Discriminating Children with SLI Among English Language-Learners How to discriminate bilingual second-language children with typical development from those with language impairment. Presentation Video
Presentation Video  |   November 01, 2014
Discriminating Children with SLI Among English Language-Learners
Author Affiliations & Notes
  • Johanne Paradis
    University of Alberta
  • Presented at the 24th Annual Research Symposium at the ASHA Convention (November 2014).
    Presented at the 24th Annual Research Symposium at the ASHA Convention (November 2014). ×
  • The Research Symposium is hosted by the American Speech-Language-Hearing Association, and is supported in part by grant R13DC003383 from the National Institute on Deafness and Other Communication Disorders (NIDCD) of the National Institutes of Health (NIH).
    The Research Symposium is hosted by the American Speech-Language-Hearing Association, and is supported in part by grant R13DC003383 from the National Institute on Deafness and Other Communication Disorders (NIDCD) of the National Institutes of Health (NIH).×
Article Information
Development / Cultural & Linguistic Diversity / Research Issues, Methods & Evidence-Based Practice / Language Disorders / Specific Language Impairment / Clinical Practice Research / Diagnostics, Screening and Assessment Research
Presentation Video   |   November 01, 2014
Discriminating Children with SLI Among English Language-Learners
CREd Library, November 2014, doi:10.1044/cred-pvd-c14002
CREd Library, November 2014, doi:10.1044/cred-pvd-c14002

The following is a transcript of the presentation video, edited for clarity.

Linguistic Diversity and Clinical Assessment
On the theme of bilingualness, bilingualism and variants my title has a primary and/specific language impairment, that's because it seems that both terms are floating around these days and in the interest of respect for variation I decided to put those in my title, however I use the term SLI. So I'm going to start by sort of laying out the problems base when it comes to linguistic diversity and clinical assessment and then move on to my research.
In the Canadian context this is kind of a picture of what our linguistic diversity looks like. We have mostly English speakers with a substantial portion of French speakers, and a little slice of the pie there in the corner that sort of peachy-looking color is what we call non-official languages. That just means not French, not English. Now certainly the children who speak non-official languages is much bigger in metropolitan areas, but for the country as a whole this is what it looks like. Now when we're talking about diversity this three-language wheel here doesn't look like much but let's take that little peach slice and blow it up and this is what we get. So we have in Canada well over 100 non-official languages and the other striking factor is that no single language dominates among the non-official languages. We have fairly big slices of people who speak Spanish, Cantonese, Punjabi, Urdu but it's quite a lot of diversity.
Now in the United States of course what comes to people's minds mostly is bilingualism means Spanish, but it's not entirely the case here either. I mean it's largely the case but not entirely the case. So, about 20% of ELL children here in the United States do not have Spanish as their first language. What I have in the bottom of this slide is a page from a very long document about language use and diversity in the United States and what I've got. In the first circled item there this is 5 to 17-year-olds in the United States and the purple bar is people who speak English in this age category. And the next bars are the other languages and this is looking at how well they speak English, those people from other language backgrounds. What struck me is this, if you look at the circle on the far side there is the, first of all there's Spanish in its own category but then the other categories are enormous. One is just called Indo-European which is basically a huge swath of languages, everywhere from Bangladesh to Portugal, and then Asia and Pacific Island languages and then other languages. So essentially what it looks like here in the U.S. is that you have English. You have Spanish but you also have that big wheel of diversity as well.
When diversity meets assessment, this is where the problem is which as I'm sure everyone in this room is familiar with. At least in Canada, the numbers are growing of children with little or no English proficiency in preschool and kindergarten and Grade 1, and their diversity is increasing as well. But then at the same time there is a lot of pressure on early assessment for language and learning disabilities. They want to catch these kids as early as they can and so forth in trying to assess them before they even have learned the majority language is a problem. And so one of the things that I receive a lot of emails about is what I do with kids who are ELL, English Language Learners, when I'm supposed to assess them? And of course we know that the prevalence of over-identification for ELL children is problematic, and it's persistent so of course it's good news to hear that people are feeling cautious about this and trying to wonder what they do. One of the reasons why I'm interested SLI in particular is that this is one of the most problematic disorders to diagnose in a diverse population because the symptoms of this disorder lie in language itself largely, so it makes it much, much more complicated to figure out what ELL kids actually have SLI versus those that don't. One of the things that has been sort of put forward as a strategy for dealing with ELL children in assessment is bilingual assessment.
Both ASHA and CASLPA put forward position papers that suggest bilingual assessment as the best practice for identification, and of course it makes sense. If you can test a child in both languages you get the full range of understanding of their linguistic competences and so on; however, bilingual assessment is rarely realistic. You need a bilingually-trained SLP or para-professional personnel. You need bilingual test materials or at least some kind of knowledge of what the child's language other than English should look like. And for many languages there is absolutely zero information on even how typical children acquire that language, so you don't have a lot to deal with. And so even though bilingual assessment is so clearly, logically, the best choice it is so rarely achieved. In fact as far as I'm concerned I'm a little worried that this is often held as the gold standard when so few people can actually do it.
So this raises the question, can you actually do an assessment and accurately identify ELL children if you're just assessing them in English or mainly English? Is this possible? Okay, so this is going to be the question that hovers over this entire talk. I think that there are two things that can go into figuring out whether this is possible and so what I'm going to talk about is research that has come out of my lab that fits both of these categories. On the one hand if we're going to look at mainly English-based assessment or really focusing on English in the assessment, we have to increase people's knowledge of what typical ELL development is actually like. Clinicians, teachers, everyone has to have a schema in their head about what they expect these kids' language development to look like. We have to know in particular what their performance is like on standardized tests, the sort of instruments that clinicians use, and we also have to know what are the characteristics of ELL children who actually do have SLI? How do they differ from typically-developing ELL children, and how do they differ from monolinguals with SLI? So that's the knowledge side of things.
Another thing, another sort of stream that we've been working on in my lab is trying to develop some practical resources for assessment with the ELL kids so we've been working on a parent questionnaire on first-language development. We've been working on developing ELL norms for some tests and that second part is available on the CHESL website.
ELL Development and Performance on Standardized Tests
So let's start with the knowledge side of things. What I'm going to do is give some highlights of the research that we've been doing in the lab for the past ten years looking at ELL development and in particular performance on standardized tests.
So, how long does it take to become a native speaker? Most people would expect that a child who immigrates to a country at a very early age or who was born there is going to eventually become indistinguishable from monolingual native speakers of that language and they're going to do it fairly rapidly. I mean some people think they soak up language like a sponge and after half a semester of kindergarten they're perfectly, fluently English speakers. Well it doesn't quite work that way. And just to start off the discussion of this I'm going to show a video tape of Cindy. And Cindy is answering a question about what she wants to be when she grows up and you'll see her at the very beginning stages when she's starting to learn English and you'll see her answering the same questions six months later.

[Video Clip]

Cindy: A doctor. A doctor.

Examiner: [Inaudible]

Cindy: Because I want to see someone inside of here. Because, because I really really want to.

[Six months later.]

Examiner: So what do you want to be when you grow up?

Cindy: A doctor.

Examiner: A doctor? Why do you want to be a doctor?

Cindy: Because I'm interested in [inaudible] a doctor, and I'm interested in the body, the things in your body.

So Cindy made a lot of progress. For those of you who were able to hear that she made a lot of progress over those six months. She sounds even quite, very English-speaking. She's lost some of her foreign accent, but she is way below age expectations on all measures of grammar and the lexicon and so forth. And this was just an illustration about how ELL kids can sound really good early on and fool us when it really takes quite a long time to become a native speaker.
So how long does it really take? Well, depending on the study it can take up between three and seven years -- that's a very long span -- for these children to catch up for oral language. And of course there are many individual difference factors that play a role in how fast or slow they will catch up. So what their first language, their L1 is, maternal education, their language learning aptitude, the richness of their English environment, et cetera, et cetera.
Another facet of their development that's important is that they have what's called profile effects. That means that they don't catch up for all linguistic domains at the same time. They catch up faster for some linguistic domains than others.
Another point is that in terms of their long-term attainment. We just completed a longitudinal study in my lab and we found that they don't become identical to native speakers in grammar for absolutely everything. There are some things where they retain a kind of a level that is a little bit more variable than what we see in native speakers.
How Long Does It Take?
I'm summarizing across all these studies here. So now I'm going to show you a bit of data to back up all these points. This is a study with looking at ELLs, it's a longitudinal study with spontaneous language sampling. They're looked at for five rounds across two years, so six months every round. And we're looking at one thing, whether they produce third person singular 's'. He walk, he walks. And we studied them from nine months' exposure to English to 35 months' exposure to English. Now what you'll see in this box plot graph is that the first round basically none of them are getting this and nothing is going on. And in the next two rounds there is tons of variability, and then toward the end we see higher scores and a bit of a plateau with still some variability. So it's a non-linear shape and some individuals even at round five, after nearly three years of exposure to English are definitely below 90% correct use in spontaneous speech, so not at what is considered kind of the standard of acquisition.
Now here's what happens when we look at the children divided into different first languages. So the panel on the far side there is all the children and then in Panel 2 we have children who have first languages that are inflecting, typologically inflecting languages. So, these are Arabic, Punjabi, Urdu, Hindi and Spanish. And the ones in Panel 3 have Mandarin and Cantonese as their first language. Mandarin and Cantonese do not mark tense grammatically, and they do not use verb inflections. They have morphological particles. They have morphology but not verb inflection and they don't have tense in their grammar. So what we see in Panel 2 is the kids, first of all Round 1 not much is going on but they really zip up to the top pretty quickly okay and get very good by Round 3. However, the ones who speak Mandarin and Cantonese, they're the ones with this enormous variation even at Round 5 there are huge amounts of variation. So, what a kid's first language is makes a difference in how fast or slow they are in acquiring English, at least this part of the grammar.
This is a study we did with vocabulary, so this is just the PPVT. What we've got plotted here are the children's mean age in months and their PPVT age as well and over the five rounds so this was the same longitudinal study and of course the child's actual chronological age goes up linearly like it should. The PPVT age however goes up a little bit, non-linearly to sort of meet age but it doesn't quite meet it at three years of exposure or 34 months of exposure to English. Even for vocabulary the kids are not quite there after three years.
This is that same study where we divided the children into two groups, one with mothers who had higher education, a post-secondary degree and those that did not, and here we find the predictable kind of pattern that one might see if this were monolingual data where the kids whose mothers only had primary/secondary education are lagging behind in the size of their vocabulary and the kids who have mothers with post-secondary education are doing better. But here's the interesting thing, the mothers who had post-secondary education were more likely to speak to their children in the first language. They placed a high value on preserving the native language and they spoke it more to their kids.
So I think what this does is it raises the issue of what is it that high levels of maternal education actually do for kids? Is it just how much speech they use with them or the quantity of speech they use with them or is it the quality of the interaction and the kind of speech they use with them? And I think that in our interpretation of this data was that this probably signals that quality is what is really going on here because the interactions are not happening in English and yet the kid's English vocabulary is higher.
Now this is another example of how first language can change test results. This is the CTOPP non-word repetition scores. And this is looking at exposure, differences according to how much exposure the children have had and according to first-language background. This is a cross-sectional study, not longitudinal. So in the first panel we have children who have had less than 12 months, less than one year of English exposure. In the middle panel it's between one to two years, and in the last panel it's over two years exposure. And then we divided them into two groups. One of the groups was children whose first languages do not put a lot of restrictions on coda consonants in syllables. So, coda consonants just means, if you have that consonant at the end of the syllable, not about the beginning but the end of the syllable that has a consonant. Some languages place restrictions on that. They don't allow any codas or they allow maybe one consonant type to be in the coda and they don't allow clusters of consonants and codas, etc. whereas English is a language that allows multiple kinds of consonants and codas and multiple clusters. And certainly the CTOPP non-word repetition task reflects phonotactics of English and there are lots of codas. So my grad student Tamara Sorenson Duncan who did this research was looking and scoring these non-word repetition tests and noticing that some kids were reaching ceiling or the critical mass of mistakes and we had to stop the task and it was always because they were dropping the codas. So she thought well maybe first language is making a difference. So that's why we're looking at two things, amount of exposure to English and first language.
What we see in the very first panel with the kids who are 12 months or less of English, we see, between the dotted lines is kind of like the monolingual one standard deviation range so we're looking to see that those, the box plots of the ELL kids, how close they are to fitting in there. And we'll see that for neither language background no coda or coda are they overwhelmingly fitting within that frame. But as time goes on, as they start to get more exposure to English, we start to see differences according to first language. So if we go to the very final panel where they have over two years' exposure we see that the kids who have first languages with codas are fitting except for a few outliers, are fitting within that normal range. However the kids who have first languages with no codas are really still variable and having a lot of, a lot of variation in their scores and therefore could look like they're perhaps in need of intervention if you looked at the scores on this test.
Profile Effects
So it takes a long time to learn English and there are certain individual variation factors like what your first language is, maternal education. There are even more of them but I don't have time to talk about them so I'm going to move on to the second point I made in the summary slide which is about profile effects. So, profile effects basically mean there's an asynchrony in development of different language and literacy skills and we know that ELLs just don't come up to speed with monolinguals for the same things at the same time.
So, if we imagine, this is not real data, but if we imagine a set of data like this where supposed you're testing a group of children, ELL children, at Year 1, Year 2 and Year 3 and you have a measure of phonology, a measure of the lexicon and a measure of syntax, and we've got zero to 100 there on there, so those are their scores. And if you imagine them sort of at the first year, they're not scoring very well, the second year they're at 50% and there's the third year. But notice they're going up evenly for all these different language subdomains. That's not what happens. This is what happens. So, what you get is unevenness in terms of how fast or slow they catch up to monolingual norms. So, some aspects of language they trail behind more than others.
So let's look at this with real data and this is from the 2011 book. This was a group of children I followed longitudinally and for one year, two years and three years of exposure to English, and this is the percentage of children in the sample who met monolingual norms on that test. So they just squeaked above that one standard deviation or higher, but I mean they weren't exactly, had to be at monolingual mean. They just had to be within normal range. And for the TEGI it was, they met criteria score. So we had a narrative story grammar measure from the Edmonton Narrative Norms Instrument. We had vocabulary size with PPVT and grammatical morphemes was with the TEGI. And what you can see is that the bars are very uneven. So, for the dark bars, which is the narrative task, by about three years of exposure every one of these kids was actually in the normal range for story grammar. And the PPVT was in between but by three years 80 some percent of them met the monolingual normal range criteria. But for grammatical morphemes, tense morphemes you can see that they're trailing behind. So they don't acquire everything in synchrony and that's something to keep in mind if you're looking at some kind of test. You have to keep in mind what subdomain of language is this test measuring because that, you have to frame your expectations according to that.
Long-Term Outcomes
Now this is a study we just finished recently, and this is a longitudinal study of ELL kids from their fourth to their six and a half years of exposure to English in full-time schooling. This is results from the TEGI, so we have third person singular 's', and we have past regular, past irregular, BE morphemes and questions and 'do' in questions. We found that even at Round 3 -- so this is six and a half years of exposure to English -- 60% of the children in the sample were below criterion on one or more TEGI probes. That kind of surprised me.
Now all of these children had a Chinese first language, and kids with a Chinese first language, either Mandarin or Cantonese or some other Chinese language tend to be slower to acquire verb morphology. We found that out in our research with kids of earlier stages of English development, and this trailing behind seems to continue even at this stage and our modelling, looking at the shape of the curve shows that most of the kids are plateauing and their scores on this task, at this age.
Which raises the question, is this the long-term outcome for bilingual kids with a certain first-language background? I mean these kids are, their scores are pretty high and pretty good. They're productive. They're producing these morphemes, but what they lack is that uniformity that we see in monolinguals. One thing about verb morphology and monolingual speakers of a language is that they nail it. They get it right almost all the time, and it seems to be that what might be possible here is that in ELLs or bilinguals in general, it might be that there's a little bit of fuzziness and variability in how precise they are with this kind of morphology. And if this is truly their long-term outcome that's another way in which one has to set one's expectations appropriately.
Development of ELL with TD and ELL with SLI
So the second part of the sort of knowledge part of the research program is looking at comparing ELLs who have typical development to those who actually do have SLI, to see how one can tell them apart in terms of characteristics. So I've got another video for you to start of this section. You're going to see a little boy who is first-language Cantonese. This is the first one and he has a specific language impairment and you're going to see him at one round, and then you're going to see him a year later and he's taken the TEGI and he's doing the third person singular s-probe.
And then the second kid is named David and he is a Spanish speaker and he is a typically-developing ELL, also taking the exact same probe. What I'm hoping that you can appreciate is that while both children progress in their English, the typically-developing child progresses from almost getting none of them right to really quickly and much more language but then the child who has SLI so that just to get an impression of the fact that kids who are ELL with SLI acquire English more slowly generally and they also have particular trouble with verb inflection.

[Video Clip: SLI, L1 Cantonese. Round 1.]

Examiner: Tell me what a baseball player does?

Wilson: Play a ball

Examiner: He ...

Wilson: Play a ball

Examiner: Here's a nurse. Tell me what a nurse does? She ...

Wilson: Tape

Examiner: Here's an astronaut. Tell me what an astronaut does. You don't know? Look at this. What's this? What do you think he does? You don't know?

[Video Clip: SLI, L1 Cantonese. Round 3.]

Examiner: Here's a baseball player. Tell me what a baseball player does.

Wilson: To hit the ball. To win the game.

Examiner: Yeah. So can you say a whole sentence for me?

Wilson: He hit the ball to try to win.

Examiner: So here's a nurse. Tell me what a nurse does.

Wilson: To [inaudible] a bandaid for the people so the people aren't hurt.

Examiner: Yeah, so can you say the whole sentence for me?

Wilson: She got a bandaid to not let the people hurt.

Examiner: So here's an astronaut. Tell me what an astronaut does.

Wilson: Put the helmet.

Examiner: Yeah, so can you say a whole sentence for me?

Wilson: He put a helmet.

[Video Clip: TD, L1 Spanish. Round 1.]

Examiner: Here's a baseball player. Tell me what a baseball player does.

David: He play baseball.

Examiner: Here's a nurse. Tell me what a nurse does?

David: He put a bandaid in there.

Examiner: Here's an astronaut. Tell me what an astronaut does.

David: He fly a spaceship up, up, up, up.

[Video Clip: TD, L1 Spanish. Round 3.]

Examiner: So here's a baseball player. Tell me what a baseball does.

David: The baseball player hits the ball, and he hits the other baseball player's head.

Examiner: Here's a nurse. Tell me what a nurse does.

David: The nurse puts a bandage on you, and then you put the other bandage on her.

Examiner: So here's an astronaut. Tell me what an astronaut does.

David: The astronaut goes out of Canada and then he crashes to a planet and the name is Pluto.

Alright, so I'm going to start off this section here with a sort of a summary of the findings, and then I'll show you some data to support these assertions.
Thinking about the relationship between typically-developing ELLs and those with SLI, I'm making a point of comparison here with what we know about monolingual children who are typically developing and have SLI. Generally speaking monolingual and ELL children with SLI are slower in general to acquire English than their typically developing peers. Tense morphology is an area of extreme difficulty for both ELL and monolingual children with SLI. Non-word repetition is also an area of extreme difficulty, so beyond the general slowness is what I mean by extreme difficulty for monolingual and ELL children with SLI. So those sorts of things are similar.
There are some little differences though. One of the things we find, it's in the early stages of development that, the gap between typically-developing children and children with SLI in their scores on a, let's say the TEGI or another task in tense morphology or non-word repetition, is smaller for ELLs. This of course makes sense because they're both in the process of learning English. So, we see the ELLs with SLI scoring lower, but we don't see these giant gaps in the scores until they've had more exposure to English. The gap widens as the kids get more exposure to English but unfortunately often at the time when one is assessing these kids they haven't quite reached that optimal gap point because they've only had one or two years of exposure.
Another interesting difference between ELLs who are typically developing and those with SLI is that they don't acquire inflectional morphology or tense morphology, I mean as a group like monolinguals do who acquire it more or less at the same time. For some reason 'BE' morphemes seems to appeal to children, ELL children and they acquire it, those morphemes much faster and this is true for ones with SLI as well. So those are differences and similarities.
Now what I've got here are two case studies compared to group data. These patterns are confirmed with larger groups of ELLs with SLI, but I like these graphs because I think they show clearly what's going on. So, with this little boy here, Wilson, the one you just saw on the videotape, what we have here with the dotted lines are his scores for tense morphology in dark circles and his scores for non-tense morphology in the open circles. And it's being compared to group data so we have typically-developing children who also speak Cantonese as a first language, and their tense scores are in the dark squares and their non-tense morphology are in the open squares. So, non-tense morphology is a composite of their abilities with plural 's' and adding 'ing' to the verb and determiners. What we see is that the child with SLI has lower scores for both types of morphemes, but the scores for tense morphemes are much, much lower. So the gap between say non-tense morphemes, the two, the open square and open circle ones is much narrower than the one for tense. And that parallels what we see for monolingual children with SLI.
The next panel is a child where the SLP wasn't sure if he had language delay or SLI because he seemed somehow resolved in his first language according to his mother, and so when we studied him in English if you look at the very bottom of the graph there which is dash-lined with the dots, with the closed circles, that's his tense scores. And you see that from 10 months to 23 months he's really, really slow in English and then he zooms up and meets the typically-developing kids in his ability with tense. So it looks like in some ways he was language-delayed in the second language as well. But the main take-home message is this pattern of what is more extremely difficult for kids with SLI versus typically-developing kids is something that seems to play out the same way in ELLs as we see it in monolinguals.
This is just the same kids looking also at this acquiring BE before inflectional morphemes so again the kids with SLI who are ELL also show this love of BE morphemes and acquire them much faster. In some ways their trajectories for BE morphemes look more like non-tense morphemes and the inflection morphology is what is lagging behind.
Now this was a study that I did with a postdoc Elma Blom who is now at Utrecht University, and she was looking at the ELLs with typical development and SLI in terms of past tense. This is cross-sectional data, not longitudinal data and so she had two groups of 24 kids and they were matched for age and exposure. As you can see by the box plots the kids with SLI were doing much worse on the past tense probe. Again this is the TEGI. She found that for both children their scores on the test were influenced by their vocabulary size and the word frequency, so the more frequent the word, the more likely they were to add the tense to it. The bigger their vocabularies in English, the better they were on this task. But there were also some intriguing differences between the typically-developing kids and the kids with SLI. She found that the kids with SLI over-regularized much less, so they were much less likely to say 'digged' for 'dug'. They mostly just dropped anything from the verb than the typically-developing kids. They were also not sensitive to allomorph so when we add 'ed' to make the past tense in English it's sometimes pronounced as a 't' or a 'd' or an 'idid' depending on the last phoneme of the verb stem, and the kids with typical development were sensitive to this and the kids with SLI were not.
So, one of the things that came out of that is that Elma and I wanted to do a second study where we looked at, more at these differences between kids with typical development and SLI. In this particular study we looked at third person singular and past tense, so this was the TEGI screener. And we looked at individual difference factors. We wanted to see does first language matter? Does amount of exposure matter? Does age matter? Again, it's a cross-sectional sample, but there was a variation in age, a variation in amount of exposure to English, and variation in first language. And what we found was that the typically-developing ELLs definitely showed positive transfer from their first language. So if their first language was an inflecting language they were better with past tense in English and third person singular 's'. But we saw no first language transfer for the kids with SLI. It's like they just were immune to figuring out that hey my other language has this too maybe I should use it in English. They just didn't seem to have that going on.
The other thing and this is the plot I have in this slide here, this is an interaction plot between group which is language impaired and TD over exposure. What this depicts visually is that for the typically-developing kids more exposure to English meant they were more accurate with verb inflection. But more exposure to English, as you can see by the flat line there, did not predict whether the kids with SLI were more accurate with English so again, they're just not making use of resources. This is how, in the article we referred to it as resources for acquiring a language. The ELLs with SLI didn't seem to be using their first language, and they didn't seem to be making adequate use of all this exposure they were getting. The only factor that predicted how well they did on the TEGI was age. And the interesting thing about this sample is age was not co-related with exposure. For monolingual children, age and exposure are always co-related because if you're five years old, you've had five years' experience with English, but not so with ELLs. ELLs allow us to actually pull apart age and exposure in the study. And so what we found here is basically the older kids did better and the younger kids did worse and that was it. So I think that what this sort of suggests is that ELLs with SLI are sort of acquiring English with slightly different resources or use of resources than typically-developing ELL kids.
Now what about these profile effects that I showed you for typically-developing ELL kids? What about those who have SLI? Do they also show profile effects? So I'm just going to go through this kind of quickly. This is a study that we did of 152 kids with typical development and 26 kids with SLI, and we gave them a battery of different tests.
So this is the part I want to show you. This is the percentage of children who fell below monolingual age expectations on that test by however that test measures that. So on the TEGI it would be below criterion, on non-word repetition it would be a score below seven, standard score below seven. So the blue bars are the typically-developing children and the red bars are the kids with SLI. So, what you can see here is the bars are uneven, so there are profile effects for both groups. There are some things that they're both acquiring faster and some things they're both acquiring slower. So this profile effect pattern also shows up for kids who have SLI and are learning English as a second language, and the typically-developing children as well.
Now this is data we've just sort of processed like a month ago so this is a longitudinal study looking at ELLs with SLI and without SLI in their fourth to sixth year of exposure. So we want to see what happens to these kids, English language learners with SLI long-term. In this case I'm going to show you about the TEGI data. So what we found long-term is that the kids with SLI were very equivalent to the typically-developing kids for their production of tense morphemes on the TEGI except for past irregular. They were behind and I'll show you they have a very unusual shape of curve. However they were consistently below the ELL children who are typically developing for grammaticality judgements. Both groups showed an increase in accuracy over time but there was more increase for the SLI. And they also showed this, they were more accurate with judging errors with BE morphemes than they were judging errors with inflectional morphemes so this precocious BE, this BE ahead of all the inflectional morphemes is still going on even at this late stage. What I found intriguing about this data is that it looks a lot like data that Mabel Rice has published for monolingual kids with SLI and the fact that these differences can persist long-term between kids with SLI and typically-developing kids with verb morphology.
These are the graphical results. For the past irregular there's a very strange shape to the curve. The TD kids basically show no change from Round 1 to Round 2 and then they show a significant change from 2 to 3 and the kids with SLI show a change from Round 1 to 2 but no change from 2 to 3. I don't know whether those kids with SLI will plateau with irregular verbs, or whether they're just late bloomers and they'll eventually have an uptick later. But the more interesting one is the panel on the far side which is the judging of whether a tense morpheme is omitted where we see there's no change overtime for either group and the SLI kids are persistently lower than the kids with typical development. So these differences persist over time.
Resources for Assessment with ELL Children
Now I don't have very much time left, and it's the end of the afternoon so I'm going to whip through just some highlights here. This is just a few things that our lab has done to try to create resources that SLPs can use in assessment.
So for this summary I cooked up a kind of a diagram for what kind of strategies one might use and first of all, one of the first things one must do is find out how much exposure to English a child has had, especially how much in a structured program, not home exposure or overhearing the television because there's lot of data to show that that doesn't really amount to all that much unless they're exposed to English at home from a fluent speaker of English.
You need to find out what their first language development history and current abilities in the first language are because nobody has a language impairment in just one language.
I've got a middle bar there about prioritizing some tests that are less biased versus tests that are perhaps more biased but I'm going to skip that because I'm not a big fan of that. I don't think you should use monolingual norms ever.
And the last box is using alternatives to monolingual norm referencing is the way to go, our research shows. And by that I mean compare ELL children to other ELL children.
So, the Alberta Language Development Questionnaire is an instrument that we've developed that you can download from the CHESL website and use and we've got a paper about its properties of sensitivity and specificity. Essentially what we found is that it had good specificity but not good enough sensitivity to be used alone which is not surprising. It's just a questionnaire after all.
So then we went on to thinking well what if we combined this first language questionnaire with some English language measures would putting the two together and using ELL norms actually work to discriminate children with language impairment from those with typical development? These were the identification properties of this particular combination of English language tests and the first language questionnaire. So we got over 90% specificity and sensitivity using that combination but the crucial thing was is that you're comparing ELL kids to other ELL kids. And I won't go through my little example here but all of this is on the CHESL website where you can enter in scores in a calculator and it will compare that one child's scores to the scores from our sample that we used in our research so you can find out where that kid lies in terms of what is typical or atypical in ELL development. So we thought that might be useful so it's available and you can check it out if you're interested.
Questions and Discussion

Audience Question

What I've been noticing with younger children, especially if they're not exposed to English, is it's really difficult to assess them as an ELL learner. When they go to preschool the speech therapist does not really know if this child needs extra support or services. Do you think a parent questionnaire is enough, especially since I know from my culture that usually parents think that children will grow out of it so they're not really concerned.

Well, certainly I think a parent questionnaire is probably the only thing you can do ,and it is worth doing because it does seem to indicate, in our research and we had a very diverse sample, obviously not every culture on the planet or language on the planet, but we had a lot of different languages represented and most of our questions aren't really about parent concern, but parent judgement about their child's abilities and early milestones and the child's activity preferences were particularly good at picking out kids who did not have typical development. So, obviously it's not perfect to do something like that, and I feel a little nervous saying anything about this because I'm not actually a clinician. But I think that in the absence of being able to access any other linguistic information a questionnaire is at least one thing that you can do.
If you can get a resource person like a cultural broker interpreter -- that's what we do in my district -- who speaks that child's language and is from that culture and who is trained to do this because that's a big factor, because they've got to know what to look for and not judge and all that. And maybe they can observe the child and talk to the child and let whether they think that child's language abilities are age-appropriate. Putting those two together, it's better than nothing right?

Audience Question

You were talking a lot about the time of exposure to L2 acquisition for English in that sense. Do you have any insight or ideas on whether the language input for L1 actually plays a role in the acquisition of L2, the second language? I feel like sometimes families, they'll start to speak English and kind of drop their L1 thinking it would enhance the language acquisition.

So that's a really important question and I'm glad you brought it up because it's so common. It's so common that immigrant families think that they should drop their home language and just speak English or they get advised by a teacher or a pediatrician or an SLP to do so. There is so much evidence to show that is a really bad idea -- if they're not fluent English speakers. If they're fluent English speakers that's a whole other story. Most of the time however they're not. Most of the time their English abilities are about beginner to intermediate level at best, and they probably will never get any better than that. And what will happen is that they won't be able to express themselves fully. They won't be able to give their children rich linguistic input. And some long-term studies have shown that conflict in families and between parents and children, and intimacy and all those things when the kids are teenagers, all are influenced by language choices when the kids were younger. For example if they keep the home language going intimacy is increased and conflicts are decreased so one of the things about this particular suggestion is not only is it bad to suggest that they switch only to English when the kids are really little, it has long-term cascading problems later on because I've often when I've met with parents and talked to them about this I say look, you think your English is good enough to control the life of a 4-year-old. It probably is. Do you really think it's good enough to control and interact with a 14-year-old? Probably not. How are you going to really talk to them about things when they're that age? So it's really something that has negative consequences and, and this is an important point because sometimes the parents come back and say, yes but English is what's really important. We really want to help them learn English. Well, all the data that I've gotten in these studies for over ten years shows that language use in the home by non-proficient speakers has zero impact on predicting children's English development. They're not helping their kids learn English. So even the goal that they think it's going to, the thing that they think it's going to do, the one thing, doesn't even happen.

Audience Question

My question kind of builds on that one too and it was to see if you've investigated linguistic attitudes towards the L1 or even L2, parents give, if that affected their rate of acquisition?

A couple of my graduate students have been looking into this. We haven't published anything about it though. They found that they didn't, -- well they found that if parents valued the first language and culture, they would more often speak it in the home. But we didn't find that this had an effect on English at all. So kids who had a very rich say Mandarin environment at home, because Mandarin and Cantonese families tend to really value home language and culture and the heritage and all that but the richness in Mandarin or Cantonese scores on any of our measures had no correlation with how good the child was in English. So I think what that speaks to is there's no trade-off going on here so that using that home language and going to activities in the home language and everything like that did not take away from the child's English development. So I don't know if that's really about attitudes as much as it's about what practices can do.

Audience Question

I had a question about the balance of the bilinguals. So you presented the data here in terms of length of exposure, and I was wondering if you had any variation or measurement in terms of the percentage of exposure that children are getting either at home or if they have L1 exposure in the community and if that differs across the different languages that you study?

Certainly. Length of exposure to English was one measure that we used. We also had a percentage of proportion of languages used at home. So it was the languages the child uses with interlocutors and the language interlocutors use with the child on a rating scale of English 'Always' mother tongue 'Never' and some point in between. We also had a mother tongue and English richness measure. So these are measures of quality, book reading and quality activities that take place in each of the languages. So we had three separate kinds of measures. They are interrelated, but they're measuring slightly different things.
So one of the interesting things that I just mentioned is that the rich first language environment doesn't take away from English. If they're using English a lot at home it doesn't seem to impact their English positively or negatively or at all, so that's a reason not to use English. English richness however significantly impacts their English, and we find this consistently with various types of language outcome measures, with vocabulary, grammar even story-telling. So richness of the English environment which is more of a quality than a quantity measure matters immensely. And one of my grad students is working on Mandarin as a heritage language acquisition, and she finds mother tongue richness in Mandarin also influences various measures of a kid's Mandarin development. So really the story it's quality input, native speaker input quality type that you, vocabulary building, grammar building types of exposure really, really matter.

Audience Question

I wonder if you could say a little bit about adults. So, we know that adult L2 learners and heritage learners also seem to struggle with morphosyntax as well. I mean what we usually say about the difference between kids and adults is that the kids catch up to native like proficiency and quickly, and the adults maybe never do, probably never do, and if they do it's certainly not quick. But you're saying that even the typically-developing ELLs don't -- or not as quick as we thought and maybe they never get to 100% in some domains. So I wonder what do you think is the difference if anything between kids and adults?

I think that's a really interesting point, and I could talk way too long about it. But I won't. I mean what you raise is the issue of age effects in second language acquisition. So does the age at which you start learning a second language, what meaning does that have for your ultimate attainment in that language? And of course it's a much-debated question, but I think that the very current research on it seems to indicate that age is probabilistic factor when it comes to outcomes. So you get some adults who learn a second language to near native proficiency, and you get some children who still have some little traces of a foreign accent in their English even though they started to learn it as a child. But on average adults will have worse outcomes and children will have better outcomes, but it's not like you can draw a line and say okay, at this age if you learn a language before that age everything will be perfect for 100% of learners, and after that age everything will be terrible for 100% of learners. It's more like a gradual decline. I don't know if a really smooth continuous line. It might have some bumps in it, but it's more of a probabilistic kind of thing than we thought 20 years ago.
Further Reading
University of Alberta. (2015). Child English as a Second Language Resource (CHESL) Centre. Available at http://www.linguistics.ualberta.ca/CHESL_Centre.aspx
University of Alberta. (2015). Child English as a Second Language Resource (CHESL) Centre. Available at http://www.linguistics.ualberta.ca/CHESL_Centre.aspx ×