Methodology in Cognitive Linguistics
What empirical research involves
Suppose for a moment that you are a professor of linguistics teaching an introductory course to a large audience of a few hundreds of students. You hardly know them personally, but when you grade their exams, there’s a funny thing that catches your attention: over the years, you notice that the students’ first names come in clusters. In one year, your attention is caught by a frequent group of Heleen’s or Ineke’s, in another, by a conspicuous cluster of Ron’s or Len’s or George’s. As a professor of linguistics, you’re an inquisitive type of person, so you wonder whether this is a coincidence. What could be the case? One possibility would be that the clustering of proper names in your class is a faithful reflection of what is happening in the population at large: there are fashions in giving first names that change over the years, and the sample of the population that attends your class reproduces the overall tendencies. In that case, there is something specific about the age of your students: they were born in a period in which a particular name was trendy. But there is at least one other possibility to consider: it could be that the frequency of specific names in your class differs from what you could expect given the distribution of names in the population as a whole. In that case, there would be something specific about your class: for one reason or another, it attracts people with a certain name.
How could you decide between the two options? You would want to make a number of comparisons: on the one hand, you would compare the distribution of names in your class with the distribution of names in the entire population of the same age group, and on the other, you would compare the distribution of names in one year or one period with the distribution of names in another year or period. In both comparisons, you would not just want to know whether the distributions differ, but specifically also whether the differences you observe are a coincidence or not. That is, of course, where statistics will help you: a statistical test will allow you to see whether a difference you detect is due to chance or not.
If it is due to chance, you will leave the matter for what it is, but if it is not, your curiosity will get the better of you, and you will start wondering about an explanation. For instance, if there are fashions that change through time (as there would be), what is it that determines them? What is it that makes Heleen popular in a particular period? A hypothesis that you are likely to come up with is that naming fashions of this kind are determined by the degree of exposure certain names get in the media and in popular culture: popular sports people or popular characters from television fiction– people “whose name gets around”– may well provide a model for naming new-borns. Again, you would want to put the hypothesis to the test: you may have to put in a considerable effort to find the right kind of data, but you would want to find out whether one or another Heleen was popular some twenty years before you noticed all those Heleen’s in your class. Or rather, you would not want to do this for just one of the names, but you would want to reach a general conclusion: if a certain name is popular in the media in a certain period, does that lead to an increase in the number of babies receiving that name in the following years? And again, you would need some statistics to ascertain whether the influence you think you see might not just be a coincidence.
We could go on like this, surely, investigating explanatory questions and research techniques for our hypothetical Christian Name Conundrum. The point is this: the procedure illustrated in the previous pages is typical of what we call empirical research, and it involves a way of thinking that we all understand– and that we all recognize as being superior to an attitude in which you would simply say: “I notice these clusters of names in my classes. I conclude that they are due to fashion”. Before you can really conclude that, you have to take a number of steps, steps of the type that we explored above.
Then, is all about the position of empirical research in Cognitive Linguistics. It will argue, first, that linguistics in general would benefit from more research of this type; second, that Cognitive Linguistics has an excellent starting-point to contribute to such an increase in the use of empirical studies; and third, that in spite of that advantageous basis, Cognitive Linguistics has not yet fully realized the breakthrough towards the massive use of empirical methods. What we will not do, by the way, is discuss the practicalities of doing empirical research in linguistics. Although, in terms of boosting empirical studies, it would be extremely useful to do so, it is simply beyond the scope of the present text to offer a practically oriented primer in experimental or corpus-based studies in linguistics. It is hoped that the references to empirical studies scattered throughout the text will help the reader to get started. Further, see Gonzalez, Mittelberg, Coulson, and Spivey (forthcoming) for a book-length introduction to different empirical methods for Cognitive Linguistics.
As a crucial backdrop for the following discussion, let us now summarize the main features of empirical research that we can derive from our simple example.
1. Empirical research is data-driven. You cannot easily draw conclusions from single cases and isolated observations, and the more data you can collect to study a particular phenomenon, the better your conclusions will get. The observations could come from many sources, by the way: you could collect them as they exist (as in our Christian Name Conundrum), but you could also elicit them by doing experimental research, or by doing survey research– we will come back to that.
2. Empirical research involves quantitative methods. In order to get a good grip on the broad observational basis that you will start from, you need techniques to come to terms with the amount of material involved. Specifically, you need statistical tests to determine whether your observations might be due to chance or not.
3. Empirical research is all about asking the right questions– the formulation of hypotheses, in other words. No perception could be more misguided than to think that once you have your database of elicited or non-elicited observations, the conclusions will arise automatically and purely inductively from the data. On the contrary, the only conclusions you will be able to draw are the ones that relate to hypotheses you have formulated and tested– so that will be your first task. In our example, you have to think creatively and inventively about the possible background of the clusters of names that catch your attention, and it is only when you have arrived at such a hypothesis that you can start thinking about testing it. Another way of saying this is that empirical research necessarily combines inductive and deductive reasoning: on the one hand, you work in a bottom-up way from data to hypotheses, but on the other hand, those hypotheses will also be derived top-down from the theoretical perspective you adopt in thinking about your data.
4. Empirical research requires the operationalization of hypotheses. It is not sufficient to think up a plausible and intriguing hypothesis: you also have to formulate it in such a way that it can be put to the test. That is what is meant by “operationalization”: turning your hypothesis into concrete data. For instance, it is fine to conjecture that media exposure could have an influence on proper name fashions, but how do you measure the popularity of a name in the media? It’s not just a question of finding the relevant data that allow you to measure the popularity of a name (like maybe broadcasting schedules), but also of settling on a method of measurement. Would you just measure the number of times a name is broadcast, or would you take into account the ratings of the programs in which it appears? And how would you deal with the existence of different television channels, or the simultaneous effect of television, radio, and the written press? In most empirical research in linguistics, it is questions of operationalization that require all the ingenuity of the researcher– and most of his or her time, because getting the relevant data and measurements is not an automatic process.
5. Empirical research involves an empirical cycle in which several rounds of data gathering, testing of hypotheses, and interpretation of the results follow each other. In the (fictitious) example, we started with an anecdotal observation, tried to interpret it, turned the possible interpretations into alternative hypotheses, tested the hypotheses, and so arrived at a new, more secure and better-established observation. But this observation was only the starting-point for a new attempt at interpretation, which led to a new hypothesis about media influence, which led to new research efforts. Just like it is misguided to think that empirical, data-driven research automatically gives you all the answers, it is misguided to think that it immediately gives you the final answer.
