Two ill-defined and controversial sound developments of late Old Spanish, namely the asymmetric diphthongizatioris ié > ì and ué > e, become better understandable if one selects as starting points a set of morphological rather than phonological conditions. Both verbal inflection and suffixal derivation can be invoked, including the rivalry of certain characteristic groups of preterits (-iemos, -ieste(s) beside -imos, -iste(s) etc.) and the competition of pairs or clusters of functionally more or less related suffixes: -ero beside -uero, -eño alongside -ueño—in addition to the pressure exerted, in the ranks of hypocoristics, by the close-knit series -ico, -ito and -in(o) on isolated -iello.

Syntactic change is viewed as a process that is critically dependent on the surface properties of language and essentially independent of grammatical derivations. In this framework, the genesis of the reflexive impersonal construction in Portuguese is shown to have no formal connection—not even a merely descriptive one—with the transformations of Agentization and Object Fronting that were involved in the derivation of its ancestor, the reflexive passive. Evidence from other sources supporting the surface framework is briefly discussed, and an explanation is offered for its validity in terms of the real-world circumstances under which language change occurs.

The negative element non appears in Italian comparatives when the speaker presupposes that his statement contradicts someone else's or his own previously held belief. Though this non appears in the comparative clause in the surface, it originates in a higher abstract S in embedded position. By proposing an abstract S in embedded position, we can account in a unified way for many sets of facts which would go unrelated in either a presupposition-dependent syntax model or an interpretive approach. This use of non is thus an example of a presuppositional fact which is accounted for by a syntactic analysis.

Various proposals for explaining the origin of the base constituent of the Latin imperfect tense formation in *bā- (e.g. amābam ‘I was loving’) are examined. With an explicit phonological derivation, as well as critical evidence from Latin syntax and word formation, it is demonstrated that the present participle, e.g. *aman(t)s, is the source of the base member of this construction; and that it is joined, as traditionally assumed, to a form of the verb *bhū- ‘be'.

The still obscure formal prehistory of the verbs of the Germanic 3rd weak class is investigated from the standpoint of their functional characteristics—rather than, as is more usual, simply their inflection. This approach is profitable because, among the statives of Gothic Cl. III, a significant number of factitives are found which, as indicated by independent evidence, probably continue PIE presents in *-o-ye/o-. Factitives are not anomalous in Cl. III, since the statives of this class are shown to be functionally archaic in Gothic. The purely formal integration of o-ye/o-factitives into originally stative Cl. III argues for *-ai- in the so-called e-forms of Cl. III in pre-Gothic, and probably in Proto-Germanic

Yuman languages allow an unusually large number of instances of non-systematic metathesis affecting a wide variety of consonant types. When we seek an explanation for such apparently random facts, various structural features of Yuman languages (both phonological and morpho-syntactic) are found to produce surface variations which look like metathesis in paradigmatically related forms. This situation, it is claimed, is the catalyst which allows other types of metathesis to establish themselves. The classes of segments involved in the process, as well as the direction in which it proceeds, are matched against some proposed universals; some of these gain additional support from the Yuman data, while others do not fare as well.

A replication of the study by Berlin & Kay 1969, controlling for saturation in the color stimulus materials, lends support to their findings about basic color category universals.

Significant insights into the organization of kinship terminologies, and into methods for their analysis, have resulted from the construction of a computer program that reduces the ranges of kin-terms to kernel kin-types, identifies the operable equivalence rules, and uses the identified rules and kernels to re-expand the full range of the kin-terms. The program is based on the work of Romney and of Lounsbury, and has been run successfully on a number of terminologies.

Findings on language use among peers in inter-ethnic work situations in three African cities are presented. A clustering algorithm is used to display relations of similarity and dominance among clusters of speakers according to their language choice in this situation. The explanation of these data, as displayed by the algorithm, promotes the development of theory regarding socio-verbal interaction in uncertain situations which call for reciprocal participant accommodation, such as conversations with peers at work. Specifically, the data suggest that neutrality strategies in language choice are perceived as necessary to make such interactions successful.