Search Results

This introductory chapter defines periphrasis and briefly discusses previous studies of the phenomenon (types of periphrasis and defining criteria previously suggested). It outlines the typological diversity of periphrasis in terms of the word classes involved and the grammatical features realized by it, both typical and exotic. Then the discussion turns to the most common functions of periphrasis, such as making the inflectional paradigm of a language complete, and to the interaction of periphrasis with other morphological and syntactic phenomena of the given language. Since several chapters in the volume employ a set of criteria for periphrasis worked out within the Canonical Typology approach, these are listed in the introduction. The criteria prove valuable for the analysis of less familiar instances of periphrasis, such as nominal case in Nenets, the large periphrastic paradigms of Archi verb, and embedded periphrasis in Bulgarian. Finally, there is an outline of the chapters which make up the volume.

The Nakh-Daghestanian language Archi has several types of verbal constructions: periphrases, complex predicates, and phenomena very similar to serial verb constructions. This chapter investigates these constructions, using the approach of canonical typology; this allows different constructions to be ranked in terms of their proximity to the canonical centre. The analysis suggested is relevant for the general typology of multiword constructions, since it identifies tests for distinguishing them: for complex predicates the test will be their syntactic behaviour, for constructions close to serialization it is the fact that they are only available for a subset of verbs, while periphrasis is exhaustive. The chapter also has a descriptive purpose: published research on Archi does not describe all the available meanings for the periphrastic constructions nor their syntactic behaviour, and so an attempt is made to fill these gaps.

The Nakh–Daghestanian language Archi is famous for its extensive phonological inventory and its very large paradigms, especially those for verbs. Against this background, this chapter shows that there is particular complexity in Archi’s relatively small paradigm of gender and number agreement. This system is complex on many levels: the realization of agreement involves all parts of speech apart from nouns, but in some instances it requires lexical specification, since not all lexemes behave alike (many do not agree). Specifying the position of the gender and number markers, as prefixes, suffixes, or infixes, is also not straightforward: there are items with infixes where prefixal marking would be felicitous on phonological grounds. That is, Archi exhibits the typologically rare phenomenon of ‘frivolous’ infixation. The chapter establishes the factors which determine the realization of agreement in Archi, focusing on the verbal system, and a set of interrelating factors (phonological, morphological, and part of speech) which predict the outcome in some 70 per cent of the instances; our analysis suggests that this apparently small system is genuinely\break complex.

Taking a page from the RoN agenda which seeks to attribute fundamentally identical syntax to nouns and verbs, Ackema and Neeleman, in their chapter ‘Unifying nominal and verbal syntax: Agreement and feature realization’ pursue the parallel syntax of nominal and verbal projections by considering agreement phenomena. Some apparent agreement phenomena within the NP behave differently in some respects from verbal agreement, an observation that has led to a view that sees it as a distinct phenomenon, labeled concord. The authors defend two claims. First, concord is not itself an instance of agreement. Rather, following Norris (2014), it consists of the spell-out of features of an XP on terminals contained in that XP. These features can be present on XP because they are inherited from one or more heads contained in XP. These heads may have these features because they partake in agreement, or because they are inherent to the head. Second, neither agreement nor concord is unique to the category of the phrase in which it is found. Following the agenda set in Remarks, the authors argue that both agreement and concord occur in nominal as well as verbal domains. They show that various instances of apparently unusual agreement in TP, such as agreement in which adverbs are targets, are better analysed as cases of concord, and conclude that the general syntax of agreement and concord does not need to refer to nominal or verbal status. <236>

This chapter presents novel data from the Nakh-Dagestanian language Archi illustrating a typologically unusual phenomenon of apparent agreement between first person pronouns and absolutive-marked arguments. Apart from their typological significance, these facts challenge current approaches to agreement, which hold that Agree relations can be established only between heads and phrases. The chapter shows that Archi agreeing pronouns do not constitute a uniform class, but subdivide into simple weak pronouns and complex forms composed of a pronoun and a focus marker. Weak pronouns lack [CL] feature specification ([øCL]), and must therefore copy a class feature from the closest v to avoid violating the constraint that all DPs must be specified for [CL]. As a result, the apparent agreement between arguments can be reduced to the unsurprising agreement between the absolutive DP and a series of verbal heads, some of them morphologically null.

Agreement is a typologically common syntactic phenomenon that should be at the core of the design of every model of syntax. Even the straightforward examples of agreement are puzzling for syntacticians, because agreement involves both redundancy and arbitrariness. The indirect relationship between semantics and sentence structure expressed by agreement is thus a significant source of syntactic complexity, exacerbated by great diversity of its morphological expression. While syntactic theories all attempt to account for the role of syntax in grammar, there are inevitable differences in the principles and theoretical mechanisms underlying each model. To assess and compare the operability of syntactic theories, an independent evaluation tool is essential. Our chosen source of data, Archi, presents a rare case of a language whose agreement system challenges major claims found in three mainstream syntactic theories: Head-driven Phrase Structure Grammar (HPSG), Lexical Functional Grammar (LFG), and Minimalism.

Archi is a Lezgic language spoken by about 1,200 people in the highlands of Daghestan. Even a brief overview of Archi phonology, morphology, and syntax reveals that the language exhibits a rich phonological inventory, a diverse range of morphological exponents (both synthetic and periphrastic), and various paradigm shapes. In particular, agreement can target almost any part of speech, and there are a large number of agreeing non-finite verb forms that head complements and other dependent clauses. Archi syntax exhibits some degree of flexibility in the ordering of constituents, with a default head-final tendency. While morphological ergativity is evident throughout transitive clauses, there is also limited evidence that both absolutive and ergative arguments can function as a syntactically privileged argument in certain control structures.

There are two distinct syntactic agreement domains in Archi, the noun phrase and the clause, each presenting different challenges for an account of the mechanisms underlying their behaviour. Within the noun phrase, modifiers can exhibit the potential for agreement with multiple controllers in different domains. Agreement controllers can also exhibit duplicitous behaviour, with different values of the same feature simultaneously being relevant in two different agreement domains. Within the clause, agreement appears straightforward at first glance, since all possible targets agree with the absolutive argument of their immediate clause. Less simple is the issue of defining a possible target; this depends on the lexical category of the target, its morphological type or sometimes the properties of individual cells in its morphological paradigm.

Archi presents a number of challenging problems for syntactic theory as a result of the (semi-)predictable non-uniformity within the agreement system. An adequate model must account for the behaviour of atypical controllers of agreement, such as coordinated noun phrases and other controllers capable of triggering semantic (rather than syntactic) agreement. Additional challenges are introduced by biabsolutive constructions in which there is more than one possible absolutive candidate as the controller for agreement. In such cases different agreement targets within an apparently monoclausal structure may be controlled by absolutive arguments with different grammatical functions. Variability is also encountered by virtue of differences in the agreement potential of items within a lexical class. This poses a particular problem for theories of syntax which assume that the agreement potential of lexical items is determined by their position in syntactic structure rather than lexically or morphologically determined properties of the word form.

In Archi, verbs and other clausal constituents agree with an absolutive argument. Attributive adjectives, demonstratives, and possessive pronouns agree with the head noun in the nominal domain. In Head-driven Phrase Structure Grammar (HPSG) the former can be analysed as INDEX agreement and the latter as CONCORD agreement. However, data from Archi pose a variety of questions about how exactly they should be analysed. While any agreeing element must agree with an absolutive argument in the clausal domain, neither a constraint on ARG–ST lists nor a constraint on the features SUBJ (SUBJECT) and COMPS (COMPLEMENTS) can provide a satisfactory account of this agreement. This suggests that a constraint on syntactic structures is required. Data from Archi demonstrate that unexpressed absolutives must be represented at the relevant level of structure, contrary to much work in HPSG.

In Lexical Functional Grammar (LFG), syntactic agreement is primarily modelled as sharing or co-specification of agreement features at the level of f(unctional)-structure, which represents abstract grammatical relations and their properties, rather than at the level of constituency or phrase structure. LFG’s projection architecture also accommodates a multi-dimensional approach. While many agreement relations refer solely to f-structure notions, others may also refer to or depend on further constraints at other levels of representation. Since evidence suggests that Archi is morphologically ergative, agreement generalizations can be modelled by means of parameterized templates called by lexical entries. This lexical treatment allows agreement across the lexicon to be specified in a succinct manner while allowing for lexical idiosyncrasy.

In Minimalist syntax, clause-level agreement is an asymmetrical connection between two elements linked via a single syntactic process, called Agree. All case and agreement licensing in Archi happens at the level of the verb phrase (vP), not the tensed clause, and agreement serves as evidence of underlying structure, providing surface indications of the functional projections inside the clause. Agreement within the noun phrase in Archi is subject to the principles of concord and cannot be reduced to Agree. While Archi data are best accounted for by separating Agree and the mechanisms of nominal concord, this result does not mean that the unifying approach to verbal and nominal concord should be abandoned. This approach may be applicable in some languages, but not in Archi, where strong language-particular evidence points to the differential approach to agreement in the two domains.

The set of agreement problems presented by Archi test the adequacy of available infrastructure in theoretical models of agreement and reveal the basic similarities that underpin different approaches. This three-way comparison of Head-driven Phrase Structure Grammar (HPSG), Lexical Functional Grammar (LFG), and Minimalism exposes the strengths and weaknesses of each theory by comparing their accounts of specific empirical problems. To this end, the behaviour of numeral phrases provides good insight into NP/DP agreement and canonical clausal agreement, while the accounts of biabsolutive constructions tackle a more unusual syntactic problem. Examining the generality of the theoretical apparatus (in terms of how much new infrastructure has to be introduced to deal with the problem sets) and predictive ability (namely, the extent to which the relative restrictiveness of the approaches is beneficial or inadequate in dealing with ‘highly complex’ systems of agreement) helps us understand fundamental differences between each of the frameworks.