Aside from a few incidental references to ‘drift’, Sapir used this word in three significant contexts (once as a chapter heading in his 1921 book. Language); one discovers certain discrepancies among the scopes he attached to this key word on those separate occasions. Most of the doubts and dissensions among Sapir's contemporaries and successors are traceable to the fact that the discussants have not discriminated among those three pronouncements. The keenest critic was Sapir himself who, in 1933, divested the substance of Chap. 7 in his book of any residual mystique—and, by implication, completely abandoned some of the ideas added at the beginning of Chap. 8.

This paper examines the claims in Natural Generative Phonology (NGP) that phonetically conditioned rules do not make reference to word boundaries, and that sandhi rules constitute a type of rule ‘intermediate between P-rules and MP-rules’. It is concluded that sandhi rules without (other) morphosyntactic information are like P-rules except for the presence of a word boundary, supposedly a non-phonological boundary. If this is the only reason for separating these sandhi rules from P-rules, then the difference is merely one of definition; and the claim that P-rules may not be conditioned by word boundaries is vacuous, since it is true by definition. The reason that proponents of NGP insist upon the exclusion of word boundaries from P-rules is their adherence to a requirement that no non-phonological information play a part in phonological generalizations. If it can be demonstrated that word boundaries do indeed play a part in P-rules—and if, further, the definition of ‘P-rule’ can be established without circularity (i.e. without including or excluding word boundaries by definition)—then one of the major distinguishing tenets of NGP is shown to be without basis, since it is clear that word boundaries are indeed not purely phonological.

This paper presents an analysis of headed and free (headless) restrictive relative clauses in Catalan in the framework of the Extended Standard Theory. Rules of grammar are considered an aggregate of independent dimensions that belong either to core grammar or to peripheral grammar. The peripheral dimensions restrict the core aspect, but do not conflict with it.

The rules involved in the derivation of headed and free relatives are (a) WH-movement and (b) deletion of the WH-phrase in COMP. One filter blocks an empty COMP, and another blocks a doubly-filled COMP. Deletion in COMP has an obligatory character in standard Catalan; this is a peripheral dimension that is unproblematic for the language learner. Catalan deletes any category in COMP up to recoverability, including PP's; this is the core situation found in French, but it is restricted by addition of a peripheral dimension that allows only NP-deletion.

The interpretation of the core/periphery dichotomy developed in this paper may provide a fruitful framework for comparative grammar and historical research.

Nominative/absolutive case and verb agreement are, in many languages, indicators of a category which is here called VIEWPOINT: the perspective from which the speaker describes the event. The order of NP constituents in a sentence encodes ATTENTION FLOW, which is the order in which the speaker expects the hearer to attend to them. Split ergative case-marking patterns are shown to reflect conflicts between the most natural viewpoint and attention-flow assignments. It is argued that the characterization and grammatical marking of an event as first-hand or inferred knowledge for a speaker, and as intentional or inadvertent for an actor, can be described in terms of whether the entire event or only its terminal phase is directly accessible to the conscious mind of the speaker and the actor, respectively; and that these categories can also be described in terms of attention flow and viewpoint.

Coahuilteco, an Indian language formerly spoken in southern Texas and northeastern Mexico, possesses a unique system of concord by which non-subject NP's, including clausal arguments, show person agreement with the subject NP of their clause. Coahuilteco is typologically SOV, with modifiers and determiners/relators following the noun in that order. Subject concord is marked by personal suffixes on the determiner/relator, or alternatively by person-marked particles or auxiliaries. The system serves to bracket constructions and permit tracking of subject scope (or verb dependency relations), fulfilling some of the same functions as switch-reference.

Negation in French can be marked redundantly—since a pre-verbal particle ne may accompany a second, usually post-verbal, marker of negation. However, the pre-verbal particle is often deleted in spoken French. This deletion may be inconsistent with the usual typological characterization of French as an SVO language. In this paper, the use of ne by 37 speakers from the region of Tours is evaluated, and the data are judged by the Sankoff Variable Rule Program (Varbrul 2). The negative particle is found to vary with a complex of linguistic, stylistic, and social factors. The historical record and the data presented here suggest that ne is being lost. This on-going syntactic change may be dependent on another, as yet uncompleted, change: the fusion of the subject clitic and verb.