Simple tips on linguistic analysis (& a change in the Observer’s Paradox?)


Here’s a brief look at ‘data collection’ in linguistic research, what it might entail, and a quick theory on how the Observer’s Paradox might be changing.

Data collection in linguistics can take a number of forms. If we want to know your opinion on accents, we can record some, and then ask you to listen to and rate them on a scale. If we want to know how language affects the brain, we can ask you to respond to questions in an MRI scanner. As a rule, it’s likely that the thing we’re looking for is not necessarily the thing that you think we’re looking for. Indeed, we might not actually tell you before the research what it is that we’re looking for! This isn’t to deceive you, but it’s to reduce the likelihood of you providing an answer that may not be representative of how you really use language.

Let’s take an oversimplified example of a question that may be asked in linguistic research:

How do you pronounce the word ‘bath?’

If you’re a British English speaker, this well-researched dialectal divide is known to be North-South, with a middle somewhere around the geographical Midlands in England. In brief, those in the North use a ‘short a’ and those in the south use a ‘long a.’ However, there are a few things wrong with asking a question like the one above.

Firstly, just by looking at the question, I’ve asked you to isolate a word rather that listening to you say it in a string of other words. In general, everyday communication, it’s extremely unlikely that you’d just be speaking one word at time, so your pronunciation of the word on its own may sound very different to how you actually pronounce it in day-to-day communication. To be more accurate in data collection, we would need to rethink the approach (e.g. digging into corpora, using extended phrases with the keyword embedded in them, or presenting visual stimuli and asking people to describe what they see).

On top of this, this split in ‘a’ pronunciation is common knowledge, and your previous knowledge of the North-South divide might come along with some ‘baggage.’ There are elements of prestige and group membership associated with the ‘short a’/’long a’ usage. So if I were to ask you the above question as it stands, you might adapt your answer based on this knowledge, and on how you would want to appear to me.

Your answer will, although you might not think it at the time, also be based on how I appear to you. This is due a huge number of factors, including how we know each other, the way I present the task to you, and even how I dress on the day! For want of a lengthy academic description, we can call this overall effect the Observer’s Paradox (click here to see a brief PowerPoint with information on this).

Over the decades, researchers have come up with ways to lessen* the effect of what Labov calls Observer’s Paradox. This ranges from the anonymisation of participants, to using people within the group being researched to act as proxy-researchers. Technology has also helped out, too: we can set up a recording device and not have to be in the room at all, therefore removing the ‘observer’ altogether…

…or does it? Even though another animate being may not be in the room with you, there’s a chance that knowing that a device is recording your language will cause you to feel observed, whether it’s as small as a modern audio recording device pointing towards your face, or as obtrusive as an MRI scanner enveloping your entire upper-half.

Research on sports concussions

“Just pop yourself in there and speak naturally, please…”

Saying that, I think a change in the effect of the Observer’s Paradox might be coming in, at least in certain social groups. This thought came about as I was reading about some work done on British Sign Language (BSL) from the late 70s through to mid-80s in Scotland. Without worrying too much how the question was posed to the participants, researchers filmed these people using BSL (with what I can only assume would be a pretty big piece of equipment, seeing as smaller handheld cameras weren’t readily available until at least the mid-80s… talk about feeling ‘observed!’). This must have felt somewhat peculiar for the participants, too, as being filmed wasn’t an everyday occurrence.

Compare that scenario to now. It’s likely that you’re reading (or could read) these words on a device that has a camera pointing directly towards you. So many devices around you can be used to take a photo, a selfie, or a quick recording of you eating as many marshmallows as possible and then downing three cans of Red Bull** before posting it on YouTube. In short, the opportunity to be caught on camera is now at least a daily experience, and some may revel in any opportunity to do so. This isn’t to say that a camera pointed at anyone nowadays would make them feel immediately comfortable, but it must be having an effect on how we perceive such devices and the result linguistic production.

The crux of the post is this: I believe that those who are more ‘used to’ using modern technology are less likely to feel ‘threatened’ (and therefore change their language) when being recorded. When recording and analysing sign language, we can’t anonymise the video as we need to be able to see the whole upper-half of the body in order to understand the communication, so we don’t have recourse to anonymity per se. However, a paper of mine that is currently under review talks about the many members of the worldwide sign language-using community who are posting videos of themselves on a regular basis, and the effects this might have on confidence and linguistic production. It effectively comes down to how ‘seasoned’ you are at filming yourself: the more you do it, the less likely you are to feel intimidated by it.

We can control certain aspects of the Observer’s Paradox more than others, as mentioned above, but the recording issue was always that one factor that was out of our hands. I don’t think it will be too long until we have more control over this, too. How this can be measured and argued for, however, might be a bit of a paradox in itself.

– – – – – – – – – –

* Important to note the use of ‘lessen’ rather than ‘remove,’ as it’s extremely unlikely that you’d ever get perfectly naturalistic data when people know you’re researching them (unless you do so covertly, which is extremely unethical and not recommended at all!)

** Again, neither recommended nor advised, but probably available on YouTube with some good searching skills…


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s