¹ For example, A. A. Hill gives an elaborate ‘morphemic’ analysis of English personal pronouns in his Introduction to linguistic structures 145–52 (New York, 1958). He says of ‘pronominals’ (551) : ‘No form class has been more confusingly defined, since the reliance is almost always on semantic characteristics. Thus, no less a scholar than George O. Curme was led to define others as a pronoun since it is a substitute. Yet since it shows all the formal characteristics of a noun, it must belong to that form class rather than to pronouns’. By ‘formal characteristics’ Hill presumably means characteristics based on (phonemic?) form, not characteristics which are abstract or mathematical as opposed to concrete, empirical, or contingent. Hill also describes a class of ‘pronominals’ in initial // and /hw/ (370–89) and makes occasional references to pronouns as nominal elements in various constructions.

² We shall not attempt to study or analyze the so-called ‘emphatic’ constructions in -self, as in John did it himself.

³ We use the asterisk throughout to designate ungrammatical expressions. Naturally our judgments of grammaticalness are based directly and solely upon our own maximally formal speech style and usage, and the rules we formulate will, therefore, characterize sentences in our own dialect only. It is not likely that corresponding rules which could be formulated for other forms of English will diverge sharply from those which we use to construct sentences, but there will be readers who judge differently certain examples we quote.

It is also important to recognize in this connection that, except for utterly impeccable short sentences or for unintelligible gibberish, absolute judgments on grammatical acceptability cannot be given with assurance unless some independent knowledge has already been acquired about other formal features of English sentences and grammar relevant to the doubtful examples in question. Thus, we would reject the ‘queer sentence’ *John astounded the dark green with some conviction, since we know from independent analyses that the distinction between ‘animate’ and ‘inanimate’ nouns is relevant in English sentence structure. In any case, the point of such studies as this is, of course, not to legislate against the use of certain expressions nor to disparage certain styles or dialects of English, but rather to account for an indisputed widespread agreement among speakers of one dialect of English that certain types of expression are structurally deviant.

⁴ Some classical grammarians also interpret the reflexive as the result of a change imposed on a more basic sentence, and thus in a sense transformationally. Cf. e.g. H. Poutsma, A grammar of late Modern English 2.836 (1916): '… to denote that the person(s) or thing(s) referred to in any enlargement of the predicate are the same as that (those) indicated by the predicate.' Or even more clearly O. Jespersen, Essentials of English grammar 111 (1937) : 'When subject and object are identical, we use for the latter the so-called reflexive pronoun.'

⁵ Although the basic idea itself is ancient, the formal particularities in this analysis of certain sentences as transforms of underlying source sentences are taken directly from well-known works of N. A. Chomsky. The details of English syntactic structure lying behind our analyses, as well as a more general discussion of the theoretical basis for this view of grammar and of language, may be found in R. B. Lees, The grammar of English nominalizations (IJAL 26:3, Pt.2, 1960).

⁶ Ibid., Rule (T68*).

⁷ And the redundant preposition for, when it occurs before the to or for of an Infinitival Nominal, is also automatically deleted; ibid., Rule (T69*).

⁸ Ibid., Rules (GT 19) and (T58).

⁹ This rule would be an addition to Rule (GT13) of Lees, op.cit., not discussed there. Also, in connection with these indirect-object examples, we shall consider sentences of the form I bought me a hat, confined to the first person, to be nonstandard colloquial variants of sentences of the form I bought myself a hat.

¹⁰ This is an addition to Rule (T2) of Lees, op.cit. Also cf. Jespersen, A modern English grammar 3. §15.13 .

¹¹ Lees, op.cit., Rule (GT10) and the preceding discussion. By ‘action type’ Gerundive Nominal we mean that type which must have an indefinite human subject and no auxiliary accompanying the verb (i.e. we have no *John's protecting himself is hard, nor *Having protected oneself is hard.)

¹² We ignore certain special expressions such as Just look at you!, which might possibly go back to a hortatory Let us just look at you or the like.

¹³ Ibid., Rules (T65*) and (T67*). Following a suggestion by L. R. Gleitman, we assume that the rules are optional for third person and obligatory for first and second.

¹⁴ Also, incidentally, they are precisely those expressions of English which N. Goodman attempted to use in formulating a theory of synonymy as a part of the theory of reference : ‘On likeness of meaning’, Anal. 10 (1949), revised in L. Linsky's Semantics and the philosophy of language 67–74 (1952).

¹⁵ In fact C.J. Fillmore has given a reasonable analysis of the double-object construction in Indirect object constructions in English and the ordering of transformations, Ohio State Research Foundation Project on Syntactic Analysis, Report #1, Prof. 1303, 13 Feb. 1962. Fillmore assumes an underlying construction of the form John Tns give to Bill a bite (exactly parallel to John Tns shoot at Bill a gun), which is later subject to the usual ‘separation’, rule, yielding John Tns give a bite to Bill, i.e. the sentence John gives a bite to Bill. There is, however, an intervening optional rule which deletes the preposition to, yielding John Tns give Bill a bite, the sentence John gives Bill a bite. He is able to account also for the existence of two different passives for such indirect object sentences, and he gives an analysis of the contrasting indirect object construction in for: John buys a car for Bill: John buys Bill a car, for which there is no passive.

¹⁶ Cf. the very similar analysis of the reciprocal in Danish given by 0. Jespersen, Sprogets logik 79–80 (1931) : ‘Nâr karen ligner maren, ligner maren også karen. I disse tilfelde er der intet i vejen for at gøre subjekt og objekt til to forbundne Subjekten, og objektets plads udfylder vi da med ordet hinanden … As v Bo = Bs v Ao = (A + B)s v hinanden-o.‘ This interpretation is again essentially transformational, as are many of Jespersen's grammatical analyses.