Studies in Language. International Journal sponsored by the Foundation “Foundations of Language” - Volume 32, Issue 3, 2008

The topic of word classes remains curiously under-represented in the sign language literature due to many theoretical and methodological problems in sign linguistics. This article focuses on language-specific classifications of signs into word classes in two different sign languages: German Sign Language and Kata Kolok, the sign language of a village community in Bali. The article discusses semantic and structural criteria for identifying word classes in the target sign languages. On the basis of a data set of signs, these criteria are systematically tested out as a first step towards an inductive classification of signs. Approaches and analyses relating to the problem of word classes in linguistic typology are used for shedding new light on the issue of word class distinctions in sign languages.

The assignment of a linguistic sign to a word class is an operation that must be seen as part of the overall transformation of extralinguistic substance into linguistic form. In this, it is comparable to such processes as the transitivization of a verbal base, which further specifies a relatively rough categorization. Languages differ both in the extent to which they structure the material by purely grammatical criteria and in the level at which they do this. The root and the stem are the lowest levels at which a linguistic sign can be categorized in terms of language-specific structure. Further categorization is then achieved at the level of the syntagm. An empirical investigation comparing the categorization of roots and stems in a sample of six languages (English, German, Latin, Spanish, Yucatec Maya, Mandarin Chinese) turns up far-reaching differences. These differences in the amount of categorization that languages apply to linguistic signs at the most basic levels throw into doubt any thesis claiming universal categoriality or acategoriality for roots. Such a static view must be replaced by a dynamic one which asks for the role of categorization in linguistic activity. At the same time, these differences raise the issue of the amount of structure — or of grammar — that is necessary for a human language.1

Late Archaic Chinese is a precategorial language, i.e., a language whose lexical items are not preclassified in the lexicon for the syntactic functions of N and V. This will be shown on the basis of structural-conceptual criteria as those developed by Croft (2000) and Sasse (1993b) as well as on the basis of methodological criteria as those suggested by Evans & Osada (2005). As is claimed in Construction Grammar (Goldberg 1995, 2005), the meaning of lexical items is derived by integrating their own lexical meaning with the meaning contributed by the construction. The construction analysed in this paper is the argument structure construction. Linking between lexicon and syntax is subject to stereotypical pragmatic implicatures (Levinson 2000) that follow a version of the animacy hierarchy. As it will turn out, Late Archaic Chinese does not strictly lack parts of speech. In fact, without the distinction of nouns and verbs at the level of syntax it would not be possible to analyse utterances in Late Archaic Chinese. The only thing that Late Archaic Chinese can do without is noun/verb distinction in the lexicon. This typologically remarkable property is due to a process of morphological change. If such a historical process can take place irrespective of parts-of-speech distinctions, precategoriality in the lexicon cannot be a robust universal feature even if most theoretical approaches take it for granted.

Examining syntactic categories in Tukang Besi, an Austronesian language of Indonesia, we find that there are additions to the traditional fixed categories. In addition to the firmly definable categories of nouns and verbs, there are many lexical items that are precategorial: they may be used, without derivation, with either nominal morphosyntax or verbal morphosyntax. Additionally, there is a class of ‘adjectives’ that display odd behaviour in terms of morphological markedness reversals and functional use, and which, under closer examination, turn out to have a variable categorial status, dependent on the structural position in which they are used, obligatorily appearing as part of the head of their phrase, V in a VP and N in an NP. Morphosyntactic tests for the different claims are given and discussed.

The Miyako-Hirara dialect of Ryukyuan is a switch-adjective language, i.e. it employs both nominal and verbal strategies for its adjectival roots. The choice of the encoding option in the predicative position depends on the information structure of the sentence. If a predicative adjective is in the focus domain, the nominal strategies (adjectival root + munu, reduplicated forms) are used. If a predicative adjective belongs to the presupposition, the verbal strategy (-kaɨ forms) is employed. We argue that Miyako-Hirara employs a switch-adjective encoding in order to retain the functional distinction between the predicates in focus and those in the domain of presupposition, in the absence of other means of morphological marking for the informational status of constituents.

This paper presents empirical evidence for a theory of syntactic categories in the tradition of categorial grammar, in which more complex categories are derived from simpler ones by means of category formation operators. In Jakarta Indonesian, almost all words and larger expressions belong to a single open syntactic category, S(entence), while a small residue of semantically heterogeneous items belong to a single closed syntactic category S/S. The theory predicts that in first-language acquisition, simpler categories are acquired before more complex ones. Thus, for Jakarta Indonesian, it predicts that the category S is be acquired before the category S/S. Examination of a naturalistic corpus of almost one million utterances provides support for this prediction, deriving from errors of overgeneralization, in which members of S/S exhibit the distributional properties of members of S.

This paper reports an experiment in which adult native speakers of Dutch were asked to categorize nonce stems. These were presented without any morphological or other information about their potential class-membership. We expected that subjects would be able to categorize these words solely on the basis of phonological information. Nouns in Dutch have a richer possible syllable-structure than verbs (Trommelen 1989) and therefore, we expected that stimuli with a ‘nominal’ syllable make-up could be identified by the subjects as nouns where the other stimuli were ambiguous between nouns and verbs. The results show that this is indeed the case, adding to the evidence that native speakers of a language are able to use phonological information to categorize the words of their language.

While nouns are thought to be more easily borrowed than verbs, this investigation demonstrates that there may be limitations on noun borrowing into certain languages. The case of Zuni is examined, in which conditions of contact similar to that of neighboring languages nevertheless result in a different treatment of the noun lexicon. The possibility of borrowing natural kind nouns into Zuni exists alongside a tendency against borrowing nouns of the artifact type. It is argued that the source of this tendency against artifact noun borrowing in Zuni is the grammatical complexity of the lexical semantic representation for these nouns: grammatical complexity in Zuni noun roots appears to be dispreferred. These findings belie the claim of Thomason and Kaufman (1988) that “any linguistic feature can be transferred to any other language” given an appropriate degree of contact.

This paper emphasises the explanatory possibilities of a parts-of-speech theory in which the notions of differentiation, flexibility, and rigidity are not perceived as mutually exclusive typological paths, but as potentially compatible grammatical strategies. The type of analysis such a theory makes possible is first applied to Spanish quantifiers and subsequently extended to degree words expressing intensification in modern Romance languages. The paper also pays attention to the formal mechanisms associated with the expression of degree words. One of the most significant conclusions is that degree words classified as differentiated adverbs tend to function as modifiers of modifiers. The modification of predicates and that of terms (especially, mass nouns) are roles usually played by flexible modifiers.

This article argues that in addition to the major flexible lexical categories in Hengeveld’s classification of parts-of-speech systems (Contentive, Non-Verb, Modifier), there are also flexible word classes within the rigid lexical category Noun (Set Noun, Sort Noun, General Noun). Members of flexible word classes are characterized by their vague semantics, which in the case of nouns means that values for the semantic features Shape and Homogeneity are either left undetermined or they are specified in such a way that they do not quite match the properties of the kind of entity denoted by the flexible item in the external world. I will then argue that flexible word classes constitute a proper category (i.e. they are not the result of a merger of some rigid word classes) in that members of flexible word categories display the same properties regarding category membership as members of rigid word classes. Finally this article wants to claim that the distinction between rigid and flexible noun categories (a) adds a new dimension to current classifications of parts-of-speech systems, (b) correlates with certain grammatical phenomena (e.g. so-called number discord), and (c) helps to explain the parts-of-speech hierarchy.

In this paper we use the Theory of Functional Discourse Grammar to show that parts-of-speech (PoS) and dependent clauses (DCs) can both be defined in terms of the functions they may express. On the basis of this parallel treatment, we predict that the functional possibilities of PoS and DCs in languages are comparable. This hypothesis is tested using a sample of 23 languages. The results do show similarities between the functional patterns of the two types of constructions.