The English (present) perfect (e.g. I have gone) has been extensively studied in the theoretical literature on tense and aspect. The pluperfect (I had gone) has by contrast received relatively little attention, and the relationship between the present perfect and the pluperfect has been virtually ignored. Descriptive grammars of English also tend to say little about the latter two issues, beyond distinguishing cases such as (1)–(6), where the pluperfect seems to be the ‘past of the perfect’, from instances like (7)–(11), where the pluperfect is the ‘past of the past’ (cf. Thomson & Martinet, 1969: 112–13; Berland-Delépine, 1971: 95–6; Guitard, 1966: 206–7; Palmer, 1974: 54):

Lexical Phonologists have made a number of claims that are directly relevant to the study of sound change in progress, two of which I wish to examine here. First, phonetically gradient patterns of variation are alleged to be controlled by rules which operate outside the lexicon. Second, phonological rules applying within the lexicon may only refer to feature values that are already marked in underlying representations. This paper sets out to test these claims against empirical data of the sort that have been reported in the sociolinguistic literature. While the first claim appears to be in tune with some informal analyses already offered by sociolinguists, the second is contradicted by at least some of the evidence.

This paper will be concerned with the analysis of sentences like those in (1), which I think are all uncontroversial examples of the construction that is generally called ‘gapping’.

It is an unquestioned assumption of all recent discussions of semantic presupposition that a presuppositional semantics admitting of truth-value gaps depends upon and therefore implies a semantic ambiguity of natural language negation (SAN). Such theories have been criticized and defended on that basis. This essay invites the reader to contemplate the idea that this picture of things is fundamentally mistaken.

I have argued elsewhere (Hengeveld, 1987b) that for a proper treatment of modality the clause model used in Functional Grammar (Dik, 1978, 1980) should be adapted in such a way that a number of different layers can be distinguished. My main argument there was that predications, used in Functional Grammar to represent linguistic expressions, have two different functions: a DESCRIPTIVE function and a CONTENT function. A predication not only gives a description of the external situation the speaker refers to within his speech act, it also represents the prepositional content or message unit processed within that speech act. Subjective and evidential modalities, which express a propositional attitude, should take a predication in its contentrepresenting function in their scope, whereas objective modalities, i.e. those modalities that are concerned with the actuality status of a State of Affairs (SoA), should take a predication in its SoA-designating function in their scope. A clause model should therefore be able to distinguish between these two functions of predications.

With a few honourable exceptions, phonologists have, until recently, generally assumed the phonological component to be organized along completely different lines from the other components of the grammar. Phonological phenomena were mainly viewed as accidental, language specific, and unprincipled. Whereas some theoretical parallels between syntax and phonology have been drawn (cf. the role of the cycle in both domains and the extensive literature on ordering), there have been few attempts to see if principles of Universal Grammar could be found in phonology as well as in syntax and semantics. Increasingly, however, phonology is now being regarded as a system of principles along with parameters defining the class of human phonological Systems. In such a framework there are no rules of the sort: A → B/C→D. Phonological phenomena result from principles and parameters governing phonological representations and structures present in a particular language. Along these lines, recent work in phonology has suggested that Phonological Form (PF), like the other components of the grammar, is subject to certain fundamental principles. For example, it was proposed by Andersen and Jones in the early 1970s (and pursued by Ewen, Durand and others) that the relations of dependency that determine how syntactic constituents are organized, also determine how segments are grouped together in a given structure. For their part Lowenstamm & Kaye (1982) proposed that a theory of government could account for certain phonological processes such as vowel shortening in closed syllables. Stephen Anderson (1982) and Levin (1985) have proposed that X-bar principles govern the representation of syllables. Specifically, they have proposed that the Rhyme and the syllable as a whole are projections of the syllabic head, the Nucleus.

There is a widespread current assumption that the semantic relations underlying linguistic constructions reflect the way human beings ineluctably, and hence universally, perceive events and situations. This assumption will be questioned in this paper. To make the point, a linguistic construction purportedly expressing the instrumental case will be examined. It will be convenient here to juxtapose my treatment of this construction with the one proposed by case grammarians, and by Fillmore in particular. However, the discussion has implications for semantic relations in any linguistic theory, whether they appear there as semantic roles, thematic relations, theta roles, or whatever. These implications will be discussed in the final section.