May 01, 2005

The meaning of eh

Sarosh Kwaja , responding to an earlier post about gender and tags, sent email this morning to ask about the function of the stereotypically Canadian tag "eh". I didn't know the answer, so I did a few minutes of poking around.

Searching on Google Scholar for {Canadian tag eh} turned up as the first hit an abstract by Elaine Gold and Mireille Tremblay, "Canadian English, Eh? Canadian French, Hein?", which in turn includes a bibliography full of interesting-looking stuff, including most recently Elaine Gold, "Canadian Eh?: A Survey of Contemporary Use", Annual Conference of the Canadian Linguistic Association, University of Manitoba, Winnipeg (2004). Searching again for {Elaine Gold} turned up a copy of the cited paper itself.

Gold's paper begins

Although eh is widely considered to be a marker of Canadian speech, there has been little research done into its use or meaning and none in the last 25 years.

and takes up the question of whether eh is really a Canadianism or not:

Avis wrote an article entitled 'So eh? is Canadian, eh?' in which he argued against eh being a Canadianism, citing examples of eh from English writings around the world. However, he undermined his own argument, in that he could find only Canadian examples of two types of eh: the narrative eh illustrated in (3), and eh used as a reinforcement of an exclamation in (4):

(3) "He's holding on to a firehose, eh? The thing is jumping all over the place, eh, and he can hardly hold onto it eh? Well, he finally loses control of it, eh, and the water knocks down half a dozen bystanders." (Avis: 103)
(4) "How about that, eh?" (Avis: 99)

The body of Gold's (excellent) paper "focuses on the results of a survey of eh usage conducted on students at the University of Toronto", covering 30 males and 61 females who were Canadian-born, and 5 males and 13 females who had immigrated to Canada within the past 5 years. All were under 30 and enrolled in the introductory linguistics course at the University of Toronto. The participants were asked about experience with, usage of, and attitudes towards various examples of eh:

Type of eh Sample sentence
1. Statements of opinion Nice day, eh?
2. Statements of fact It goes over here, eh?
3. Commands Open the window, eh?
Think about it, eh?
4. Exclamations What a game, eh?
5. Questions What are they trying to do, eh?
6. To mean 'pardon' Eh? What did you say?
7. In fixed expressions Thanks, eh?
I know, eh?
8. Insults You're a real snob, eh?
9. Accusations You took the last piece, eh?
10. Telling a story This guy is up on the 27th floor, eh? and then he gets out on the ledge, eh?

You can read the paper to see the details. There were few sex differences among native Canadians in answers to the "have you heard" questions -- significance levels are not presented, so maybe there were none, it's hard to tell. There was little overall sex difference in answer to the "do you use" questions -- 49% sometimes+often for males, 45% for females; 15% often for males, 14% for females.

There were larger differences in the responses for some particular cases: 92% of the females said they used the expression "I know, eh?" sometimes or often, while only 72% of the males admitted to this; on the other hand, only 13% of the females reported use of eh in commands like "Open the window, eh?", while 45% of the males did. (Though this last difference may have more to do with the rather brusque tone of the command, rather than the use of eh...)

It's clear from the attitude survey that some of the uses of eh are stigmatized. In particular, about half the Canadian respondents reported a negative attitude towards "narrative eh" (34% of males, 57% of females, 49% overall), while very few reported a positive attitude (3% of males, 2% of females, 2% overall). In general, the negatively-evaluated cases tend also to be the ones where males reported higher usage than females did.

The recent immigrants seem to be picking up eh pretty fast, although their overall rate of (self-reported) usage was somewhat lower, and they mostly don't get some of the subtleties, especially the use with fixed expressions:

As was noted with recognition rates, the immigrants' usage rates do not reflect the pattern of use they hear from native speakers. This can be seen in Table 14 where the responses of the native Canadian speakers regarding use (Table 4) are compared with the use reported by the new immigrants. On the one hand, immigrant speakers do not seem to be picking up set expressions like Thanks, eh? [11% vs. 53%] and I know, eh? [28% vs. 85%]. On the other hand, their use of eh for pardon is higher than that of native speakers [56% vs. 39%].

Gold offers an interesting and plausible theory about what's going on with the immigrants:

One explanation for the immigrants' different pattern of use might lie in their interpretation of the function of eh. It is possible that the immigrants are interpreting eh as strictly as a question particle, equivalent to tags like 'isn't it' or 'don't you think'. This interpretation of eh is consistent with eh following statements of opinion, accusations or fact; these can all be rephrased as questions, such as Nice day, isn't it? or It goes over here, doesn't it? However, this question particle meaning of eh is not compatible with expressions like Thanks or I know which make no sense when rephrased as questions. This would explain why immigrants are not picking up these expressions as quickly as some of the others, even though they are exposed frequently to them.

Gold's survey did not address the questions that Sarosh asked me about, which dealt with the distinction, originally made by Holmes 1984, between modal tags (which indicate speaker uncertainty) and affective tags (which may be softeners, conventionally mitigating the force of possibly-impolite or aggressive remarks, or facilitative tags, which invite the listener to take a conversational turn to comment on the speaker's contribution). Nor did Gold deal with the effects on eh usage of the structure of the interaction, which Cameron et al. (1988) found to interact with speaker sex in a crucial way. In their study, men were found to use modal tags more often than women, and affective tags less often; while for both sexes, the affective tags were only used by "powerful" participants (those "responsible for the conduct of the talk", like a teacher or talk-show host). (See the cited gender and tags post for more details.)

There are several different sorts of things at issue in these different studies: the form and function of the phrase to which a tag like eh is appended (e.g. commands, questions, fixed expressions, etc.); the discourse context (e.g. narrative eh); the function of the tag itself (e.g. Holmes' distinctions among uncertainty, softening and turn-taking); the relationships among the speakers (e.g. whether one of them is institutionally "in charge" of the interaction). Cross this with formality (and other aspects of "register"), age, sex, geography, ethnicity and so on, and you've got plenty of topics for empirical research. More important, the results of such research bear on basic questions about language, communication and identity.

Surveys, though very useful, are not by themselves an adequate way to study such patterns of usage. There are several serious problems with trying to resolve such problems using intuitions, whether the intuitions of a linguist or the intuitions of survey respondents. It's not just that people tend to underestimate their usage of stigmatized features, though this is true. It's also true that people may have no useful intuitions at all about some crucial factors. An even bigger problem is that the survey designer may not know to ask the most important questions. And the biggest problem of all is that we humans are incorrigible theorizers. As soon as we start reflecting about our own behavioral dispositions, we start to organize our impressions and reactions into more or less coherent patterns, whose relationship to our actual behavior can be remote.

This used to be a daunting and even depressing line of thinking, but now it's become invigorating and exciting. For those of use who are interested in language use, this is a great time to be alive and working. We've got large digital archives of speech and text, we've got (semi-)automatic techniques for search and analysis, we've got great tools for exploratory analysis of large bodies of data. Go to it, eh?

Posted by Mark Liberman at May 1, 2005 01:11 PM