ETHNOLINGUISTIC NOTE 37

Links to pages: 557, 561, 562, 563

Comments Welcome

George W. Grace

University of Hawaii

1 October 1989

ETHNOLINGUISTIC NOTES

Series 3, Number 37

RECOGNITION STRATEGY AND ANALYSIS STRATEGY IN LANGUAGE USE

In Grace 1989 I described several kinds of evidence which suggested to me that our knowledge of our language involves "some kind of direct linkage in our 'minds' between sentence-sized linguistic expressions and 'situations'" (1989: 516). I felt that this evidence suggested, "something like a dictionary which contains an inventory of linguistic expressions and an inventory of 'situations' and which matches the members of the first inventory and the members of the second inventory in some approximation to a one-one mapping" (Grace 1989: 518).

I went on to hypothesize (1) that "dictionary-type" (or "holistic") knowledge of linguistic expressions was both phylogenetically and ontogenetically prior to "grammar-type" ("analytic") knowledge, and (2) that the extreme prominence given to "grammar-type" knowledge by what I called the "school language" of modern Western culture seriously misrepresented the nature of the linguistic knowledge of most of the people of the world, even of members of Western societies. (I think this is especially true of non-"mainstream" members).

My purpose here is to propose that the use of "dictionary-type" knowledge involves a particular strategy which I will call recognition strategy. This strategy is used for understanding something which has been said by someone else and is also used by a speaker-to-be for selecting a linguistic expression to serve as the vehicle of his/her speech act (i.e., for selecting a linguistic expression which matches the situation to be signaled).

As used for understanding, the strategy consists in nothing more than attempting to "recognize" the linguistic expression which one is seeking to understand, where "recognizing" it means identifying it with a linguistic expression for which there is a dictionary entry.

As used to select a linguistic expression for uttering, it consists in nothing more than attempting to recognize the situation which the speech act is to signal--i.e., to identify an appropriate situation from among those for which there are dictionary entries.(1)

I don't know how we go about trying to recognize something, but I know that we do do so. Making such an attempt seems to involve both (1) trying to get a better, or at least another, look (substitute equivalent words for other senses which may be involved) at whatever it is, and (2) trying to figure out where one might have seen the same person or thing before--i.e., trying to provide oneself with the right expectations.

The recognition strategy is complicated by the fact that some of the situations and some of the linguistic expressions in the dictionary are abstract or general in nature (I don't as yet understand what is involved well enough even to be sure how I want to talk about them). In Grace 1989: 529ff., I suggested that there might be "groupings" of similar situations signaled by similar linguistic expressions. Examples mentioned there were (1) the members of inflectional paradigms and (2) speech formulas whose lexical content is not fully specified. I now think that it is a mistake to assume that every different linguistic expression (no matter how small the difference) has a separate dictionary entry. Rather, it makes more sense to think of some expressions--such as schematic speech formulas--as conceived in somewhat more abstract terms.

One consideration which leads me to this conclusion is that it is clear that the entries on the situation side of the dictionary must represent abstractions since only abstractly-defined situations could ever occur more than once, and therefore only abstractly-defined situations would be suitable for matching with linguistic expressions. Once it is acknowledged that even the situations associated with the most invariant linguistic expressions are abstractly defined, it becomes relatively easy to accept the existence of (presumably even more) abstractly-defined situations corresponding to formulas such as "The X-er, the Y-er", and the like--formulas which are, in their own turn, abstractly defined.

Of course, I don't understand very well how the recognition strategy actually functions in language use. I'm sure that it involves memory searches and pattern recognition, both of which are widely used for non-linguistic purposes (and neither of which is well understood). However, to make the discussion somewhat more nearly concrete, I will offer a speculative account of how a recognition strategy might proceed in attempting to find an interpretation for a linguistic expression. I suppose that it would go something like this:

1. Try to recognize the exact expression--that is, try to find the whole expression in one's memory. The application of the recognition strategy is simply a series of memory searches, of course, and presumably it ultimately works like any other memory searches (however that may be).

2. If the exact expression cannot be recalled, try to find some recognizable pattern in it. Just how this works is again unclear to me. I would guess that the kind of recognition involved is something like the kind we use when we recognize an animal or plant as being of a particular breed or species, or a painting or piece of music as representing the style of a particular school or artist, or a language as of a particular family or subgroup.(2)

3. Having recognized a pattern in the linguistic expression (i.e., having identified [some portion of] the expression with an [abstractly-defined] expression in the dictionary), try to recognize the remaining portion. Here again, I must be vague. Sometimes the abstractly-defined expression will constitute something like a frame in which fillers are inserted; sometimes it will constitute a lexically-specific sequence which is less than a whole sentence, etc. I don't know what all of the possibilities are.

4. Repeat the process until no unrecognized portion remains. Interpret the entire linguistic expression as signaling the situation which results from the superposition of the various abstractly-defined situations associated with the abstractly-defined linguistic expressions recognized.

I should emphasize that I am not suggesting that the foregoing represent separate steps which are taken in this, or any other, sequence. I think, rather, that several of the steps distinguished in my description are likely to be going on at the same time.

I am proposing the recognition strategy as an alternative to what I will call the analysis strategy. My "analysis strategy" is supposed to represent the strategy assumed by current linguistic theory to perform the same functions that I am here attributing to the recognition strategy. As I understand it, the analysis strategy is supposed to employ three kinds of information in interpreting a linguistic expression. These are: (1) the syntactic structure of the linguistic expression, (2) the "meanings" of the constituent lexical items as they are given in a dictionary which matches lexical items with "meanings" (and various other information), and (3) a "compositional semantics" (as it has been called) which is used to determine the contribution to meaning made by the syntactic structure. (It will be apparent that my understanding of this process is very incomplete).(3)

However, where there seems to be no danger of confusion, I will use this same designation (viz. analysis strategy) in a loose sense for any strategy that involves recourse to a partial analysis of the syntactic structure of a linguistic expression and/or the dictionary definitions of some of its constituent lexical items in figuring out an interpretation for it.

It is my opinion

(1) that the analysis strategy is almost never used alone (even by the most highly educated members of Western societies), that its main use is as an adjunct to the recognition strategy, that the situation in which it is most likely to be used alone would be in deciphering a text in a language (or jargon) not known to the decipherers, and

(2) that only a very small fraction of the language users of the world have more than the most rudimentary command of the strategy, that it is really something invented in connection with what I have called the "school language" of Europe, and even in Western countries not mastered at all well by more than a small fraction of the population (which fraction, however, does include pretty well all linguists).

FURTHER COMMENTS

1. I suspect that we may eventually come to understand the structure of a language as consisting of a number of grammatical constructions, each with its own functions (of the sort that would now be thought of as "compositional semantic" and "pragmatic" [and maybe "semantic", as well]). In that event I would expect the dictionary of linguistic expressions to include these grammatical constructions among the abstractly-defined linguistic expressions.

2. I want to suggest that the analysis strategy was necessitated by the invention of what I have called "formulated linguistic expressions" (cf. Grace 1987: 434, 1988a: 480). I see formulated linguistic expressions and the analysis strategy as necessary parts of the "school language" which I discussed in Grace 1989 (cf. 523ff.). School language, in turn, is an essential instrument of the modern world "monoculture" (cf. Grace 1988b: 487). The monoculture would be impossible without the the-meaning-is-in-the-text assumption. That assumption, since it is an inextricable part of school language, is as unquestioningly accepted as school language itself.

I do not want to suggest that no individuals outside this tradition ever discovered analysis techniques or that none were ever capable of any use whatever of them. What I want to suggest is that they played at most a very peripheral role before school language.

3. Recognition strategy is coming to be acknowledged in standard linguistic theory under the rubric of "pragmatics". However, it is assumed to play only a complementary role to analysis strategy (associated with "syntax" and "semantics").

4. Recognition strategy makes much use of the context of the speech act and of what is expectable. Analysis strategy as such makes none. That is one reason why children (or adults) who have no knowledge of analysis strategy find reading most materials difficult.

5. Recognition strategy is very inadequate for dealing with texts in which a familiar word or sequence of familiar words is given a special ad hoc definition for the purpose of that text. People equipped only with recognition strategy have great difficulty with such texts.

6. Recognition strategy is very inadequate for dealing with artificial phrases (e.g., "foot bird-house", foot-bird house", bird foot-house", etc.) of the sort invented by the Gleitmans for the experiments reported in Gleitman and Gleitman 1970. I propose that the fact that Group A in the Gleitmans' study, composed entirely of Ph.D.s and graduate students, did better than Group C, none of whom had gone past high school, is due to the availability of analysis strategy to the former.

NOTES

1. It is probably worthwhile once more to remind readers that the situation to be signaled is what I have called a "subjective situation"--ultimately a state of mind of the speaker. To what extent the speaker may have things outside him/herself on his/her mind--i.e., contributing to his/her state of mind--will vary from case to case. However, humans do have much interest in--and a very detailed understanding of their external environment--and often are interested in communicating about it.(Back up)

2. Of course, it is not necessary to employ a recognition strategy to make any of these identifications. Theoretically at least, any of them could also be made by means of an analysis strategy. And, of course, there have been frequent attempts to explain recognition in general as employing analysis, whether consciously or not.(Back up)

3. It is not clear, at least to me, just what form the results of an interpretation by this analysis strategy are supposed to take. Some discussions seem to suggest that the interpretation is simply a change from one "encoding" to another--the latter in some kind of code used by the "mind". It is not clear how this new coding would be any closer to something which could be acted upon than the original linguistic expression.

On the other hand, I do believe that the "situations", identified by my proposed recognition strategy, do have practical implications for behavior.(Back up)