The correlates of finiteness in universal grammar
There are two basic traditions in studying finiteness. One, which explains the etymological basis of the term, sees finite forms as 'completed' by their morphological marking for person and tense. The other treats finite forms as those which can occur in independent clauses, whereas non-finite forms only occur in embedded contexts (with due allowance being made for such strange grammatical animals as 'historical infinitives' and the like). We will first seek to show on cross-linguistic grounds that the only defensible avenue of investigation is the latter. En passant, this will involve a criticism of mainstream generative views on the question that have, until recently at least, largely opted for the former definition. The parameter of nominative case and verbal agreement that has attracted so much generative attention is a real one but is orthogonal to the question of finiteness. In more positive vein, we will then move on to a case study of the evolution of the finite/non-finite distinction in the transition from Latin to Romance, showing in particular how the morphological changes in that family have led on the one hand to the loss of the Accusative and Infinitive construction and on the other to the rise of infinitives, gerunds and participles all bearing person/number agreements. We conclude by suggesting that the underpinnings of the traditional categories of finiteness are to be found in the area of modality rather than of tense or of person agreement.
Sonia Cristofaro (University of Verona)
Finiteness and the cross-linguistic coding of subordination
The opposition between finite and non finite verb forms is usually defined in terms of a number of features such as absence of tense, mood, aspect or person marking (which corresponds to loss of verbal properties) and possibly case marking on the verb or coding of verb arguments as possessors or obliques (which can be regarded as nominal properties). These features need not occur in exactly the same combinations cross-linguistically, and none of them can actually be said to be necessary for the definition of finiteness. However, cross-linguistic analysis shows that the various features involved in the notion of finiteness occur in predictable correlation patterns.
This paper presents the results of a cross-linguistic study of subordination systems, based on an eighty language sample. Many of the features that are relevant to the cross-linguistic coding of subordination are the same involved in traditional definitions of finiteness. Verb forms coding the dependent event may display nominal morphology such as case/adposition marking, or elimination of verbal morphology such as tense, aspect, mood or person marking. Verb arguments may be coded as possessors or obliques.
The cross-linguistic distribution of these phenomena responds to the following implicational pattern
(1) Agreement elimination, Case/adposition marking on the verb, coding of arguments as possessors or obliques -> Tense, aspect or mood elimination
Moreover, these phenomena can be ranged along an overall frequency hierarchy (>> = `more frequent than'):
(2) Elimination of TAM distinctions >> Elimination of agreement distinctions >> Case/adposition marking on the verb >> Coding of verb arguments as possessors or obliques
The implicational pattern, as well as the frequency hierarchy, can be explained in terms of the cognitive status of subordination. Following Langacker, subordination is regarded here as a cognitive situation whereby one of two linked events (the dependent event) is construed not in its own perspective, but in the perspective of the other event. As a result, the dependent event is not conceptualized as an autonomous process any more (as is usually the case with events), and is possibly conceptualized as a component of the main event (that is, in Langacker's terms, as a thing). This is reflected at the morphosyntactic level by loss of verbal properties and acquisition of nominal properties on the part of the verb coding the dependent event. This approach allows to account for the two most significant facts about the implicational pattern and the frequency hierarchy, namely (i) acquisition of nominal properties entails loss of verbal properties, and (ii) loss of verbal properties is more frequent than acquisition of nominal properties.
Walter Bisang (University of Mainz)
Finiteness, obligatority and grammaticalization
Finiteness has been discussed by different functional typologists in terms of scales (Noonan 1985: 57; Givón 1990: 853 - 854; Hengeveld 1998). The categories typically involved are tense, pronominal agreement, case marking, nominalising affixes and articles/determiners. In my approach, I shall try to look at finiteness from the perspective of obligatority and grammaticalization and I shall try to show how functional and formal approaches can be related to each other. Obligatority will be understood in terms of Lehmann (1995: 139) who departs from the freedom of the language user "in either specifying the category by using one of its subcategories, or leaving the whole category unspecified". Obligatority thus means that the language user has to refer explicitly to a certain grammatical category by selecting one of its subcategories. If this type of obligatority applies to independent clauses, languages can create asymmetries in the sense that a finite clause has to be marked for a given grammatical category which cannot occur at all or which can occur only with a reduced set of subcategories in the non-finite clause. From a cross-linguistic point of view, it seems that the only categories that can be used for this purpose are tense, mood, person and markers of illocutionary force. The categories of tense, mood, person and illocutionary force are characterised by their high degree of generality in terms of Bybee (1985). The principle of generality states that the semantics of a marker need to be maximally general if it is to be combined with a large class of lexical items. This criterion is particularly pertinent for a category which is obligatory in every independent clause. The fact that the above four grammatical categories are highly recurrent as the final products of processes of grammaticalization as observed by Talmy (1988) and the fact that they are linked to the finite/non-finite distinction can thus be accounted for in terms of generality. From a formal perspective, the categories relevant to finiteness belong to INFL, or more precisely to T or AGR. A very interesting approach to finiteness is presented by Shlonsky (1997) on the basis of T. "[T]he essential difference between full and small (or reduced) clauses is that the former are CPs, and are hence endowed with a TP projection, while the former are clausal chunks that may vary in size--that is, in the number of functional projections they include--but they crucially lack TP and hence CP" (Shlonsky 1997: 6). In my view, the central syntactic status of T is its high degree of generality which allows it to be linked to any independent clause. Once this category becomes obligatory, it can be reinterpreted for purposes which are purely syntactic and thus arbitrary in the sense that they are no longer linked to the semantics of tense (on the relevance of obligatority and frequency of occurrence in the context of online access and language acquisition cf. Slobin 1997, 2001). The same can also be said with regard to person (in particular the marking of the subject) and its obligatority. To conclude my paper, I would like to show that it is the obligatority of the categories of tense and person, no matter whether expressed morphologically or syntactically, which is relevant to the finite/non-finite distinction. I shall do this by discussing data from Sranan (Plag 1993) and Chinese (Li 1990, C.-T. J. Huang 1989, Y. Huang 1994). In Chinese, neither tense nor person is obligatory, whereas in Sranan person is obligatory in independent clauses. As was shown by Plag (1993), there is a clear-cut distinction betweeen finite and non-finite clauses (pro vs. PRO) in Sranan. In Chinese, the situation is much more controversial and depends on a number of pragmatic criteria. Some aspects of what I am discussing here is presented in some more detail in Bisang (forth.).
References:
"Finite" and "nonfinite" from a typological perspective.
The differentiation between finite and nonfinite forms is among the most complex domains of grammatical description. This is probably the reason why this category plays only a marginal role in typology. Clarification is further handicapped by the tradition of school grammar, which simply copies its categories from specific languages. In order not to be trapped by this tradition, a distinction between semantic and morphologic finiteness is proposed, one which makes it possible to clarify notoriously complex cases as e.g. movable predicate affixes in Amerindian and Munda languages, the "finite infinitive" in Portuguese, etc. Two areas are investigated: Complex predicate formation, where auxiliary formation is distinguished from multiple finite coverbal modification, and complex sentence formation with degrees of finiteness in secondary predicates (masdar, attributive and converbal constructions).
Christine Dimroth (MPI, Nijmegen) Petra Gretsch (MPI, Nijmegen) Peter Jordens (Vrije U. Amsterdam) Clive Perdue (U. Paris VIII) Marianne Starren (U. Nijmegen)
Finiteness in first and second language acquisition of Germanic languages
This paper investigates the linguistic operations involved in the child's and the adult's acquisition of finiteness in Germanic languages. We follow a very broad definition of finiteness and take the notion to refer to all kinds of devices whose function is to relate the state of affairs described in an informationally-structured proposition to its spatio-temporal anchorpoint. By assigning finiteness to an utterance, a speaker takes on this abstract relation and gives it a value. This value is basically polar (yes/no), and allows the speaker to validate, question or request a state of affairs in relation to its anchorpoint.
At subsequent acquisitional stages we examine
(i) the devices which express a spatio-temporal reference point and anchor the utterance in the context
(ii) the devices expressing the relation between this anchorpoint and the state of affairs expressed in the utterance
For purposes of exposition, we call "topic" the linguistic correlates of the operations of (i), "linking" those of (ii), whereas the term "predicate" represents the state of affairs expressed in an utterance. These structural units reflect the very transparent information organisation of early learner language, i.e. topic element(s) before linking element(s) (if not let implicit) before predicate. They give us a theory-neutral terminology that allows to compare the linguistic and cognitive operations involved in child and adult language acquisition.
Intuitively, there seems to be a big difference in the way in which the expression of the linking operation is acquired by children and adults. It will be argued, however, that in fact, both groups of learners go through a so-called "conceptual ordering stage" where linguistic structure mirrors information structure and the linking operation is expressed with the help of a closed class of "linking words" (uninflected lexical items expressing a definable range of values). In other words, we argue that the information organisation of early - child and adult - learner language is directly reflected in the use of anchoring and linking devices. The significant linguistic difference between first and second language learners lies then exclusively in the acquisition of inflexional morphology: L1 learners acquire topic- and linking-expressions in parallel, whereas adult L2 learners do well with topics, but are often unsuccessful in acquiring the morphology associated with the linking-operator (grammaticalized finiteness).
>From this starting point, we bring together a number of individual studies on the acquisition of Germanic languages, and describe the L1 and L2 acquisition process, comparing as closely as possible the internal organisation of topic and linking, and how they interrelate over time.
Elena Kalinina (Moscow State University)
Non-finite forms in finite sentences: finiteness and grammatical structure.
a.. Preliminaries. Finiteness is usually associated with a set of "finite" categories - person, tense, modality. In the languages where the opposition of finite vs. non-finite is the property of the verb form these categories are marked on the finite verb only. Non-finite forms and predicate nouns cannot be marked for finite categories and require the presence of a copula or copular verb. Now finiteness is assumed to be the property of a clause rather that of a verb form. However, the validity of the distinction between finite and non-finite verb forms has never been tested on a cross-linguistic scale. So, the focus of my cross-linguistic investigation was the universality of the finite vs. non-finite distinction at the level of verb forms. The tentative hypothesis was that the absence of this distinction would affect other parts of grammatical structure, namely the set of finite categories and the way these categories are expressed.
b.. Cross-linguistic investigation. My sample contained 40 languages where this distinction turned out to be invalid. Two kinds of forms were investigated: a) those functioning in relative and complement clauses (referred to as participles and/or nominalizations in grammars); b) those functioning in adverbial clauses and generally termed converbs. Two groups of phenomena have been observed: a) the same form can be combined with finite markers (person, tense) when functioning in finite clauses and nominal markers (case, possessive markers, genitive complement) when functioning in non-finite clauses; b) a non-finite form, when functioning as the main clause predicate, expresses some grammatical meaning. For instance, in Bagwalal participles in main clauses express the meaning of habitual. In Nenets, converbs in main clauses express remote past.
c.. Theoretical implications. The absence of the finite vs. non-finite distinction entails revising theoretical and descriptive views on certain areas of grammar.
a.. Expression of grammatical categories. The fact that the same forms take both verbal (person) and nominal (case) morphology markers renders them problematic for adequate description. However, some clues are provided by the behaviour of predicate nominals: in the languages under analysis, they can directly combine with verbal markers, like person and even tense. This set of related phenomena can be treated in two ways. On the one hand, the treatment can be designed so as to allow for the word class change. Under the alternative approach, word class remains unaltered, whereas grammar markers are permitted to be attached to both nouns and verbs.
b.. Another basis of distinction. Bagwalal participles and Nenets converbs (mentioned in 2) can also be handled in two ways - the case can be made for homonyms or for the invariant grammatical meaning. I argue that the former way is more adequate. This invariant meaning can involve information structure, so that the opposition of verb forms can be described as the opposition of focus-bearing or assertive forms vs. non-assertive forms expressing presuppositions. This distinction is fully independent from that of finite vs. non-finite.
c.. Copulas and copular verbs. In the languages of the sample, they could no more be regarded as carriers of finite categories, accompanying items unable to express them. I am going to demonstrate that copulas and copular verbs in the languages of the sample express information structure categories (topic, focus, illocutionary force) and relative tense. As such, they are subject to special rules and are not primarily used to distinguish between predicate nouns and verbs, nor finite and non-finite verb forms.
d.. Conclusions on finiteness. a) finiteness as the property of verb forms is not universal; b) finiteness is the property of a clause/sentence and does include a set of categories that could be denoted as finite; c) these categories include information structure and illocutionary force; d) the fact that finite categories are not marked on finite verb forms entails theoretical and descriptive decisions regarding the way grammatical structure is organized.
Can finiteness be gradual? - Degree of finiteness and infinitive categories in languages without person conjugation/indexing and without an obligatory subject -
In Indo-European languages, the inflectional categories which are traditionally regarded as constitutive for the finiteness are: person(/number), tense, and modality (and, in addition, genera verbi in languages with special diathesis conjugation such as Sanskrit and Greek). A further criterion crucial for the finiteness in these languages is, beside the inflecional property of the predicate itself, the case of the "subject" of the clause: a non-nominative case for an/the infinitive form, and the nominative for the finite form of the predicate, which indexes ("agrees with") the person (and the number) of the referent of the argument in the nominative, i.e. the so-called "subject". But how can we identify finiteness and, vice versa, an infinitive in languages without, firstly, these inflectional categories, either those lacking inflection at all like Chinese, or those with inflection that is, however, not reflecting the category of person. Japanese belongs to the latter type. Moreover, for both Chinese and Japanese it is not necessary to mention the "subject" of the clause, so that a predicate is capable of forming a clause, and even a sentence, alone. Furthermore, not only for Chinese, a prototypical aspect language, but also for Japanese that can be regarded as showing a tense inflection, aspectuality of the same $B%a(Btense inflection$B%b(B overrides its temporal indexicality, depending on the type of the clause. The higher the degree of subordination, the more aspectual is the function of the predicate form. In my own paper (Ono 1993) I tried to clarify the situation in Japanese, introducing a set of factors for finiteness: e.g. absolute tense (reference) and politeness besides the inflectional forms and sentence positions. Based on this, I will try to develop a multifactorial scale of finiteness and infinitivity.
Reference:
Ono, Yoshiko 1993 "Does Japanese have coordination?" In: Ebert, Karen (ed.) Arbeiten des Seminars fuer Allgemeine Sprachwissenschaft der Universitaet Zuerich.