Although else is recognized as a cohesive device (e.g. Halliday & Hasan, 1976: 76), it has received relatively little attention in the literature. To help remedy the situation, this article investigates how native speakers use else in both written and spoken discourse. We first review the literature for a description of the meaning(s) of else, then identify its most common meaning as reflected in its frequency of use in American English native-speaker discourse. More specifically, we undertake a quantitative analysis of the frequency and distribution of different types of else constructions, as well as a qualitative analysis of how else functions as a referential and cohesive marker. The unique referential role of else is then highlighted by contrasting it with its closest synonym, namely other. It is shown that these two referential and cohesive markers complement each other, with else playing a less specified semantic role than other.

The article discusses the distribution and meaning of ‘locally free reflexives’ (‘untriggered reflexives’, ‘viewpoint reflexives’, ‘perspective logophors’) in English. After a thorough assessment of the contextual constraints on the use of such locally free self-forms, three analyses are discussed and compared: (a) the view that such self-forms are bound by a minimal subject of consciousness within their discourse (Zribi-Hertz, 1989); (b) the view that the relevant forms exhibit essentially the same distribution as pronouns and that the choice between the two is motivated by discourse considerations (Reinhart & Reuland, 1993); and (c) the view that locally free self-forms are intensifiers without pronominal heads (Baker, 1995). It is shown that the third analysis is by far the most adequate one, and that both the distribution and the meaning of such expressions can be explained on the basis of this analysis if it is combined with a suitable semantic analysis of intensifiers.

In this article the authors examine six different analyses of the meaning of unless that have been argued in the literature, and present a new analysis in which a careful distinction is made between the semantic meaning and the pragmatic interpretations of unless. Contrary to the common belief that unless cannot be used in irrealis conditionals, it is shown that unless can actually be used in two different senses there, one in which it alternates with except if, but not if . . . not (e.g. He wouldn't have done it {unless I had / except if I had / *if I hadn't} asked him to, and I didn't) and one in which it alternates with if . . . not but not with except if (e.g. You did do well, but you wouldn't have succeeded {unless I had / if I hadn't / *except if I had} helped you). It is shown, finally, how the various interpretations of unless can be reconciled with a common semantic meaning and how the different constraints on the use of unless, except if and if . . . not can be accounted for.

Actually and in fact have been generally regarded as interchangeable without leading to any significant differences in the meaning of the containing utterances: as a result, no serious attempt has been made to discover potential differences between the two. The present study focuses on the differences as well as the similarities between actually and in fact in their distribution and use in spoken and written American English. Based upon an analysis of tokens from the Switchboard Corpus and the Brown Corpus, I propose that `unexpectedness’ is the core meaning shared by actually and in fact, and that the difference between the two lies in the typical association of each with one or the other way of signalling ‘unexpectedness’. The study also shows that in real discourse contexts, actually and in fact develop a number of different uses that are more or less remote from this core meaning.

Previous work in syntax, semantics, pragmatics, and psycholinguistics has shown adjectives to take an intermediate position between nouns and verbs. This claim is made more specific in this article through a phonological (and to a lesser extent, morphological) test of the ‘equidistance hypothesis’. On seven criteria, English adjectives are found to exhibit a much greater affinity with nouns than with verbs. This pattern of results is brought about by the syntagmatic context in which the three word classes typically occur. The rhythmic context in which adjectives are embedded is similar to that of nouns but dissimilar to that of verbs. This bias gives rise to suprasegmental differences which in turn lead to segmental differences. The proposed account rests upon the assumption, independently verified, that nouns and verbs differ in length, a disparity that might have a syntactic cause. A tentative case is made for a cross-level harmony constraint according to which the similarity between adjectives and nouns is not particular to the phonological level but characterizes the linguistic system more generally.

In the course of their history, English wh-relatives are known to have undergone a syntactic change in their prepositional usage: having originally occurred only with pied-piped prepositions, they came to admit preposition stranding as an alternative pattern. The present article presents an overview of this process, showing a modest beginning of stranding in Late Middle English, an increase in Early Modern English, and then a clear decrease in the written language of today, against a more liberal use in spoken English, standard as well as nonstandard. The drop in the incidence of stranding is thus not an expression of a genuine grammatical change but due to notions of correctness derived from the grammar of Latin and affecting written usage. The general trend of the development outlined is mirrored by relative that, with which the pied piping attested in Middle English completely disappeared from the language.