Diversification: A solution to the dichotomy of intuition versus observation
Ecole Normale Supérieure Paris-Saclay, France
Sociolinguistic investigations that primarily rely on elicited judgment data are often easy prey for the contention that intuitions (can) fail, entailing that “the use of intuitive judgments of acceptability as the sole basis for linguistic generalization may incorporate serious errors into the argument” (Labov 1996: 100). There is much strength in this idea. If even the highly specialized knowledge that a linguist summons in the analysis of his native language cannot be taken for granted in an epistemological sense, how could we trust the insight of everyday speakers, who may be prone to prescription, vagueness, or both, in their appreciation of individual features? Theoretically, intuitive judgments thus seem hard-pressed to stand on their own.
Yet judgment elicitation is still around, and far from being extinct: Labov predicted it to be something on which “linguistic analysis will always rely to a large extent” (77). Elicitation-based interviews and questionnaires remain regular methods in sociolinguistic studies, as suggested by Krug et al. (2013: 8). These are frequently deployed to collect substantial amounts of fine-grained quantifiable judgment data. The key question that needs to be addressed is whether the elicitation-based approach is necessarily a demerit that should be avoided. If not, can we devise better ways of implementing this potentially controversial method in future linguistic research? Although I do not claim to offer entirely novel and comprehensive answers to these questions, I would like to propose a few ideas for a useful and continued conversation which, by the way, is not restricted to dialectology, or even sociolinguistics more broadly.
The question posed above seems to imply that intuition-based judgments are not the ideal source of data in linguistic research. They are viewed as intrusive due to the degree of control they require in experimental design and execution. The method that is preferred over intuition-based judgments concerns corpus studies and the recording of spontaneous speech, as they help more easily in “reducing the effects of observation, defining the linguistic variable and variable constraints upon its frequency, [and] quantitative measures of the fit of theoretical model to observation and significance of the results” (Labov 1996: 78). We may add that even then, all problems are not solved; it is much more difficult than one might think to stay clear of the observer’s paradox when recording speech.
However, if we rigidly endorsed this methodological dichotomy (intuition versus observation), it would follow that a large number of linguistic phenomena are entirely out of our reach, due to various factors such as rarity, recession, and the like. Unless we devised an uncontroversial way of collecting data (recorded speech) from communities without them being aware of being observed – which might subsequently evince a kind of Orwellian impulse more than better linguistic fieldwork – then we would have to give these phenomena up altogether, especially in the field of syntax, an area traditionally sidelined in research on language variation and change, until recently at least.
This seems to explain why judgment elicitation is still used frequently in such circumstances: it would essentially be the least ethically problematic source of collecting data. But there may be ways of turning these scientific and ethical negatives into a positive, and reasons not to consider the method simply as a plan B anymore. How can we do this? The elements I have been thinking of could be united under the provisional label of “(intensive and extensive) diversification”. The ideas are fairly simple and I will describe them below.
First of all, proper reflection is due on the very act of elicitation, not least on the kind of information that constitutes an “intuition”. An attempt to precisely define intuition would take us on a difficult cognitive and philosophical route, but we could provisionally agree that it is a reasonable assumption about a concrete or abstract object; in our case for instance, a linguistic construction. The obvious problem that arises is that we cannot easily distinguish intuition from a judgment induced by subjective values, or by mistaken preconceptions more generally. For example, the speaker of a dialect might indicate that a construction is unacceptable simply because they were told at school it was part of “incorrect grammar” in spite of its actual entrenchment; or their answers could be predicated on a different interpretation of the instruction than the one the researcher set out with.
Although this is a major drawback, one should not entirely dismiss intuitions. A possible solution would be to look for other ways of investigating the reliability of intuitive knowledge instead of probing the speaker’s mind. This is where I introduce the notion of “intensive diversification”. Namely, we could try to increase the number and diversity of tasks inside an elicitation set to reveal sizeable parts of the “rational backbone” of an intuition: in other words, designing a tailored set of sub-tasks aimed at cross-examining the main judgment, and establishing its coherence, or lack thereof. Arguably, the typical orientation of these subtasks would be quantitative; but in fact, qualitative components could be welcomed, and even encouraged, as apt to bring unique contextual information in the evaluation of candidates for intuition status, for instance in terms of pragmatic and social dimensions of a feature.
Through a balanced appraisal, and a tailored design of elicitation for each specific objective at hand under this principle, it may thus be possible to accomplish two seemingly exclusive, but actually complementary aims: on the one hand, to convincingly circumscribe the “limits of awareness” (a theoretical enterprise introduced by Silverstein 1981, and developed, for instance, by Preston 1996 and Mertz & Yovel 2009) in order to avoid the biggest pitfalls; on the other, to pinpoint and excavate the true benefits of good intuitions, in a perspective comparable to that of “folk linguistics” (Niedzielski & Preston 2000) specified for grammatical enquiry. Multidimensionality seems to be the key for better elicitation per se.
The principle of diversification could then be developed and made “extensive” by incorporating other methods. Recent research in language variation and change has suggested that combining methodological approaches may lead to richer results in quite a few instances (Krug et al. 2013). From such a viewpoint, judgment elicitation gains added value insofar as it throws its own specific light on results from protocols controlled to a lesser degree. In the case of rarer syntactic constructions for example, elicitation can bolster or tweak a working suggestion into a hypothesis which was prevented from coming into being for lack of sufficient quantitative data (e.g in a corpus). Again, any conceived combination would be highly dependent on the specificity of the object of study we are interested in. But from this perspective, methods are not seen so much as isolated and competing with one another, but rather as constituting different angles used to sharpen a multidimensional image of the feature(s) in question; although, to be sure, hierarchies based on priority can and must be drawn accordingly.
So, in conclusion, should we exorcise intuitions and elicited intuition-based methods? Admittedly, they cannot remain fully satisfactory instruments in the long run, if they are deployed on their own; but their strength may lie in their possibility to be associated with more objective and distanced types of observation; to be used as valuable prompts and pilots for larger, more ambitious collective research projects in sociolinguistics.
Krug, Manfred, Julia Schlüter & Annette Rosenbach. 2013. Introduction: Investigating language variation and change. In Krug, Manfred & Julia Schlüter (eds.), Research methods in language variation and change. New York: Cambridge University Press, 1-13.
Labov, William. 1996. When intuitions fail. In McNair, L. et al. (eds.), Papers from the Parasession on Theory and Data in Linguistics 32, 77–106.
Mertz, Elizabeth & Jonathan Yovel. 2009. Metalinguistic awareness. In Sandra, Dominiek, Jan-Ola Östman & Jef Verschueren (eds.) Cognition and Pragmatics. Amsterdam & Philadelphia: John Benjamins, 250–271.
Niedzielski, Nancy & Dennis Preston. 2000. Folk linguistics. Berlin and New York: Mouton de Gruyter.
Preston, Dennis. 1996. Whaddayaknow? The modes of folk-linguistic awareness. Language Awareness 5, 40–74.
Silverstein, Michael. 1981. The limits of awareness. Sociolinguistic Working Paper 84. Austin, Texas: Southwest Educational Development Laboratory.
About the author
Cameron Morin is a postgraduate student in sociolinguistics at the Ecole Normale Supérieure Paris-Saclay.
Edited by: Abhimanyu Sharma
Views represented in the article (and copyrights) belong to the author. The purpose of posting is to encourage discussion and not to endorse a certain point of view.