2003 IFA Congress: Montreal, Canada

Judgments of Disfluency in a Familiar vs. an Unfamiliar Language

Brian D. Humphrey
Programs in Speech, Language, and Communication Disorders, Nova Southeastern University, 3301 College Avenue, Fort Lauderdale, Florida 33314, USA


This study examined whether bilingual English-Spanish speaking judges may be better at making disfluency judgments in Spanish than monolingual English-speaking judges. Both groups judged a Spanish-language narrative to contain a greater percentage of disfluencies than an English-language narrative by the same speaker. However, neither group identified a significantly greater percentage of disfluencies in the Spanish-language narrative. Implications for treatment and directions for further research are discussed.

  1. Introduction
Although the prevalence of stuttering in many languages and cultures is well documented, there are relatively few investigations of the relationship between stuttering and cultural and linguistic differences (Cooper & Cooper, 1998; Finn & Cordes, 1997, Van Borsel et al., 2001). Studies of the relationship between stuttering and bilingualism are even fewer (e.g. Humphrey, 1999, 2003; Humphrey et al., 2001; Jankelowitz & Bortz 1996; Karniol, 1992; Nkowah, 1988; Roberts, 2002).

Finn and Cordes (1997) suggested that stuttering behaviors and listener reactions to stuttering show significant correspondence across languages and cultures. They stated that they were “aware of no empirical evidence concerning whether or how well clinicians are able to make reliable and valid judgments about stuttering in languages or dialects other than their own” (Finn & Cordes, 1997, p. 225). Van Borsel et al. (2001) asserted that judging disfluencies in other languages and making appropriate clinical decisions about bilingual stuttering demand extraordinary levels of clinical knowledge and skill. Because less than one per cent of clinically certified speech-language pathologists in the U.S. speak more than one language fluently (Holliday, 2001), it is important to study clinical judgments of disfluencies in unfamiliar languages and to identify strategies for monolingual clinicians to assess and treat bilingual people who stutter.

  1. Purpose
The purpose of the study was to examine whether bilingual judges who are familiar with a given language would be better at making clinical judgments about disfluent behaviors in that language than judges who are unfamiliar with that language.

  1. Subjects (Judges)
Subjects were six bilingual English-Spanish-speaking judges and six monolingual English- speaking judges. Prospective judges were graduate students in speech-language pathology who had taken a course in fluency disorders. Using an investigator-designed questionnaire, bilingual English- Spanish speaking judges were assigned to Group 1, and English-speaking judges with very little or no proficiency in Spanish were assigned to Group 2 (Table 1). Those who failed these criteria were excluded from the study.


Table 1. Average Self Ratings of Language Proficiency (I = No Proficiency; 5 = Full Proficiency)

  1. Method
After completing the investigator-designed questionnaire about language proficiency, the judges were instructed on how to identify disfluencies. Then they were asked to view two videotaped narrative samples featuring a disfluent bilingual English-Spanish speaker in order to identify disfluencies in the samples. The English sample and the Spanish sample were each viewed and scored twice. Judgments of disfluencies for the two groups were tallied. The overall percentages of disfluency obtained from these tallies were compared to each other and to the percentages of disfluency determined by two trained graduate students who previously transcribed and scored the same samples (Humphrey, 1999, 2003).

As the judges viewed the samples (Sony SLV N51 video recorder output to a Quasar 13” color television monitor), they heard the audio portion through the left side of a pair of stereo headphones (Sony MDR V100). To record disfluency judgments, the judges pressed a spring-loaded toggle switch. The switch sent a software-generated marker tone (N CH Tone Generator: http://www.world- voices.com/software/nchtone.html) to the right side of the headphones and to the right track of an audiocassette tape recorder (Tandberg TCD310 Mk II). The audio portions of the communication samples were sent to the left track of the audio tape recorder. Finally, each judge completed a questionnaire about of the taped speaker’s fluency and language proficiency.

Each recorded marker tone was tallied as one disfluency judgment. Each group’s percentages of disfluency for the English sample were compared to their percentages of disfluency for the Spanish sample and to the percentages of disfluency determined previously by the two trained transcribers. Test-retest data were also analyzed. The results of the final questionnaire were tallied.

  1. Results
Group 1, the bilingual judges, and Group 2, the monolingual judges, both scored the Spanish sample as more disfluent than the English sample (Table 2). This finding is statistically significant (F=112.595, df=l, 10, p<0.001). It is consistent with the work of the trained transcribers.


Table 2. Percent Disfluent Syllables Across Judges and Language of Sample

The judges’ severity ratings on a clinician-designed scale, completed after Viewing, hearing, and scoring the sample, were consistent with this finding (Table 3). Bothgroups rated the speaker on the videotape to be more disfluent in Spanish than in English.


Table 3. Judges’Averaged Estimations of Severity of Disfluency (I = Fluent, 5 = Severe)

However, the bilingual and the monolingual judges did not identify a significantly different percentage of disfluencies in the English sample or in the Spanish sample (F:1.377; df=1, 10; p<0.268). In this study, familiarity with Spanish made no significant difference when judging disfluencies in the Spanish sample.

The judges identified significantly more disfluencies upon retest (F=10.03l; df=10, 1; p<0.0l). Interestingly, upon retest, the monolingual judges increased their judgments of disfluency to a significantly greater extent than the bilingual judges (F=11.137; df=1, 10; p<0.008) (Table 4).


Table 4. Test-Retest: Percent Change Across Judges and Language of Sample

After judging the sample, the bilingual judges commented in writing that the speaker on the videotape appeared to have greater proficiency in English (n = 3); that poorer proficiency in Spanish appeared to affect fluency (n = 3); that the speaker had word retrieval issues in Spanish (n = 2); that there were many interjections in both languages (1); and that there were more part-word disfluencies in Spanish (n = 1).

After judging the sample, the monolingual judges wrote that severity seemed greater in Spanish (n = 2); that the speaker had word retrieval issues in Spanish (n = 2); that the English interjection “um” was used during the Spanish sample (n = 2); and that there was a greater variety of disfluency types in Spanish (n = 1).

  1. Discussion

The primary finding, that familiarity with Spanish made no significant difference when judging disfluencies in Spanish, is a first step in gathering evidence about clinical judgments of stuttering in unfamiliar languages. The finding that both groups of judges identified a greater percentage of disfluency in the Spanish sample than in the English sample lends additional support for the idea of correspondence of stuttering behaviors and listener reactions to stuttering across languages and cultures.

Because the monolingual judges’ percentages of disfluency were higher than the bilingual judges’ percentages of disfluency, the question arises whether the monolingual judges may have made number of false positive judgments. That is, the difference in scores for the two groups raises the question whether the monolingual judges may have had a greater tendency to misidentify fluent moments in Spanish as disfluent moments. Even though an item analysis was not conducted, other observations suggest that a significant number of false positives may be ruled out.

First, the monolingual judges consistently identified more disfluencies in both languages, not just in Spanish, and in the same proportion. The monolingual English-speaking judges identified 30.9% more disfluencies in the English sample than the bilingual judges did, and they identified 28.6% more disfluencies in the Spanish sample than the bilingual judges did. Rank ordering the percentages of disfluency determined by the individual judges led to a similar conclusion: for both the English sample and the Spanish sample, five of the highest six percentage scores belonged to the monolingual judges, and five of the lowest six percentage scores belonged to the bilingual judges.

Second, the percentages of disfluency determined by the trained observers were 33% to 58% higher than the average percentages determined by the judges in the present study. Given these data, false positive responses appear less likely. More likely, both the monolingual and bilingual judges would have a high percentage of false negative identifications if their individual judgments were to be compared with those of the trained scorers.

It was not surprising that the subjects (judges) who participated in the present study identified less disfluency than the trained scorers who previously analyzed these samples. This observation is consistent with reported findings that less experienced judges tend to identify less stuttering than more experienced judges (Evans et al., 1999). The judges in the present study were less experienced than the trained scorers, and they did not have access to transcripts or to an unlimited number of replays of the communication sample.

It is possible that the monolingual judges, the bilingual judges and the trained scorers may have judged certain disfluency types differently. It would be interesting to see whether an item analysis of the responses from the three groups would reveal any consistent patterns of identification or misidentification, with regard to types of disfluency.

Both the monolingual judges and the bilingual judges showed a greater change in scores for the English sample than for the Spanish sample (Table 4). Since the severity of disfluency was greater in the Spanish sample, perhaps disfluency judgments were easier, resulting in more consistent scores for the Spanish sample.

The statistically significant difference in test-retest performance between the bilingual judges and the monolingual judges is puzzling. It is not clear whether monolingual versus bilingual proficiency may be a relevant factor. If this finding is replicated, it may be evidence of factors as yet unidentified.

This study focused on differences in quantitative judgments of disfluencies. However, inspection of the judges’ written comments suggests that differences in qualitative judgments of disfluencies should also be investigated, since the bilingual judges in this study appeared to provide more qualitative information.

It is possible that the results of this study may have turned out differently in an area with different demographics. In South Florida, English-speaking residents are fairly likely to hear Spanish and a variety of other languages. The monolingual English-speaking judges in the present study may have developed some familiarity with certain features of Spanish, even though they may not understand or speak the language.

Linguistically, it is possible that English and Spanish may be related closely enough for monolingual English-speaking judges to draw parallels between English and Spanish prosodic, grammatical, or lexical features. The ability to recognize analogous linguistic features in a related language may possibly improve judgments about disfluencies even though the language may be unfamiliar to the clinician. Research is needed to investigate whether clinician judgments about disfluencies may be affected by the degree of similarity of a foreign language to one’s own language.

Research is also needed to determine the abilities of monolingual clinicians to identify linguistic features of unfamiliar languages.

Humphrey (1999, 2003) and Humphrey et al. (2001) reported the transfer of certain treatment effects from one language to another. The results of the present study open another avenue of inquiry that may lead to guidelines for monolingual clinicians who work with bilingual people who stutter.

The author extends his appreciation to Laurel Milo, to all those who served as participants, and to Teri Hamill for statistical assistance.

Cooper, E., & Cooper, C. (1998). Fluency disorders. In D. Battle (Ed.) Communication disorders in multicultural populations, (2"“ ed)., (pp. 247-274). Boston: Andover Medical Publishers.

Evans, J., Brundage, B., & Cordes, A. (1999, November). Clinician judgments of stuttered and nonstuttered speech samples. Poster session presented at the annual convention of the American Speech-Language-Hearing Association, San Francisco CA.

Finn, P., & Cordes, A. (1997). Multicultural identification and treatment of stuttering: A continuing need for research. Journal of Fluency Disorders, 22, 219-236.

Holliday, P. (2001); Demand may exceed supply in future job market. ASHA Leader, 6, 18.

Humphrey, B. ( 1999, November). Bilingual stuttering: Can treating one language improve fluency in both? Poster session presented at the annual convention of the American Speech-Language-Hearing Association, San Francisco CA.

Humphrey, B. (2003). Bilingual stuttering: Transfer of treatment effects across languages. Poster session presented at the annual convention of the American Speech-Language-Hearing Association, Chicago IL.

Humphrey, B., N atour, Y., & Amayreh, M. (2001, May). Bilingual stuttering: Comparing treatment studies of children vs. adults. Paper presented at the annual convention of the Florida Association of Speech-Language Pathologists and Audiologists, Orlando FL.

Jankelowitz, D., & Bortz, M. (1996). The interaction of bilingualism and stuttering in an adult. Journal of Communication Disorders, 29, 223-234.

Kamiol, R. (1992). Stuttering out of bilingualism. First Language, 12, 255-283.

Nkowah, E. (1988). The imbalance of stuttering behavior in bilingual speakers. Journal of Fluency Disorders, 13, 357-373.

Roberts, P. (2002). Disfluency patterns in four bilingual adults who stutter. Journal of Speech- Language Pathology and Audiology, 26, 5-19.

Van Borsel, J ., Maes, E., & Foulon, S. (2001). Stuttering and bilingualism: A review. Journal of Fluency Disorders, 26, 179-205

join button


The IFA implemented Japanese translations of some pages on the site for the 2018 Joint World Congress. 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


Journal of Fluency DisordersBrowse the current issue

The official journal of the International Fluency Association
IFA Members receive online access to JFD as a member benefit.

Read more: JFD

Twitter Feed @official_IFA

IFA on Facebook