As Chomsky has shown (1965:22), the set of sentences S₁...S_n of a language L are generated by the syntactic component of the grammar of L. After the selection of lexical items to fit the syntax, the sentences are assigned a semantic representation by the semantic component,¹ and phonetic interpretation by the phonological component of the grammar of L, providing a systematic phonetic representation of the sentences of L for hearers to listen to. In this paper I shall show that all of the above processes apply to grammatical sentences such as (1), but not all of them apply to ungrammatical sentences such as (2).

Sentence (2) is clearly ungrammatical and as such could not be uttered by an idealized speaker-hearer. The question now arises of how the production of sentences like (2) is inhibited, and how this inhibition is shown formally in a grammar.

Recent experiments² have shown that sentences such as (2) are prevented, by a blocking transformation, from entering the phonological component of the grammar. That is, after passing through the semantic component and being found ungrammatical because of the well-known PICTURE+POUR constraint (Wassermahlerstromer 1892:50), these sentences undergo a further transformation which blocks assignment of a phonetic interpretation to the structure of ungrammatical sentences.

This is accomplished by the prefixing of an asterisk to the formative dominated by the leftmost node of sentences like (2) (see Figure 1). A universal convention, already needed in other parts of the grammar (de Asterisco 1588.xvii), states that no sentence whose leftmost formative is immediately preceded by an asterisk can be pronounced; it must either be suppressed or enter the graphological component and appear in an article about linguistics. The existence of this component of grammars has been suspected for some time (see, e.g. Kateb 1343:0), but until the discovery of the blocking transformation, empirical evidence for its existence was lacking, and it was not clear how sentences such as (2) could be generated at all.³

W	X	POUR	Y	PICTURE	Z
1	2	3	4	5	6	→  *1  2  3  4  5  6

¹The speaker is thus able to mean what he says.

²In a double-blind experiment, 48 idealized speaker-hearers were shown a slide of a man labeled JOHN pouring water from a pitcher, followed by a slide of the same man pouring water from a picture. 47 subjects either produced sentence (1) or had no opinion. The one aberrant subject produced the utterance ['komo se 'dise pin'tura ɛn iŋ'gles].

³This work was supported in part by Grant No. 913-864-4100 of the Onanism Foundation. I am grateful to my colleagues N. J. Marr of the Moscow Institute for the Wrong and P. S. Pallas of the Katzenhaus Staatgymnasium for their suggestions and for the experimental data described in fn. 2. They do not, of course, necessarily agree with my conclusions, but all errors are theirs.

Chomsky, Alfred E. 1965. Aspects of the What? Granuloma 913.22-22

de Asterisco, Raul. 1588. Gramática asteriscana, o como Drake no ha jugado justamente. Carne Orgulloso 42.xv-xviii

Kateb, Nevisandeh. 1343. Newishtan-i-dastur. Gosht-i-nau 2.24-0.

Wassermahlerstromer, Graf Kitschi. 1892. Wie giesst man einem Bild aus? Wildes Fleisch 12.49-2234.

	The Anguish Lenglage II—Tom Ernst
	On Some Acoustic Correlates of Isoglossy—Robert L. Rankin
	Lingua Pranca Contents