Consideration of the problem of alternative agreement forms (§ 1) leads us to postulate a hierarchy of agreement positions (§ 2) which allows us to make predictions as to the possibility and relative frequency of semantic as compared to syntactic agreement. The hierarchy is justified by data from a variety of languages which permit alternative agreements (§ 3). Confirming evidence comes from a study of the instances where agreement in case and person is required in different languages (§ 4). The status of the hierarchy and the type of prediction it makes are discussed. It is suggested that the hierarchy determines the main divisions of a measure called SYNTACTIC DISTANCE and that other factors (word order, distance between controller and agreeing element, depth of stacking) determine relative degrees of distance within these main divisions (§ 5). The basic claim of the paper is that as syntactic distance increases so does the likelihood of semantic agreement.

In a number of works up to 1974 Jerrold Sadock advanced and defended the claim that there are a limited but distinct number of sentence structures in English which exhibit true illocutionary ambiguity. Structure (1) belongs to this limited number.

Most linguists agree that there is a mismatch between the notions of grammaticality and acceptability. A widely held view is that there are several types of sentence which are technically well-formed, and consequently grammatical, but which are unacceptable to the ‘typical’ speaker. A more recent, and less widely held view is that the reverse situation is also possible: there exists a class of sentences which are ungrammatical, yet acceptable. In other words, it is claimed by some that the notions of grammaticality and acceptability cross-classify, since we get sentences which are:

1.Consider a string s = a1 a2 …an over a finite vocabulary Va, such that s is structured by an asymmetric binary relation R: i.e. there are ordered pairs of elements in the string, ai, and aj, where a, R aj.1 S is a simple structured string if (a) there is a unique element ak such that there is no a1 for which it is true that a1 R ak, and if (b) for all am am, ≠ akit is true that there is a unique a1 such that a1 R am. In graph-theoretic terms these conditions define a tree where a1, a2 …a are paired with nodes in the graph and each a, R a is represented by an arc, and where the tree has a unique root or centre associated with a. Figure i is an instance of such a tree, where k = 3. In Figure a, condition (a) is infringed, in that an arc converges on each node; i.e. for all a, there is an aj such that aj R ai. Figure 3, on the other hand, violates condition (b), in that two arcs converge on a2 i.e. there is not a unique ai for whch it is true that a1 R a2.

The attention by ‘modern’ linguists to the problems of Old English diphthongal spellings and their phonetic realizations may perhaps be dated (without doing anyone an injury) from the study by Robert P. Stockwell and C. Barritt (1951). Since that time much ink has been spilled (or typewriter ribbons worn out) in attempts to specify in some precise and principled way the phonological structure of the entities spelled in Old English ea, eo, io, Ia and ie. I will spare the reader a review of these articles and commentaries since such a review would prove both tiresome and uninformative.

My main purpose here is to examine the basic colour vocabulary of modern standard French in the light of recent research on colour vocabularies and in particular to account for the fact that two highly salient terms, brun and marron, are used to designate a single colour category which I shall call BROWN. (The notion of saliency is defined in § 1.2 below.) Before the publication in 1969 of Berlin and Kay's Basic color terms with its universalist hypothesis, the fact that the English term brown has no one equivalent term in French would generally have been regarded as evidence for the structuralist view that languages are anisomorphic in their semantic structures (Lyons, 1968). Linguists before 1969 turned to colour vocabularies to demonstrate not only that languages are anisomorphic in their semantic structures but that the colour continuum is cut up in a completely arbitrary way in different languages (Hjelmslev, 1943; Gleason, 1955). Recent research, however, suggests that some modification of this view may be necessary.

Many linguists describe the underlying structure of sentences in terms of cases (e.g. Fillmore, 1968) or relations between the verb and nouns (Chafe, 1970), the, usually tacit, assumption being that this reflects the way we view the world. The underlying linguistic structure would accordingly be a formalization of cognitive structure. Elsewhere (Schlesinger, 1977: Ch. 5 ), I have argued that a distinction has to be made between cognitive structures and semantic deep structures. Cognitive structures correspond to the way we perceive the world.

Since Postal (1971: 27–31) first used the term Tough Movement, a great deal of material on this phenomenon has appeared in print. In all of this research, there has been a widespread (often tacit) assumption that the characteristic - in fact, exclusive - triggers for Tough Movement are adjectives (e.g. tough, hard, interesting) and nouns (e.g. bitch, bear, snap). For example, Lasnik and Fiengo (1974: 568) list 21 common Tough predicates. Although their list is not meant to be exhaustive, it is a characteristic one, especially in that adjectives outnumber noun triggers two to one and half the nouns are limited to vernacular usage (the most frequent use of noun triggers probably involves obscene usage).

Harris (1978) argues against the ‘descriptive’ interpretation of performative utterances on the ground that on a particular occasion of its use I apologize may not constitute a ‘genuine instance of apologizing’, although the use of that utterance on that occasion may be ‘correctly described’ by using the verb APOLOGIZE non-performatively, as in (these are Harris's examples): (1) S apologized to the lamp-post, and even the police sergeant laughed. (2) S apologized on T's behalf, although T himself flatly refused to.

In an earlier paper (Huddleston, 1977) I analysed the would of (I) If he had stayed in the army, he would have become a colonel as the unreal mood form of an epistemic modal verb WILL that figures in the underlying syntactic representation and falls within the semantic scope of a past element realized by HAVE. Palmer (1978) criticizes this analysis, claiming that WILL is ‘obviously’ semantically empty, a mere ‘dummy’ inserted to carry the mark of unreality – that would have become is ‘the Unreal and Past form of BECOME’.