Where do sentences come from?

The other day, intent on demonstrating something obvious (but not trivial) about the language of Simply Describing, I went to my graduate seminar in psycholinguistics armed with a plastic ring (orange), one ball (small and black) and two plastic cups (one red and one green), all in a shopping bag. I told the some thirty students that I would ask them to close their eyes when I gave the number of each of five little demonstrations, then to open them when I said ‘open’ and to close them again when I said ‘ close ’ - and then they were to simply describe, in a single sentence that a hypothetical six-year-old boy ‘just outside the door’ would understand, what they observed during the eyes-open period. The only props were a bare table (desk) and myself, whom I instructed them to refer to as ‘the man’. The following sequence of events transpired:

1. ‘Close your eyes for #1.’ I place the orange ring in the middle of the table. I say ‘open’ and then in a few seconds ‘close’, and I ask them to ‘simply describe what you saw in a sentence ’.

2. ‘ Close for #2.’ I remove the plastic ring, take the black ball, and I stand holding the ball in front of me about shoulder high. I say ‘open’, then ‘close’ and ‘ simply describe what you saw ’.

3. ‘ Close for #3.’ I place the black ball in the middle of the table. And then again, ‘open’ (pause) ‘close’ and ‘simply describe’.

4. ‘ Close for #4.’ I remove the ball, take the red cup, and I stand holding the red plastic cup in my hand. ‘ Open ’ (pause) ‘ close ’ and ‘ describe ’.

5. ‘Close for #5.’ I place the red cup back in the shopping bag, take the green one, and I place the green plastic cup in the middle of the table. Then finally, ‘open’ (pause) ‘close’ and ‘simply describe’.

Note that each of the three critical demonstrations, #1, #3 and #5, is identical except for the particular object which is in the middle of the otherwise bare table, an orange ring, a black ball, or a green cup. Yet the types of sentences they produced varied markedly.

The sentences produced by # 1 were typically either

(1) An orange ring is on the table or

(2) There is an orange ring on the table

with at least one and often more adjectives included. Sentences with definite articles.

(3) The ring on the table is orange

or sentences making explicit the adjectival transformation

(4) A ring is on the table and it is orange

almost never occurred. On the other hand, after seeing #2 (usually described as the man is holding a small black ball), demonstration #3 did regularly yield sentences with the definite article along with adjectival pronominalization

(5) The black ball is on the table

but (except for a few philosophers in the class) types (1) and (2) rarely occurred and

(6) The ball on the table is black

never occurred. Yet demonstration #5, following the man is holding a red cup, did typically produce

(7) The cup on the table is green

but never sentences analogous to (5), which in this case would have been

(8) The green cup is on the table.

It is obvious that non-linguistic, perceptual antecedents can create cognitive presuppositions in the same way that previously heard or uttered sentences do and that these presuppositions influence the form that descriptive sentences take. If a speaker has already seen a particular BLACK BALL, ¹ and assumes that his listener is familiar with it also, then it is absurd for him to say the ball on the table is black, since it is its new location which is now informative, not its color or size. Perhaps not so obvious is the fact that such demonstrations call into question the semantic equivalence of even such common transformations as N is A into AN. In demonstration # 1 above, not a single sentence included an N is A assertion (ring is orange), yet this was the first exposure to this or any other object in the situation. Furthermore, it is apparent that the green cup is on the table is not semantically equivalent to the cup on the table is green; demonstration #3 produces the former type but not the latter, whereas #5 produced the latter type but not the former. The answer yes to the question is the cup on the table green? refers to the location of the object (its color being presupposed).

Where do sentences come from? In a generative grammar (linguistic competence model) the symbol S represents the set of all possible grammatical sentences in language L, and S can be re-written or expanded in all of these possible ways - infinitely given recursive devices.² But linguists ordinarily start with some actually uttered or potentially utterable sentence or non-sentence and proceed to analyze it in a posteriori fashion. The term ‘generative’ can thus be misleading. In any case, it is clear that any theory of language behavior (psycholinguistic performance model) must inquire into the antecedents of S and relate these antecedents to the forms and contents of particular sentences. Neither the syntactic bone nor the lexical flesh of sentences created by real speakers is independent of the non-linguistic contexts in which they occur. The body of this paper will be concerned with demonstrating, in some detail, how the form as well as the content of sentences can be influenced by manipulating the perceptual context in which they are produced.

Philosophers and linguists have given precious little attention to pre-linguistic cognitive behavior, in any of its three senses - prior to language in the evolution of human species, prior to language in the development of the individual or prior to language in the generation of particular sentences* by ordinary speakers. A philosopher-linguist may claim that there can be no thinking without language,³ but it is difficult to imagine how pre-linguistic humans types survived without ‘ thinking to say nothing of dogs and untrained deaf-mutes. Lenneberg, in his Biological Foundations of Language (1968), devotes exactly three pages (276-8) to pre-linguistic development, and this exclusively to articulation; philosophers of language typically talk about ‘objects’ (e.g., the traditional APPLE) as physical givens when they are really talking about perceptual signs which themselves have to be learned. And only quite recently have philosophers of ordinary language - and only very recently linguists - begun to take an interest in the cognitive pre-conditions for uttering sentences of certain types. Psychologists like Piaget, on the other hand, have perhaps paid too much attention to cognitive development and too little to language development (cf. Mehler and Bever, 1967, vs. Piaget, 1968, in Science).

My arguments in this paper will be as follows: (1) That pre-linguistic perceptuo-motor behavior displays many (even if not all) of the characteristics of linguistic behavior, including semantics (meanings of perceptual signs like APPLE or BALL), syntactics (appropriate interpretations of DADDY HITS BALL vs. BALL HITS DADDY) and pragmatics (appropriate behaviors with respect to gestured TAKE BALL vs. GIVE BALL). (2) That the ability to paraphrase perceptuo-motor events in language (Simply Describing), and vice versa (Simply Acting Out), implies both (a) that perceptual signs and events must have meaning and structure (one cannot paraphrase a meaningless, structureless string like colorless furiously ideas sleep green) and (b) that perceptual and linguistic signs and sequences must, at some level, share a common representational (semantic) system and a common set of organizational (syntactic) rules.⁴ (3) That that which is shared by both sign systems is not linguistic but cognitive in nature, and therefore that the constructs and rules of generative grammars are, in principle, incapable of dealing with it.⁵ And, finally, (4) that this non-linguistic cognitive system is ‘ where sentences come from ’ in sentence creating by speakers and * where sentences (finally) go to ’ in sentence understanding by listeners.

¹ Throughout this paper I will use all small capitals when referring to perceptual signs and their interactions in perceived events.

² Given the proliferation of subordinate embedded Ss characteristic of contemporary linguistic analysis, and the mutual constraints placed by one S upon others, I wonder how this symbol actually could have a constant reference.

³ As Zeno Vendler did at a conference on Language and Thought at Tucson in 1968 (see J. Cowan (ed.), Studies of Language and Thought, 1970).

⁴ This is not, of course, a claim that perceptual and linguistic semantics and syntactics are co-extensive with each other, merely that the one must at least overlap the other: presumably there are some things which one can ‘ think ’ with language which he cannot ‘ think ’ without it.

⁵ I shall also argue that this cognitive system is to be identified with the ‘ semantic com¬ ponent ’ of grammars and the latter in turn with the ‘ deep structure ’ level of syntax - not an outrageous argument any longer, apparently.