How Accents Work

I wrote the following article by request from the Ragged University, who will be hosting my upcoming talk at the Counting House on the evening of 11 September. The basic idea behind the Ragged Project is what some call ‘knowledge exchange’ — basically, bringing academics into pubs for a chat with non-academics about the research academics do. My participation was motivated by my receipt of a Small Grant award from the British Academy with funds from the Leverhulme Trust. You can find the original version of my Ragged article here. (Their version is nicely formatted; mine has footnotes!) Special thanks to my PhD students Daniel Lawrence, Ruth Friskney, and Zuzana Elliott for their feedback on earlier drafts.


 

How Accents Work

by Lauren Hall-Lew

 

The way we speak suggests to the world something about who we are: where we come from, what our gender is, what age group we’re in, and even what kind of job we might have. The parts of our identity that most define us are often marked most strongly in our accents and voices (whether we want them to be or not). Often, the aspects of our speech that do this social signalling can be very subtle. Sometimes we’re aware of them, but often they just pass by, unremarkable and unnoticed at a conscious level, but nonetheless shaping the way we interact with one another. Understanding how and why this happens is part of sociolinguistics: the study of the relationship between language and society. Sociolinguists pull together work from linguistics, sociology, psychology, and anthropology, among other things (such as speech acoustics). The results give us insights into the nature of society, the nature of speech communication, and the relationship between the two.

 

How accents are documented and analysed

Imagine that you stop a stranger to ask for directions and the person points and says simply: “It’s over there.” If you are very fluent in English, then you will be able to infer something about that person’s social identity based only on how they pronounce that phrase. The presence or absence of the ‘r’s, the quality of the vowels, or even the quality of the ‘th’ in ‘there’ (does it sound like a ‘d’?) might all serve as clues. As linguists, we need to be able to describe these clues in a systematic way, before describing how this process of social inference actually works.

Linguists call all these possible clues ‘variables’, categories composed of possible ‘variants’ that then can signal particular social qualities. The ‘r’ in words like ‘over’ or ‘there’ is an example of a very well-studied linguistic variable (called ‘rhoticity’) that’s important to descriptions of different accents of English within the UK and across the English-speaking world. The quality of ‘rhoticity’ is the variable; one variant is ‘rhotic’ and the other variant is ‘non-rhotic’ (and there might be even more variants, still). But although the presence or absence of a final ‘r’ is one of the qualities that distinguishes, for example, most Scottish accents from most English accents, it’s far from the only variable that does so (consider, for example, the vowel in the word ‘house’). Linguistically then, an ‘accent’ is just a list of those variants that distinguish one way of speaking from another way of speaking. Everyone has at least one accent, and how ‘weak’ or ‘strong’ a person’s accent either reflects how different that person’s variants are from your own, or how different that person’s variants are from some recognised ‘standard’ accent. If the person who says, “It’s over there” says each vowel and consonant differently from you, then you might think that person has a ‘heavy accent’. But if their variants are all the same as the ones you would use, you might not think about their accent at all, or you might even (mistakingly!) think that they have ‘no accent’.

A dialectologist is the kind of linguist who might produce a description of all the variables that distinguish one regional accent from another one. A sociolinguist is also interested in this, but might instead look closely at only one or two variables and see how those vary between speakers within a community: are there differences in their use according to the speaker’s age, social class, or gender? In Edinburgh, for example, recent research by one of my undergraduate students has confirmed our impression that working class men are more likely to roll their ‘r’ in a word like ‘road’ than middle class women are. At the same time, that tendency gets more complicated and interesting when you consider middle class men versus working class women, or younger people versus older people. Of course age, class, and gender are only three aspects of people’s identities, but when we hear the voice of a stranger on the street, not only do we usually infer those traits, but we also infer any number of other social qualities. Looking at which variants are used by which groups of people helps us describe what those social qualities may be, but it also raises the question of how we define a ‘group’ – which aspects of identity matter in determining which accent we use?

Practically speaking, sociolinguists conduct research by recording people talking, listening to the recordings, and counting the proportion of times different variants appear. For cases when the difference between variants is gradual rather than either-or, we might make use of acoustic measurements. For example, some people say that the stereotypical pronunciation of the vowel in the word ‘bat’ by residents of the Morningside neighbourhood of Edinburgh is ‘bet’-like (making possible slightly off-coloured jokes about ‘sacks’ in Morningside). A sociophonetician is the kind of linguist who might take precise measurements of someone’s vowel in words like ‘bat’ to see just how close it really is to their vowel in words like ‘bet’. If this is done across a representative sample of members of a community then we often find very regular statistical patterns between peoples’ accents and their social identities. We also find that these patterns exist even when people aren’t consciously aware of them.

That kind of research can be described as the sociolinguistic study of speech production. It’s also common to study the listening side of things, which is the study of speech perception. Sociolinguists who study the perception of accents might look at what conscious attitudes people have, playing recordings of speech for listeners and then asking listeners to comment on the kind of person they think the speaker is. We might intuitively think that we start out with neutral opinions about people and then judge them positive or negatively based on how they speak. But linguistic research has shown that if we view a group of people negatively, then we’ll also often extend those negative evaluations to the way they speak. In many cases, saying disparaging things about a particular accent is just a socially acceptable way to say disparaging things about the people who speak with that accent. In addition to this kind of language attitudes research, there has also been some interesting work on subconscious speech perception, including work on ‘accent hallucination’, which I’ll say more about in the next section.

 

How an accent provides clues to a person’s identity

So what do we think we know about how accents signal social identities? Without getting too deep in the theory of it, the basic idea is that when we acquire language, we not only learn what words mean or refer to (‘referential meaning’) but we acquire information about how those words are most often used, such as which contexts they occur in (‘pragmatic meaning’) or what kind of person is likely to use them (‘social meaning’). As we learn how to speak, we learn the pronunciations of those we spend the most time with, and as we get older, we adjust those pronunciations as our social networks change (such as when we shift from spending more time with family, as a child, to spending more time with friends, as a teenager). Although adults take on new ways of speaking less often, or slower, than children do, nearly all of us do shift a bit when talking with different kinds of people, regardless of whether we notice that we’re doing so or not.

One of the most famous examples of this is work by Jonathan Harrington and colleagues on changes in the speech of Queen Elizabeth II over her lifetime.[1] Their study looked at her Christmas broadcasts since the 1950s and compared the pronunciations of all of her vowels over time. They found that, overall, her speech had shifted gradually away from Received Pronunciation and towards a more widely-spoken variety linguists refer to as Southern Standard British English.

But changes over the course of one’s lifespan are not remarked upon nearly as often as changes that happen from one moment to the next, or what sociolinguists call ‘style-shifting’ or ‘code-switching’. This is the phenomenon of someone’s accent changing when they speak to a stranger on the telephone, or speak lovingly to a baby or a pet. Observers often take notice of someone else’s style-shifting when it happens, and can poke fun at it.[2] In some instances an observer may accuse someone of putting on a false accent, and thus a false social identity, even though style-shifting is a ubiquitous feature of human communication.

While it’s perhaps not so surprising that linguistic forms can point to social identities, here’s an example of just how deeply ingrained these associations are in all of us: ‘accent hallucination’. In a study conducted by Okim Kang (Associate Professor, Northern Arizona University) and Donald L. Rubin (Professor Emeritus, University of Georgia) and published in 2009,[3] 158 American university students sat together in a lecture hall and listened to a recorded lecture on a generic science topic. The lecture was given by a native speaker of U.S. English who had been described by his friends as “a particularly clear speaker.” While they listened, the students saw a picture of someone who they were meant to think was the person giving the lecture. Part of the time they saw a White face and the description “native speaker” and part of the time they saw an East Asian face and the description “non-native speaker” (there were various measures taken to make it believable that the speakers were actually different people). The listeners were then asked to write down some of the sentences that they heard the speaker say. The listeners were also asked how much regular interaction they had, personally, with non-native English speakers. Not only did they find that listeners said the “non-native speaker” had a non-native accent, but they also found that those listeners who interacted less often with non-native speakers were actually less successful at writing down the exact words that the “non-native speaker” said.

Remember: they were actually listening to the same speaker the whole time; it was only their perception of who they were listening to that changed. This study reveals that our social expectations alone are enough to make us think that someone has a non-native accent, even when they don’t, and thinking that someone has a non-native accent may actually make it more difficult for us to comprehend what they’re saying, regardless of what they sound like. The findings take the relationship between accents and social identities to another level, where it’s not only just a question of how a particular variant conveys something about the speaker’s identity, but the listener’s own inferences about the speaker also influence what variants the listener hears.

 

Uncharted areas of accent and identity research

So how does speech actually convey speaker identity? In attempting to understand this question, sociolinguists have looked at the relationship between linguistic variants and regional, class, age, and gender identities. These are the staples of sociolinguistic research, and others such as ethnicity, race, religion, and sexual orientation have also been studied in great depth in some places. Some work has even taken a very fine-grained look at language use in very specific social groups, such as a friendship cohort of six middle-class Muslim high school girls in Sheffield whose friendship was primarily defined by an interest in the ‘Twilight’ novels, and see how their language use compares to that of members of other similarly specific cohorts.[4]

Another angle we can take for approaching this question is to look at social identities that have not been previously considered, such as what political party a person belongs to. In this talk, I will present work-in-progress from my latest project that asks if politicians’ accents can tell us anything about their politics. Some recent research shows that politicians present an interesting case, presumably because their political identity is particularly important to their social identity, and also possibly because they speak professionally and therefore present their identity in an on-record, public way more often than most of us do.

In my talk, I’ll first present work from an earlier study I did on Members of the United States House of Representatives,[5] and then I’ll move on to talk about my current work on Scottish Members of the Parliament of the United Kingdom.[6] While the specific identities of the political parties are very different in those two contexts, in both cases political party is a strong predictor of the way those politicians pronounce their vowels. Typically when we think of the speech of politicians we think of conscious rhetorical decisions designed to manipulate and spin. In contrast, in my talk I will present the argument that politicians subconsciously convey their political identity through their pronunciation patterns, just in the same way that all of us subconsciously convey various aspects of our identity through our speech. In short, politicians are people too, and what their speech patterns look like provides us with insights on how speech works for all of us.

 

[1] http://www.nature.com/nature/journal/v408/n6815/full/408927a0.html

[2] For a great example, see: http://www.youtube.com/watch?v=JzprLDmdRlc

[3] http://jls.sagepub.com/content/28/4/441 (Note that ‘accent hallucination’ is my term, not theirs, and not an official term.)

[4] http://etheses.whiterose.ac.uk/5864/

[5] http://americanspeech.dukejournals.org/content/85/1/91.abstract

[6] http://www.political-voices.lel.ed.ac.uk/

Advertisements

About vocalised

http://www.laurenhall-lew.com
This entry was posted in Uncategorized and tagged , , , , , , , , . Bookmark the permalink.

Leave a Reply

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

WordPress.com Logo

You are commenting using your WordPress.com 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