A., Courcey, A., & Polomeno, L. (1998). The impact of bilingualism on developing fluency in a preschool child. In E. C. Healey & H. F. M. Peters (Eds.), Second world congress on fluency disorders proceedings, San Francisco, August 18-22. Nijmegen: Nijmegen University Press.
van Borsel, J ., Maes, E. & Foulon, S. (2001). Stuttering and bilingualism: A review. Journal of Fluency Disorders, 26, 179-205.
Wingate, M. E. (1984). Stuttering events and linguistic stress. Journal of Fluency Disorders, 9, 295-300.
Wingate, M. E. (2002). Foundations of stuttering. San Diego CA: Academic Press.