Expressive or receptive language?

Tests of language development generally distinguish between ‘expressive’ or ‘productive’ language’ and ‘receptive language’ or ‘comprehension’. But if these tests use different criteria according to what is being tested, as many tests do, it is not clear that the same cognition is being tested in the two cases. The distinction may be an artifact of the testing.

Small children can often respond appropriately to instructions before they give equivalent instructions themselves. The apparent discontinuity with reception before expression in almost all areas of language in almost all children may be due to any number of intervening factors – including memory, familiarity, confidence in the use of language, access to the necessary words, and the demands of the context. Where there seems to be a developmental issue, the difference can seem even more stark between comprehension and expressive language. So there is a notion of Specific Expressive Language Delay or Disorder or SELD. This is sometimes used as a diagnostic category, often based on a difference between children’s scores on tests of their comprehension and their expressive language. But this is not sufficient to force the conclusion that expressive and receptive language represent separate areas of cognition. Part of the problem here is that it is difficult to test these things symmetrically. For instance, many of the most naturally plausible ways of testing comprehension are with respect to answers to questions beginning with who, what, where, when, why and how, or requiring Yes or No as an answer. But it is difficult to devise test protocols which require children to ask corresponding sorts of questions in similarly natural and plausible ways. Take the statement “She seems to know what she is doing”. What is clearly the object of doing. And she is clearly the subject of both know and do, as well as seem. The construction here is much discussed in various competitive theories of generative grammar. But it is not obvious how to test either whether this is understood or how to elicit such structure in productive speech or what cognitions are involved in either case. So tests of comprehension and expressive language may be testing different things. And the differences on the scores may be an artifact of the differences between the protocols. A similar difference may underly the common observation about more normally developing children to the effect that they understand more than they can say. This would seem to be just an unverifiable claim.

It is possible to test ‘reversible’ structures, such as “The boy is pushng the girl” or “The girl is being pushed by the boy” or “The boy is being pulled by the girl” or “The girl is not being pulled by the boy” with corresponding pictures. These just happen to be things which seem to be easily testable. But it is not clear exactly what is being tested. The pragmatics of “The girl is not being pulled by the boy” are decidedly odd. It is not hard to devise a scenario in which a negative passive is conversationally natural. But in varying all the properties, the words themselves and these particular constructions, negative and passive, the individual being tested may be jumping between different areas of cognition, syntax and pragmatics. What the test may be really showing is how well they can make that jump.

It is possible, as the framework here would suggest, that there are just different degrees of competence, yielding different results according to how these are tested.

It may well be that there are psychological aspects of production and comprehension which are specific to these aspects. But that is not the same as saying that production and comprehension are different things. In the framework of generative grammar, neither production nor comprehension is well-defined other than as one aspect of a single cognition.

One area where there may be a substantial difference between what is said and what is understood may be with respect to the lexicon. Children may confidently and accurately point to pictures of lions, tigers, rhinoceroses and other exotic animals from their pictures, long before they are using these words in their own speech. The passive vocabulary is likely to be several times larger than the active one. But as the lexicon expands, both of these things are very difficult to measure with any accuracy.

More generally, it may be that the notion of an ‘expressive language disorder’ is less well-defined than sometimes assumed.

It might seem possible in principle to devise a fully symmetrical test of expressive and receptive language. But in the absence of such a thing, the difference here may be artifactual. If a test is designed on the basis that speech and language are behaviours, rather than capacities, the test may be doing little more than testing this assumption. Tests of ‘expressive’ and ‘receptive’ language may may be just tapping different cognitions.

Apart from the fact that reversibility is very difficult to test in a way that is natural, specific, and accurate, it is also only one small part of everyday syntax. There are also relative clauses, as by “Find me the book we looked at yesteday”, conditionals, as by “If you do that again, I will be cross”, object control, as by “I want you to be good now”, and a lot more, all naturally starting to emerge between two and three. The focus on testing whatever is most easily testable misses key phenomena and risks understating real problems.