Dr Neil W. Kirk — Abertay University
Previous language switching studies with “bidialectal” speakers of closely related language varieties has revealed evidence for very similar language control processes as used by more traditionally recognised bilinguals. In this registered report study, we extended this investigation to determine whether speakers of (Scottish Standard) English and Dundonian Scots also displayed a mixing benefit as has been found for bilinguals when they can mix their languages freely. Using an online voluntary language switching paradigm, participants were cued to use one language variety at a time (i.e. naming pictures of objects in English only, and Dundonian only) and were also given blocks in which they could freely name the objects in either variety. While previous research has revealed faster picture naming for bilinguals in the voluntary vs cued language blocks, this was not the case for “bidialectals”. This suggests that the language control processes used by these speakers is not entirely similar to those used by bilinguals. In this talk, I will also discuss the strategies and experiences of using the media to recruit these participants whose language experiences would not be captured by traditional recruitment platforms.
Jo Evershed 0:00
We’ve got Neil, Neil do come across and share your slides. Neil and I first met, oh, gosh, a few years ago, Neil has been up in Dundee studying different dialects of different languages. He’s done all sorts of amazing things, including using the local radio stations to recruit his participants, which is one of the reasons I’ve asked him to speak today. So don’t let him leave the stage without sharing how he got that to work and what impact it had on his research. Neil, over to you.
Neil Kirk 0:30
And I just check that you’re seeing my title slide only? Yeah, yeah, no, no. Perfect. Because, yeah, I’ve not been using zoom that much now that we’re able to get out the house again. So bit rusty. Yeah, thanks, Jo. I’m going to talk today about some work that we’ve been running with. I’ve explained this term in a minute, but what we refer to as bidialectels, and we’re running language switching paradigms, essentially trying to investigate the cognitive processes that are involved in using more than one language variety. And the work I’m going to present today has been published as a stage one registered report.
So as someone who was alluding to the things like the design hypothesis, or analysis plan or peer reviewed, I’m not going to go too heavy into sort of theory or anything in my talk, this Papers, Please, if you’re interested in more, it’s a good starting point. So maybe I will go into a little bit of theory. So models of bilingual language control propose that a bilinguals languages are always active, and they’re always competing for selection to some extent. So when a bilingual plans to speak in that particular target language, they have to inhibit or suppress all the items from the other non target language from…they have to inhibit them to prevent any cross language interference.
So in this example, if you want to name this animal, the bilinguals have to inhibit the German word to allow them to select the English word dog. And the amount of inhibition that’s required to do this is relative to how strongly represented or dominant that language is. So in this case, let’s say English is the more dominant language, it would take more energy, in a sense to suppress this to allow them to speak the German word. However, it can take a while for that inhibition to wear off, meaning that in relative terms, it can take longer to switch back to your dominant language, compared to switching into your less dominant language.
So these effects are commonly measured using language switching paradigms. And a typical setup here is that a participant is asked to name pictures of objects, and usually given some kind of cue as to which language variety they should name the object in. We’re interested in how accurate they are, but predominantly, we’re interested in how quickly they can do this. And by measuring the response times, we can observe the effects of things like naming in your dominant versus your non dominant language, and the effects of switching from language A to B or from language B to A and so on.
Generally, what you find in these paradigms is that switching between varieties takes longer to name the object than if you had already sort of remained within the same language setting. So, there are different types of language environments a bilingual could find themselves in which might require different applications of language control. So for example, a bilingual could find themselves in a single language environment. And what they can do here is proactively apply language control to keep a higher activation level of one language, because they might have no expectation that they will require the use of the other language.
And in language switching paradigms, we can tap into this by using single language blocks. So this is where the participant would be informed that they will only name in one language have it throat and again, they would have no expectation of having to shift gears and to use a different variety.
A bilingual could also find themselves in a mixed language environment where they have to reactively switch languages on the fly depending on who they are communicating with at any given moment. And we can measure this in language switching paradigms by using mixed language blocks. Usually this involves the participant being cued are still which language variety the to name, the object, then often that’s a colour cue.
And these are often presented in an unpredictable sequence so the bilingual can’t anticipate what language we’ll have to use on the next trial until that picture, and its corresponding border appear. These kinds of blocks give us different types of trials. So we have within these, what we’d call repetition trials, which have trials in which you have named in a particular language having stayed within the same variety compared to the previous trial. And we also have switch trials that require you to switch from one variety to the other, and basically shifting these different mental settings. And we can also, this can involve switching from language A to B, or from language B to A, and again, trials that involve switching tend to take longer than trials that require you to remain within the same variety.
Now, a common finding in bilingual research is that there is a mixing cost associated with having to use more than one variety. And what this… basically what we generally find is that trials in single language blocks are named faster than repetition trials from mixed language blocks, despite both types of trials basically involving having named within the same variety from one trial to the next. But basically, in a mixed language block the anticipation of potentially having to use your other variety. And to switch to that means that both languages maintain a higher level of activation, which is causing these slower responses. So that’s the mixing cost, you’re slower when you’re having to mix your languages.
There’s also another type of language setting a bilingual could find themselves in where they’re presented with speakers who speak both their languages. So they’re basically able to use the languages freely and mix them, sometimes even within the same sentence. So in a sense, this kind of setting might not require any kind of language control, because it’s just a case of an anything goes, it doesn’t matter what word I use, from what language, who I’m communicating with will understand.
And in terms of language switching, we can investigate this using voluntary language switching blocks. So this is where the bilingual be told to name, each picture with whatever comes to mind doesn’t matter which language just whatever, whatever names comes to mind first. So in this case, they might name that first picture in language A, they might then stay within that language and name the next picture within languages as well, they might then switch into language B. And again, that could then remain within that language. What we then do is subsequently code, what language they used for each trial, and whether that trial forms a repetition or a switch.
And what we find here is a mixing benefit of in some studies that we saw previous studies by by linguist find a mixing benefit when you are able to voluntarily choose the language. And so basically, in this case, the bilinguals are often faster to name repetition trials from the mixed block than they are to name their single language trials from this single language block, presumably because they’re not having to employ any language control in the voluntary block. And in the single language block, they are having to suppress the other variety, because they’re, they know they won’t be using it. Whereas in the mixed voluntary block, they’re free to use whatever comes to mind.
So we’ve previously investigated whether similar effects exist for, as have been found for bilinguals, for speakers of closely related language varieties. And in our research, we’ve used speakers of Scots, which is a Germanic variety, closely related to English it has from the last census, something like 1.5 million speakers. It’s recognised by the European Charter for regional or minority languages as a minority language. But its status as a language is often questioned and even the speakers themselves might not regard themselves as bilingual.
In our previous studies, using adaptations of language switching paradigms, we found lots of similar effects as are found for bilinguals. So essentially, we’re seeing in these participants bilingual effects, even though they would regard themselves as monolingual. And why why do they do that? Well, Scott is a sort of stigmatised variety, it’s often considered quite low status. I’ve just thrown up a couple of examples of tweets where in response to somebody who was speaking Scott. So you can see you know, it’s not considered you know, people don’t are not very favourable towards it sometimes.
And even speakers themselves, don’t necessarily regard it as a different language and this government report from a few years back so 64% of respondents said that they didn’t think of Scots as a language, it was just a way of speaking. And in a sense, we can think of it more like a register, you know, it’s kind of like a formal or informal way of talking, in that sense more of an informal way of talking.
So in the present study, we wanted to investigate whether our so called bidialectal speakers would display this mixing benefit that has been found for more traditionally recognised bilinguals. So to do this, we used speakers of Scottish Standard English, which is essentially what I’m speaking though, and dundonian Scots, which is the the Scots dialect spoken in Dundee.
And, as I mentioned, this was a registered report, we ended up with 46 participants, which was our target. They reported using dundonian around 26% of the time. As Jo has said, I’ve used the local media, radio, newspapers, to and social media to recruit our participants, and we paid them 15 points for participating. I should say that recruitment platforms like prolific do actually have questions in the demographics about Scots, we decided not to use that because from our experience, a lot of speakers themselves don’t even know what Scots is, they might think of it as something that other people in other parts of Scotland speak. And because we were using stimuli that was related to dundonian Scots, it doesn’t, prolific doesn’t drill that deep. So you know, you don’t have to travel very far for people to start using different names and stuff. So to keep it you know, within this dialect region we’ve recruited ourselves.
One drawback to using the local media is that you can never be sure of what headlines are going to put your picture next to. So just a couple of unfortunate incidents where my face has appeared next to these headlines, I really should get lawyers involved at some point.
So we use gorilla to run this language switching paradigm and materials are available as open materials. And this consisted of participants completing a single language block in one variety, they then moved on to completing four voluntary blocks with they could pick whichever item came to mind. And then they would finish by completing another single language book in the other variety, and we counterbalance the order of those across the participants.
We used in this study, only used non-cognate items. So these items that don’t have any phonological overlap with each other. So that’d be things like girl and Lassie, ears and lugs, earache and forkytaily, and sandwich and piece. And participants completed this at home. It only required the use of a device that had a microphone and mostly desktop PCs and laptops. And because we’re capturing audio, we required a quiet environment.
So what we then did with the audio trials is we took them and re uploaded them into a new task so that we could code their accuracy for the single language blocks where they were instructed what they had to name. And for the voluntary blocks, we could then code which variety they had used in that, for that trial. And we also uploaded the audio files to chronset which extracts their response times.
And combining these together basically gives us what we need for our analysis. So what did we find? In terms of this mixing analysis, we are comparing here, the trials from the single language blocks with the repetition trials from the voluntary blocks, and we use linear mixed effects models to do this. We find a main effect of language variety in that the standard English items were named more slowly than the dialect items. Bearing in mind, prior to the task, we’ve controlled for things like word length, and things like that. So so people were slower to name in the standard than they were the dialect.
We didn’t find any effects of block type. So the single language trials and the voluntary repetition trials were approximately pretty much the same. And we didn’t find an interaction between language variety and block type either.
So our take home here is that we didn’t find any evidence for a mixing benefit in this group of speakers. But nor did we find any mixing cost. I haven’t reported it here but we also measured the switching costs in the voluntary blocks and we did find the same effect that you would find for bilinguals. So You know, some some aspects replicated, but we didn’t observe any mixing benefit or cost here. And although all our previous studies that we’ve done with these types of bidilectcals have shown effects that we find for bilinguals, this is the first time we’ve not found a similar effect. So it suggests that potentially the by dialectal, and bilingual language control processes are not necessarily identical. But this has given us more work for us to investigate this further. So I just like to thank my collaborator, Dr. Mathieu Declerck, and the Carnegie trust, who funded this research. And that’s me.
Jo Evershed 15:44
Neil, that’s fabulous. You snuck it in there in the middle, did I get it right that you got these auditory recordings online? And then you extracted those auditory recordings, and then you put those back on prolific and got prolific workers to score them from
Neil Kirk 15:57
No, no, no. In this case? I mean, we could do that. It’s, it’s crossed my mind to do that. No, we for this purpose coded them ourselves. But used a task in gorilla to do that, because then it’s easy to share amongst people. And it’s Yeah,
Jo Evershed 16:16
and you can get three ratings for each one and see how that
Neil Kirk 16:18
Exactly, yeah, so I’ve done that. We’ve done that in all our experiments, basically create a, it’s also available as open materials. Yeah, like an accuracy coding task that, again, just uses the platform, because, you know, gorilla is so easy to use.
Jo Evershed 16:34
Ah, thanks, Neil. No, I do know other researchers who do put their accuracy coding back onto prolific and get their prolific workers to do it for them, which seems to work really well as well. And then they are the ones where they, they sit together, they let them through. And then when they only when there’s a disparity, they go and check them manually. They do them themselves. So that works quite well.
Now, you didn’t say a lot about what it was like recruiting with. Using media, I know you’ve recruited in the newspapers and using radio, can you just say a little bit more about that. And while Neil is chatting, if you’ve got other questions, put them into the q&a. And hopefully, I will get around to them.
Neil Kirk 17:09
So one aspect of this work is it seems to capture sort of the public’s interest. And anytime we put a press release about this, we always get some kind of coverage. So it always makes the sort of, I guess, a bit of a quirky new story sometimes. So there’s been a couple of experiences when I first got this funding and launched my lab, and we were running some studies, I did some radio interviews on national radio, which led to, I think, two participants responding and neither had usable data. So almost like the bigger that we the the sort of broader the reach, the less beneficial it was because you know, then it wasn’t so relevant to everyone.
When we’ve done this, for, we actually run two projects at the same time. So we recruited for both of them this way. I certainly found a lot of interest, a lot of emails coming from, from people who are interested, but actually the follow through of people to to complete it was it was maybe only about 50%. So I’m here today to hear other people’s tips for recruitment, because I’m finding when you when I don’t use a recruitment platform, even when I’m paying, I have to say quite generously participation. Even using media, sometimes there are still challenges with getting people who have shown an interest to follow through to completion. So I tried a few little tips like bit of nudging a bit of you know, this. You know, that might only be the study might only run until we’ve got an x number.
Yeah, kind of didn’t work. I think all it did was backfire slightly, and people thought, Oh, I’ve missed, I’ve missed it. I should have done it yesterday. So it’s, it has its challenges, but certainly we wouldn’t have I think recruited in the time period that we did without that media coverage. It was a huge help.
Jo Evershed 19:07
Okay, so it helps but it’s not it’s not the the Nirvana that we’re all hoping for.
Neil Kirk 19:14
Potential. I think the other aspect is what I’ve noticed, with this type of speech production research, I think people are just that bit more hesitant about doing I think people can be slightly more self conscious about doing this type of work. It also requires that they have, you know, a particular type of setup, and that they can do it in a place where there’s no distractions and you know, often at home, that’s that isn’t the case. And so that can sometimes I think, be a slight barrier. It’s yeah, I think it’s with more comprehension based or where they weren’t having to produce I think we would have got a, been more successful,
Jo Evershed 19:52
I think, yeah, it’s sort of the death by 1000 cuts, isn’t it that you lose a few because of this and a few because of that, and each time you’re like, oh, so I’ve only got a few. Have you ever tried recruiting with charity partners like are there other Scottish charities that focus particularly on regions?
Neil Kirk 20:07
No, that’s really a good shout. I’ve got a few connections because because this this language variety forms part of the Scots language, I’ve got a few connections like that would have been able to recruit. Basically, with any of these tasks, we use every tactic we can think of. So there’s a few things like that. But yeah, that’s certainly something, especially now that we might be able to not everything has to be done online. In terms of how you reach people in the first place. Yeah, something to think about for sure.
Jo Evershed 20:40
There’s a question for you in the q&a from Sachin to questions about the procedure, did you record a participants responses in a block from the first or last trial? And how did you decide the onset of each vocal response?
Neil Kirk 20:52
So what the audio, was it called the audio recording zone does is it captures every response on a trial by trial basis. So we didn’t have to then splice them into individual trials, we get an output that is basically one an audio output for every single trial. And we use the chronset system, which is an online server, you can upload your files into an extracts the response times, in some of our previous studies, what we’ve done is taken a sample of the audio and we’ve manually measured them as well. And we’ve shown that, you know, they’re pretty much bang on. So I’m quite trusting of the chronset system to give me some accurate responses. In our accuracy coding, what we also have is if we can detect extraneous noise, we’ll code that as a, basically a no or other signs so we know to exclude it because the reaction time measurement might have been messed up. So we have that kind of extra quality check when we’re doing an accuracy coding.
Jo Evershed 21:59
Brilliant. I’m sure there are lots of auditory resources here who are going to be delighted to hear about concepts. So audience in the chat if you can share what was the one message you took away from Neil’s talk today that you’re going to take back to your research what was helpful, that would be fantastic. That’s your thank you to Neil. Neil, you’ve been great.