2.1 Aspects of typological classification in spoken languages
Typology can be defined as “the classification of languages or components of languages based on shared formal characteristics.” (Whaley 1997:7). Typological classification is thus based on the grammatical variation found across languages. Data from a representative sample of languages is required in order to allow for reliable generalizations (Whaley 1997; Payne 1997). Based on a large amount of data from spoken languages, different typologies based on syntactic and morphological properties have been proposed.
Word order, or basic constituent order to be more precise, is a very common property on which a typological classification is based. Firstly, languages can be classified as to whether they have a flexible or fixed constituent order (Dryer 2007; Whaley 1997). Cavineña (Tacanan; Bolivia) is an example of a language with flexible word order, as is illustrated in (1) with just two of the possible orders. However, languages with a fixed constituent order may also allow for other orders under specific circumstances. It is for this reason that the notion of basic word order is used. The most widespread order among languages is Subject-Object-Verb as in Imbabura Quechua (Quechuan; Ecuador) (2), followed by Subject-Verb-Object as in English (Germanic; United Kingdom), followed in turn by Verb-initial orders as in Maori (Polynesian; New Zealand) (3). Languages with Object initial word order (OVS/OSV) exist, but are extremely rare.
To determine basic word order, linguists take into account which is the most frequent occurring word order, the one that is least marked, and the one that is pragmatically most neutral (Whaley 1997). Also, it has been proposed that there are correlations between basic word order and other properties (Greenberg 1963). When a language is of the SOV type, for instance, it is often found to have postpositions and the genitive modifier following the noun. In contrast, languages of the SVO type tend to have prepositions and genitives generally preceding the noun.
In addition, languages are commonly classified based on their morphological typology, i.e. the amount of affixation and fusion. A language with monomorphemic words is called isolating, a language with polymorphemic words is synthetic. If the morphemes in a word are easily segmented, the language is called agglutinative; if not, it is called fusing (Payne 1997; Whaley 1997). The extreme case of agglutination also allows for noun incorporation, that is, a word can contain several grammatical stems. Languages of this type are called polysynthetic languages. The word order type of a language does not appear to be related to its morphological typology. The free order language in (1) and the SOV language in (2) are both agglutinative whereas the VSO language in (3) is isolating.
Other areas of grammar that have received considerable attention from typologists are sentential negation and noun categorization. A typological classification has been proposed for negation by, for instance, Dahl (1979) and Payne (1985). They suggest three different strategies for encoding sentential negation: negative affixes, negative particles, and negative auxiliaries. The first two strategies are exemplified by the examples in (4) and (5). In Turkish (Altaic; Turkey), sentential negation is expressed by the verbal suffix –mI (the vowel harmonizes with the stem) while in Russian (Slavic; Russia), the negative particle ne is used.
Aikhenvald (2000) describes a typology of noun categorization devices, where she distinguishes six types of classifiers – noun, numeral, genitive, verbal, locative and deictic classifiers. Noun classifiers are free morphemes, which classify the noun in a specific, generic class. Numeral classifiers are devices of quantification, which appear either as free or as bound morphemes while verbal classifiers combine with verbs and classify one of the (nominal) arguments of the verb. The Cherokee (Iroquoian; United States) examples in (6) show that the verb néé’a (‘give’) combines with different classificatory morphemes depending on physical properties of the object given (Aikhenvald 2000:161; CL = classifier).
Genitive classifiers appear in possessive constructions. For locative and deictic classifiers, more examples seem to be needed “before their typological profile could be fully established” (Aikhenvald 2000:172). This is probably the reason why Grinevald (2000) leaves out these last two types, and bases her typology on morphosyntactic properties of the classifiers.
One other grammatical phenomenon that has been studied extensively is verbal agreement, and again, interesting typological variation has been found. First, languages with verbal agreement have to be distinguished from languages in which verbs do not agree (null agreement languages). Second, within the former group, we find languages with a poor agreement system, e.g. Dutch (Germanic; Netherlands), and languages with a rich agreement system, e.g. Spanish (Romance; Spain). The difference between the two types of languages is illustrated in (7). Note that only in Spanish is every feature combination (person and number) spelled out by a different phonological form.
Languages with rich agreement, such as Spanish, commonly allow for pro-drop: the pronominal subject can be dropped as “the information can be determined by the agreement morphology on the verb” (Whaley 1997:289). However, pro-drop may also be observed in languages without agreement, such as Chinese for example (for a discussion, see Lillo-Martin 1986).
So far, we have only been concerned with subject agreement. However, in addition to subject agreement – and independent of their classification as a poor or rich agreement language – some languages also display verbal agreement with the object. This is illustrated by the Itelmen (Chukotko-Kamchatkan; Russia) example in (8), in which the verb agrees with its direct object by means of the suffix –um.
Agreement typically operates “according to a hierarchy of relations (Whaley 1997:153). If the verb of a certain language agrees with only one of its nominal arguments, this will typically be the subject. If it agrees with two arguments, these will be the subject and the direct object. In the rare cases of languages where the verb agrees with three arguments, the third argument it agrees with is its indirect object. The agreement hierarchy (Whaley 1997), given in Figure 1 predicts this situation.
A fully-fledged typology of agreement is still under development. Corbett (2006) attempts to create a typology of agreement using grammatical relations, but concludes that these are not nearly sufficient.
It is interesting to investigate whether the different morphological and syntactic properties sketched above are also helpful in constructing a typology of sign languages, or whether other (or additional) classifications are necessary. In the following section, a brief overview of research on variation across sign languages is given, focusing on a number of phenomena for which sufficient data are available.