The grammar of Kasem is being analysed in terms of a hierarchical model (cf. Halliday, 1961; Bendor-Samuel, 1963), the (ascending) levels of relevance to this paper being word, phrase, clause, string, and sentence. (The ‘string’ is the unit which functions in the sentence and consists of a tightly bound sequence of clauses, the non-initial clauses operating under extensive restrictions with respect to their structure. This type of syntactic unit appears to be characteristic of many Gur languages and is often referred to as a ‘series’ or ‘serial construction’. For a recent discussion in some detail of this unit, under the term ‘clause clusters’, see Pike, 1966: 55–78.) At the word level in this hierarchy, a system of categories can be set up to account for the form of every verb word, i.e. so that every verb word can be completely ‘parsed’ or described in terms of these categories. But one of the characteristic features of this system of categories is that there is extensive neutralization between the categories. The question, then, that this paper seeks to discuss is this: are the ambiguities inherent in the verb word resolved within the language as a whole? An answer is here presented which seeks to relate the resolution of the ambiguities to the different levels of the grammatical hierarchy, as set up.

Our purpose is simple: to apply a rudimentary kind of logical analysis to the problem, which sometimes arises in linguistic typology and in comparative linguistics, of whether there are any languages with fewer than two vowels. The principal reason why this question has occupied the attention of linguists is probably the fact that Proto-Indo-European is sometimes said to have had only one vowel phoneme (see especially Hjelmslev (1936–7), Borgstrøm (1949, 1954) and Lehmann (1952), following Saussure (1879). Yet this was precisely one of the features of PIE reconstruction that Jakobson selected for attack on typological grounds, reasoning that ‘a conflict between the reconstructed state of a language and the general laws which typology discovers makes the reconstruction questionable’ (1958: 23).

We explore the implications of a particular approach to learning for an architectural question in phonology. The learning approach follows the principle of Minimum Description Length (MDL), which has recently been used for learning in both constraint-based and rule-based phonology. The architectural question on which we focus is whether the grammar allows language-specific statements to be made at the level of the lexicon, as was assumed in early generative phonology, or whether such statements are prohibited, as is commonly assumed within more recent work. We show that under MDL, the architectural question has real empirical implications: across a range of seemingly natural representational schemes, an ability to make language-specific statements about the lexicon is needed to ensure the learnability of an important aspect of phonological knowledge.

In this paper, we present an analysis of so-called determinerless PPs in German, i.e. prepositional phrases that allow singular count nouns to occur without an accompanying determiner, despite other rules in the grammar requiring the presence of the determiner. The analysis is based on annotated corpus data, which are fed into a statistical classifier (applying logistic regression). Superficially, the syntax of bare prepositional phrases is difficult to capture, and intuitions cannot be easily elicited. The analysis is based on data sets for two pairs of German prepositions: mit ‘with’ and ohne ‘without’, and über ‘over, above’ and unter ‘under, below’. The results of the classifiers applied to annotated data indicate which syntactic, morphological and semantic features are responsible for determiner omission. We are able to detect common properties of all four prepositions, as well as preposition-specific, and idiosyncratic properties. The apparently unsystematic conditions for determiner omission can be discerned by tracing the interaction of these properties.

This is a book written by a philosopher for philosophers; it needs a philosopher to review it properly. A review by a non-philosopher in the pages of a linguistics journal is justified, however, by the fact that the core of the book consists of a series of proposals concerning one of the fundamental questions of lexical semantics: what are we to make of the apparent variability in the meaning of a word from context to context? (This fact is not immediately inferrable from the title.) The present review focuses exclusively on this topic; it should be borne in mind, however, that the author's exposition of his views on word-meaning occurs as part of a more general discussion concerning the nature of analogy, and its role in literary, religious and legal discourse.

The syntax of Clitic Left Dislocation (CLLD) has been widely debated due to its mixed properties, which in some cases indicate movement (e.g. island sensitivity, certain connectivity effects) and in other cases base generation of the CLLD-ed phrase (wide scope, lack of weak crossover). In this paper we discuss scope facts with CLLD in Greek, revealing a contrast depending on the type of quantifier. We present experimental evidence that whereas CLLD-ed plain indefinites take wide scope, CLLD-ed numerals can get a low scope interpretation. We argue that the inverse scope interpretation with CLLD-ed numerals is only apparent, presenting, in fact, an instance of split scope between the degree quantifier and the existential operator. This analysis presents evidence in favor of a movement analysis for CLLD, thus patterning with the observation that binding reconstruction is possible. At the same time, the non-availability of scope reconstruction with CLLD is attributed to stricter locality constraints which have been discussed for quantifier raising as opposed to other types of movement and dependencies.

Langendoen and Bever (1973) contended that both expressions in (1) are ungrammatical in English, despite the acceptability of the first to native speakers of English.

(1) (a) a not unhappy person

(b) a not sad person

We came to this conclusion because we believed, first, that all grammars of natural languages should be subject to a constraint ‘M’ that no syntactic rule is able to make use of the morphological structure of lexemes (1973:402); and, second, that we had shown that no grammar that satisfies M can distinguish the grammaticality of(1a) from that of (1b). Our exact wording of M was as follows: ‘no syntactic transformational rule is permitted to make use of the internal morphological structure of lexical items’. Since the only syntactic rules of the standard theory, within which we were operating that are not transformations are base-categorial rules, which by definition make no use of the internal morphological structure of lexical items, the word ‘transformational’ in our formulation of M is unnecessary.

The English Preposing in PP construction (PiPP; e.g., Happy though/as we were) is extremely rare but displays an intricate set of stable syntactic properties. How do people become proficient with this construction despite such limited evidence? It is tempting to posit innate learning mechanisms, but present-day large language models seem to learn to represent PiPPs as well, even though such models employ only very general learning mechanisms and experience very few instances of the construction during training. This suggests an alternative hypothesis on which knowledge of more frequent constructions helps shape knowledge of PiPPs. I seek to make this idea precise using model-theoretic syntax (MTS). In MTS, a grammar is essentially a set of constraints on forms. In this context, PiPPs can be seen as arising from a mix of construction-specific and general-purpose constraints, all of which seem inferable from general linguistic experience.