Does bilingualism cause stuttering?

« previous post | next post »

You might think that it does, from the headlines "Bilingual children stutter more – study", "Stutter risk for bilingual kids", "Study: bilingual kids more likely to stutter", etc. These stories report on a recently-published study: Peter Howell, Stephen Roger Davis, and Roberta Williams, "The effects of bilingualism on stuttering during late childhood", ADC Published Online 9 September 2008.

Howell et al. studied 317 children who stuttered, all of whom started school in the UK at age four or five, first came to a speech clinic between 8 and 10, and lived in the greater London area. For 69 of the 317, "at least one language other than English was spoken in the home". Thus 21.8% of the stutterers were raised in a home where a language other than English is spoken. In comparison, among schoolchildren of comparable age in the greater London area, the London Education Authority reports that 28.4% come from such homes.

But wait a minute: the percentage of stutterers who come from non-English-speaking homes is smaller than the percentage of kids in general who come from such homes? How does this translate into "Bilingual children stutter more", or "Stutter risk for bilingual kids"?

Well, hang on, we'll get to the connection between (a certain kind of) bilingualism and stuttering that the study suggests may exist. But to understand the argument, you're going to have to put up with some slightly complicated analysis and reasoning.

For 38 of the 69 bilingual stutterers, caregivers "primarily or exclusively" used a language other than English. This excluded the 31 children whose parents "opted to speak a language other than English in the home for their child’s educational/social advancement".

The researchers further subdivided the 38 selected stutterers into 15 LE ("late English") kids, who weren't exposed at all to English until they went to school, and 23 BIL kids, who already spoke both English and some other language when they started school.

And they selected a control group (FB) of fluent (non-stuttering) bilingual kids, age-matched to the stutterers, and also divided into LE and BIL subgroups in the same way.

Now let's look at the 2×2 contingency table for stuttering vs. non-stuttering and LE vs. BIL:

Stuttering 15 23
FB controls 28 10

So if we accept that the FB controls were really matched in every other relevant way, we have some indication that a particular kind of bilingualism might be a risk factor for stuttering. Remember, we're talking about situations where 1) the caregivers speak "primarily or exclusively" another language, but 2) the kid learns English anyhow, for example from playmates or siblings — that's the case they call BIL. This is distinguished from the situations where 1) the caregivers speak "primarily or exclusively" another language, and 2) the kid doesn't learn any English from other sources — that's the situation that they call LE. (Again, we're entirely excluding the cases where the caregivers speak English as well as another language.)

Now, as the contingency table shows, among the stuttering bilinguals, 61% learned English before going to school (i.e. were in sub-group BIL rather than LE); among the fluent bilinguals, only 26% learned English before going to school.

If this were the question period after a presentation, I'd want to ask the authors these questions:

If (some forms of) bilingualism tends to cause stuttering, why is the proportion of non-English-speaking homes among stutterers (in your study) actually lower than in the general school population population? I know that you say the difference is not statistically significant, but you report p= 0.054, which is pretty darn close to p=.05. Shouldn't we expect a difference in the other direction — which you surely didn't find?

If early bilingualism tends to cause stuttering, why did you remove from the study those early-bilingual kids whose caregivers spoke both English and some other language? Does early bilingualism tend to cause stuttering only if the socially-dominant language (here English) is learned from siblings or playmates rather than from parents? If so, why?

The family situations of LE and BIL kids are likely to be rather different — for example, BIL kids presumably tend to live in more ethnically mixed social situations than LE kids do. Could this (rather than the early bilingualism itself) be responsible for the apparent increase in stuttering?

Prof. Howell and the other authors may very well have good answers for these questions. Perhaps they even answered them in their paper, and I missed it on a quick read this morning. But those are the questions that the paper left me with.

[I should note in passing that there hasn't been a lot of media uptake on this story, which surprises me a bit given the generally high level of interest in bilingualism. Maybe the press release didn't hit the right wires; or maybe there's been too much other news, what with tropical storms, financial meltdowns and U.S. politics. Or then again, maybe most editors took a look and decided to pass this one up. If you have any insight into this process from the inside, please share it with the rest of us.]


  1. Robert F said,

    September 17, 2008 @ 9:58 am

    I can't view the paper, but could this be a case of choosing what thing to test for after you've got the data? The data taken as a whole (nearly) shows the opposite effect from what was wanted, so a specific subset of it is selected that does show the required effect. Because of the birthday paradox, the significance test will underestimate the probability of finding any such correlation.

  2. john riemann soong said,

    September 17, 2008 @ 10:23 am

    I like how one news source is framing this as, "A new study might have some parents wondering if bilingual education should start later in a child's life. "

    Exactly what parents should NOT do, if they ever want their children to have a language instinct … [for the desired target language]

  3. Mark Liberman said,

    September 17, 2008 @ 11:08 am

    @john riemann soong: The issue is whether to start the dominant-culture language (English) soon after birth or at age 5, which is still well within the critical period. The authors' claim is that "late English" (i.e. starting at age 5) is less likely to result in stuttering, and doesn't cause any scholastic problems in their sample. This could all be true — but I'm puzzled about a few things, as explained in the body of the post.

    @Robert F: I don't think that the authors were going "subset shopping". The idea that various kinds of childhood bilingualism might (or might not) have different connections to stuttering has been around for a while, and (I think) was not invented for this paper. But as I said in the body of the post, it's curious that early bilingualism apparently doesn't associate with increased stuttering if the parents are involved; that makes me wonder whether it's not early bilingualism itself, but something else in the social/family situation that's responsible for the effects.

  4. James Wimberley said,

    September 17, 2008 @ 12:53 pm

    "For 38 of the 69 bilingual stutterers, caregivers "primarily or exclusively" used a language other than English. This excluded the 31 children whose parents "opted to speak a language other than English in the home for their child’s educational/social advancement"."

    This looks like a distinction between "poor" and "middle class".

  5. Jonathan Mayhew said,

    September 17, 2008 @ 1:16 pm

    "Monolinguism increases risk for stuttering." That's the conclusion I draw. Do they give a suggested reason why languages spoken in the home might lead to stuttering?

    Only a little more than 10% of the stutterers come from bilingual homes, once the exclusion of middle class parents talking to their kids in French "aspirationally" is excluded. The Late English effect shows up in 23/38 of these kids, so it is not a major cause of stuttering among the 317 total stutterers. Without a theory of why monolingual kids stutter, or why anyone does in the first place, it is hard to find a context in which to interpret this result. What are the other known risk factors, and do these 23 LE kids have those? How do those factors correlate with socio-economic status?

  6. marie-lucie said,

    September 17, 2008 @ 2:00 pm

    My understanding is that stuttering linked to bilingualism is more likely to occur if the children are not sure which language they are supposed to use, for instance if the adults use either language indiscriminately when addressing the children. It seems to me that where the family members are consistent (eg parents or grandparents in mixed families each speaking their own language to the children), the children are much less likely to have a problem. I wonder if the family or even the school situation was considered in the study: e.g. if the children had siblings, playmates or classmates who were bilingual and were likely to use either language.

  7. Michael Roberts said,

    September 17, 2008 @ 2:45 pm

    Perhaps stuttering increases the risk of bilingualism.

  8. Lance said,

    September 17, 2008 @ 3:32 pm

    "Hey, Ethel, the kid's stuttering when he speaks English. Wanna try Spanish on him, see if he's any better at that?"

  9. Julie said,

    September 22, 2008 @ 11:46 pm

    This is article is timely to me… I speak English to my 3 year old son, my husband speaks Zapotec to him, we speak Spanish to each other. My son is bilingual in E and Z, and understands but doesn't S, E is his dominant language. He stuttered some in E starting several months ago, but not in Z. 6 weeks ago, we left our home in Mexico and came to a predominantly E-speaking place. Since that time, my son's stuttering has become severe in E and he has started to stutter some in Z. What to make of this, I'm not sure, but I'm curious to see what happens when we return to our S- and Z- dominant home in Mexico.

  10. Shynell said,

    September 24, 2008 @ 1:01 am

    Hi all,
    I do not have access to the reference study. I also have a situation with my son stuttering. He is bilingual English/Japanese. I am an english speaker. From the age of 2 months to three he socialized with Japnaese speaking caregivers. We returned to America in Apr 2008(5months) he began stuttering within in a month of our return.
    Does the article offer any solutions?

  11. Does Bilingualism Increase Stuttering? | 2 Languages 2 Worlds said,

    May 25, 2013 @ 11:40 pm

    […] are less likely to stutter than bilinguals who start using both their languages from an early age. As pointed out however, it seems that the prevalence of stuttering in bilinguals no higher (or maybe less) than that of monolinguals. I think however we […]

RSS feed for comments on this post