2003 IFA Congress: Montreal, Canada

Distribution of Disfluencies According to Word Class Categorization in Brazilian Portuguese

Fabiola Juste and Claudia Regina Furquim de Andrade
Department of Physiotherapy, Speech-Language and Hearing Science and Occupational Therapy, School of Medicine, University of Scio Paulo, Rua Cipotanea 5], Cidade Universitciria, Siio Paulo -S.P., 05360-160, Brazil


This study aimed to verify the influence of word class in the speech disruptions of fluent and stuttering children, speakers of the Brazilian Portuguese. Participants of this study were 20 stuttering children (GI) and 20 fluent children (GII), 26 males and 14 females, whose ages ranged from 4.0 to 11.11 years. Speech samples were collected and the distribution of frequency of disfluency were classified by type and grammatical class. The results indicate that both groups present a higher number of speech disruptions in closed class words and for GI a small difference was observed for SLD in open class words.


According to Colburn & Mysak (1982), during the years of language development and acquisition, the young children can present varying degrees of disfluent speech. This variation results from morpho-syntatic-semantic uncertainties and from the maturation of the neuromotor speech processes. Most of the children overcome this period with success (Yairi, 1997; Andrade, 1999). Several studies suggest that the occurrence of speech disruptions is related to different aspects of language formulation (Wingate, 1979, Postma & Kolk, 1993; Bernstein-Ratner, 1995, 1997, 2000; Yaruss, Newman & Flora, 1999).

The influence of word grammatical classes over speech disruptions have been periodically studied for the English language. Wingate, in 1979, presents a brief retrospective about the subject. According to Brown, in a series of research papers published between 1935 and 1945, adults present speech disruptions more frequently in long words, positioned at the beginning of an emission and in content words (nouns, adjectives, verbs and adverbs). In a study with 33 adults stutterers, Wingate found results similar to Brown’s. The subjects presented more speech disruptions in content words, and these disruptions occurred independently of the word’s position in the emission. Speech disruptions in function words (articles, conjunctions, prepositions and pronouns) occur mainly in the first positions of an emission.

For children, unlike the findings observed for adults, the speech disruptions occur more frequently in function words (Bernstein-Ratner, 1997, Au-Yeung, Howell & Pilgrim, 1998, Howell, Au-Yeung & Sackin, 1999, 2000).

In a first study of Au-Yeung, Howell and Pilgrim (1998), the authors found that for children who stutter, speech disruptions occur more frequently in function words, especially when these precede content words. For these authors, content words (open class of words) have a fundamental role in the transmission of semantic information. Function words (closed class of words) lack in semantic content, but exert an important grammatical function. According to the authors, speech disruptions in function words occur when the planning of the subsequent content word is not ready for execution.

In a posterior study, Howell, Au-Yeung and Sackin (1999) found that with age the predominant grammatical class of speech disruptions changes from function words (during childhood) to content words (adolescents and adults).

Although only one Brazilian study exists about the influence of the grammatical classes on the occurrence of PWS speech disruptions, this study (Pereira, 2001) found results similar to those for the English language (i.e. research performed with 21 stutterers, male and female, with ages ranging from 12 to 55 years, points as a result that for this age range speech disruptions occur more frequently in content words). For Pereira, the predominance of speech disfluencies in open class words occurs because in the Portuguese language, words with lexical content are longer than the function words and, in many cases less familiar, making the elaboration of the articulatory level more difficult.

The purpose of this study was to verify the influence of word class in the speech disruptions of fluent and stuttering children, native speakers of the Brazilian Portuguese language. In this study, the adopted terminology will be open class for content words and closed class for function words.

The hypothesis of this study are:
  1. the occurrence proportion of speech disruptions in closed class words is higher than the occurrence of speech disruptions in open class words for both groups (Howell, Au-Yeung & Sackin, 1999)
  2. the occurrence proportion of stuttering-like disfluencies (SLD) and other disfluencies (OD) has different distribution in open and closed class words for GI based on the assumption that the process of language generation interacts with children’s ability to produce fluent speech (Bernstein-Ratner, 1995)
Material and Method

This study had prior approval from the Research Ethics Committee of the Institution (CAPPesq HCFMUSP - protocol no.262/03) and informed consent was obtained from all participants parents.


Participants of this research were 40 children aged 4.0-ll.ll years, 26 male and 14 female, native speakers of the Brazilian Portuguese language, divided in two groups: the research group (GI) consisted of 20 stuttering children (13 male and 7 female) and the control group (GII) consisted of 20 fluent children (13 male and 7 female).

All participants had no history of other communication disorders, hearing, neurological, psychological, or intellectual problems (parents report and examiner observation) at the time of testing.


Table 1. Participant’s distribution and characterization

Classification and inclusion criteria

A child was included in the research group (GI) if he/she met the following criteria:
a) presented fluency scores off the confidence interval for the reference values of the age group according to the Fluency Profile Assessment (Andrade, 2000a’b) a.1) disfluencies types
a.1.1) Stuttering-like Disfluencies (SLD) - sound and/or syllables repetitions, sound prolongations, blocks, pauses and broken words;
a.1.2) Other Disfluencies (OD) - hesitation, interjection, revision, unfinished Words, whole word repetition, segment repetition, phrase repetition.
a.2) speech rate
a.2.1) Words per minute - WPM - represents the index of information production;
a.2.2) Syllables per minute - SPM -â  represents the index of articulatory rate.
a.3) Frequency of disruptions:
a.3.l) Percentage of speech discontinuity - reflects the percentage of the discourse that is disrupted by the total number of speech disfiuencies (SLD and OD);
a.3.2). Percentage of stuttered syllables - reflects the percentage of the discourse that is disrupted by the total number of SLD.


Table 2. Reference values for the Fluency Profile


Table 3. Stuttering severity of GI


Data Collection

The following equipment was used to gather and analyze data: digital videocamera (Gradiente, GC-16C); tripod, microphone (Leson, ML8), videotapes (VHS-C, JVC, TC-30), digital chronometer (Casio, CPW-200), television , videocassette recorder and computer (Toshiba Satellite Pro 425 CDT). A picture stimulus was used to elicit self-expressive speech.


In the clinic, a certified speech-language pathologist trained in the assessment of stuttering administered all formal tests and analyzed all speech samples. During these assessments, all participants were audiotaped using a Panasonic RQ-L309 cassette recorder and videotaped using Panasonic VJ98 videocamera. The camera was focused upon each participants head, arms and upper torso. Data collection sessions lasted approximately 10 minutes. Procedures for gathering spontaneous speech sampling were those standardized by (Andrade, 2000ab). Briefly, each speech sample contained the minimum of 200 fluent syllables/100 fluent words (the Portuguese language is mainly dissyllable).

Audio and videotape

After parents signed the informed agreement protocol, a case history was obtained, followed by a speech screening test and the recording of a self-expressive speech sample elicited by a picture stimulus.

This method for collecting speech samples was selected because of its use during clinical assessments of stuttering and other disorders, when first analyzing cases as well as during treatment follow-up assessments. The use of a visual stimulus to elicit speech monologues allowed data to be collected without the influence of a conversation partner into a dialogue.

Analyses of speech samples followed the methodology described by Andrade (2000a,b). For the categorization of word classes, the speech disruptions were identified and classified according to their grammatical class and later divided in open or closed word classes (open class - nouns, adjectives, adverbs, verbs and numbers; closed class - pronouns, prepositions, conjunctions, articles and interjections). The analyses of the results, according to the formulated hypotheses, did not consider tie total proportion of open or closed class words produced in each speech sample. Only the disrupted words were analyzed.


The obtained results are illustrated by the tables below:


Table 4. Distribution of disfluencies regarding word class


Table 5. Distribution of stuttering-like disfluencies (SLD) according to word class

Table 5 illustrates that the distribution of SLD, according to the word class categorization, was similar for both groups. GI presented a slightly higher number of speech disruptions in open class words when compared to the occurrence of speech disruptions in closed class words. For GII the distribution between speech disruptions in open and closed class words was the same. Comparatively GI e GH are different regarding the number of occurrence of SLD (GI presented 25 times more SLD than GII). These data are compatible with the findings of Yairi (1999).


Table 6. Proportion of SLD in closed and open class words

Table 6 illustrates the proportion of SLD in closed and open class words. For GI the proportion of prolongation in closed class words is higher when compared to the other SLD. Blocks and sound repetition demonstrated the same proportion of occurrence in Closed and open class words. AS for syllable repetition, the occurrence proportion of SLD was low in closed class words.

The same performance is observed for both groups regarding the occurrence of prolongation (higher in closed class words) and of syllable repetition (higher in open class words).


Table 7. Distribution of other disfluencies ( OD ) according to word classes

Table 7 illustrates that the distribution of OD, according to the word class categorization, was similar for both groups (a higher occurrence in closed class words). Comparatively GI.e GI1 are different regarding the number of occurrence of OD (GI presented 2 times more OD than GI1). These data are not compatible with the findings of Yairi (1999).


Table 8. Proportion of OD in closed and open class words

Table 8 illustrates the proportion of OD in closed and open class words. The groups are not proportionally different (for word repetition and revision the proportion of occurrence is higher in closed class words).


It can be observed, according to the design of the research, that the tested hypotheses indicate:
  • Hypothesis 1 - Partially confirmed. Although the results in absolute numbers indicate that for both groups the occurrence of speech disruptions is higher in closed class words, it is observed that for GI the distribution of disfluencies between the word classes is practically the same (proportion 1.2). However, for GI1 the proportion of speech ' disruptions in closed class words is almost twice the proportion presented in open class words.
  • Hypothesis 2 - Confirmed. The occurrence proportion of SLD and OD has different distribution in open and closed class words for GI.
This study is part of more extensive research that plans to approach, besides the analysis of the grammatical aspects on speech disruption, also aspects of prosody and word extension in fluent and stuttering children speakers of the Brazilian Portuguese language.

This is an initial study, descriptive, lacks statistical analysis, and therefore its results cannot be generalized. The final version will have twice the number of participants. Besides that, this is a pioneer study for children - ï¬ uent and stutters - native speakers of Brazilian Portuguese language. All of the found methodological and theoretical support is based in non-Latin languages. For this reason we believe that the closure of this study will be more descriptive than normative. Even though the comparison with other findings is somewhat limited, it is of extreme importance for the speech and language science to study language variables and its influence in the speech and language processing and pathologies.

Andrade, C.R.F. de (1999). Diagnostico e intervengéio precoce no tratamento das gagueira infantis. Carapicuiba, Pro-fono. 112p.

Andrade, C.R.F. de (2000a). Fluéncia (Parte C). In: Andrade, C.R.F. de; Befi-Lopes, D.M.; Femandes, F.D.M.; & Wertzner, H.F. (Org), ABFW- Teste de Linguagem Infantil nas Areas de F onologia, Vocabulcirio, F luéncia e Pragmatica (61-75). Carapicuiba: ed. Pr6-Fono. _

Andrade, C.R.F.de. (2000b). Protocolo para avaliagao da fluéncia da fala. Pro-Fono, 12(2), 131-134

Au-Yeung, J ., Howell, P., & Pilgrim, L. (1998) Phonological words and stuttering on function words. Journal of Speech, Language, and Hearing Research, 41, 1019-1030.

Bernstein-Ratner, N. (1995) Language complexity and stuttering in children. Topics in Language Disorders, 15(3), 32-47.

Bernstein-Ratner, N. (1997) Stuttering: a psycholinguistic perspective. In: Nature and Treatment of Stuttering: New directions, (99-127) Org. Curlee & Siegel. Boston, A1lyn & Bacon.

Bernstein-Ratner, N. (2000) Performance or capacity, the model still requires definitions and boundaries it doesn’t have. Journal of Fluency Disorders, 25, 337-346.

Colburn, N. Mysak, E.D. (1982) Developmental disfluency and emerging grammar I: Disfluency Characteristics in early syntactic utterances. Journal of Speech and Hearing Research, 25, 414-420.

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.

Pereira, M.M.B. (2001). Analise Linguistica da Gagueira. Belo Horizonte (Doctoral theses presented at the Linguistics Department of the Federal University of Minas Gerais, Brazil). 336p.

Postma, A. & Kolk, H. (1993) The covert repair hypothesis: Prearticulatory repair processes in normal and stuttered disfluencies. Journal of Speech and Hearing Research, 36, 472-487.

Riley, G.D. (1994) A Stuttering Severity Instrument for children and adults. Austin, Pro-Ed, 20p.

Yairi,E. (1997) Disfluency characteristics of childhood stuttering. In: Nature and Treatment of Stuttering: New directions, (49-78) Org. Curlee & Siegel. Boston, A11yn&Bacon.

Yairi, E. (1999) Normative disfluency data for early childhood stuttering. Journal of. Speech, Language and Hearing Research, 42, 895-909.

Yaruss, J. S., Newman, R. M., Flora, T. (1999) Language and disfluency in nonstuttering children’s conversation speech. Journal of Fluency Disorder; 24, 185-207.

Wingate, M. E. (1979) The first three words. Journal of Speech and Hearing. Research, 22, 604-612.

This research was supported by a grant from CAPES ( F undaccio Cooerdenaccio de Aperfeigoamento de Pessoal de Nivel Superior). F abiola Juste is the main investigator This presentation is part of a doctorate project.


In preparation for the 2018 World Congress the IFA is implementing Japanese translations of some pages on the site. Choosing Japanese below to see these translations.

Not all pages are translated, but you can use Google translate to see a machine translation using the switch below

Google Translate

Follow the Joint World Congress