Studies in Language. International Journal sponsored by the Foundation “Foundations of Language” - Volume 36, Issue 2, 2012

The specificity of linguistic structures must be explicitly acknowledged and the investigation of such structures should not be lost in the mass of cognitive sciences. It is claimed that “pure linguistics” meets that requirement. It is founded on the Saussurean definition of a language and on its interpretation by the epistemologist Granger. He showed that language defined in this way is a concept similar to those upon which natural sciences were founded, with similar consequences. That concept entails strict limitation of the field of research, but because of this very limitation, linguistics is made more able to reach precise and secure results and thus to overcome the common shortcomings of the humanities. The legitimacy of pure linguistics does not in the least invalidate that of other language sciences and the quest for cognitive explanations. Moreover, the perspective of pure linguistics appears to be converging with a trend which has been emerging in linguistics towards more strict procedures.

This paper takes up the recent discussion of what controls speakers’ use of the -s and of- genitives, as in the university’s budget and the budget of the university. Most of the paper deals directly with control of that “genitive variation”. It first argues that there is a third variant, as in the university budget. It then discusses why all three just given, for example, are grammatical, but, with a litre of petrol, the alternatives (*a petrol’s litre and *a petrol litre) are not grammatical, arguing that the grammaticality is controlled by whether the word and phrase meanings are compatible with their syntax. It then argues that, where there are equally grammatical alternatives, the speakers’ choice often depends on the same issues of semantic and syntactic compatibility. The last main section of the paper deals with the grammatical status of that third variant, arguing that the university budget truly is genitive.

This research is a first attempt to survey ideophones in the Amazonian Arawak language Alto Perené (a.k.a. Ashéninka Perené). Based on fieldwork data, this study shows that ideophones constitute a separate class of words in Alto Perené in view of their distinctive phonological, morphological, syntactic, and semantic properties. The study also draws on primary and secondary data from three other genetically related neighboring language varieties (Ashéninka Pichis, Asháninka Tambo-Ene, and Kakinte) to demonstrate a moderate degree of interdialect variation. The study suggests the possibility that the following properties may be regional affinities: non-canonical stress assignment; word class-specific reduplication of the word-final syllabic segments -ri, -re, -ro, -pi, -po expressing spatial distribution, intensity, or repeated/durative/open-ended temporal structure of the reported event; productive (V)k-suffixation contributing to the expression of punctual/perfective aspect; syntactic functions of appositional or coordinated predicate, co-verb, complement, and adverb; prevalence of Gestalt packaging of sensory events; a dearth of ideophones describing states.

In the frequency literature, it has been argued that high text frequency is the motivating factor in grammaticalization. This claim is questionable for two reasons: linguistic items with high frequency do not necessarily undergo grammaticalization, and it is not unusual for grammaticalization to occur to linguistic items with low frequency. The problems are that neither context nor pragmatic inferring has been taken into serious consideration. The development of new grammatical meaning, the defining feature of grammaticalization, takes place only in contexts in which pragmatic inferring is possible. Therefore, depending on the properties of the contexts, frequency may or may not directly trigger the development of new grammaticalization. The more reliable independent variable and catalyst in grammaticalization is “critical frequency,” i.e., the number of occurrences of a grammaticalizing item in contexts with the similar pragmatic inferring conditions.

On the basis of Japanese data, this paper proposes a five-level classification of clause linkage. Level IV and Level V of this classification correspond to the epistemic domain and the speech-act domain, respectively, of Sweetser’s (1990) classification. Level I corresponds to the content domain of her classification. In addition, the present paper sets up Levels II and III, as distinct from Level I. It shows that different clause-linkage markers (hereafter CLMs) have different distributions in terms of the five levels. For example, some CLMs can be used in all of the five levels. There are also CLMs that are confined to the lower levels, and also those which are limited to the higher levels. Furthermore, some CLMs have a continuous distribution, while others have a discontinuous distribution. Some of the CLMs exhibit different distributions in terms of Levels I, II, and III. This furnishes evidence for setting up Levels II and III, as distinct from Level I and also as distinct from each other.