Against some approaches to long-distance agreement without AGREE

With the introduction of AGREE into Minimalism by Chomsky (2000), the relationship between the two elements in an agreement relationship went from being strictly local (Specifier-Head) to being unbounded (c-command with no intervening strong phase boundary) in order to accommodate long-distance agreement phenomena. Concern over the less restricted nature of the new approach led researchers to propose alternatives that eschewed the unbounded reach of AGREE , in the hope that a more restrictive theory might yet be salvaged. This paper scrutinizes some of the most widely cited and fully developed of these alternative proposals (employing predicate inversion of expletives, restructuring, covert movement), applied to extensively studied spheres of data (English existentials, Icelandic agreement), and concludes that they are deeply, perhaps fatally, flawed. While Chomsky’s version of AGREE is far from providing a complete and satisfactory theory of agreement, it has yet to be shown that it can be eliminated.

Contrast in syntax and contrast in phonology

This chapter investigates the parallels and asymmetries between contrastive features in phonology and syntax, with particular reference to the notion of a contrastive hierarchy. In phonology, contrastive features can be assigned to segments by a recursive procedure that applies to the underlying phonemic inventory of a language. To apply such a procedure to morphosyntactic features, it is necessary to identify the inventory of items that the features serve to distinguish. This chapter argues that the relevant morphosyntactic inventory is the inventory of functional lexical items (in the sense used in Distributed Morphology), and not the inventory of vocabulary items. It further proposes that contrastive specification of functional lexical items is done separately on different dimensions of contrast—for example, that person features are specified separately from number features. This approach is illustrated by application to phi-features in Mi’gmaq, where it is shown to be consistent with patterns of agreement.

The Tłı̨chǫ syntactic causative and non-nominal CPs

Study of the periphrastic causative in Tłı̨chǫ has its origins in community-based research supporting a literacy manual and dictionary database. It is shown that the causative verb ats’ele ‘cause, let, do to’ takes two complements: the causee and a second expression of varying complexity that indicates the caused process or result. At its most complex, the second complement can be a clause which bears an adverbializing suffix and expresses the caused situation. Ats’ele selects a TypeP, which is independent of the higher verb in aspect, tense, and polarity. This leads to a broader range of interpretations of the relationship between causing and caused situations than is found in languages where the caused situation is expressed as a verb phrase. The complementizer used in the causative construction contrasts with a nominalizer in the language, in that it forms an adverbial clause. The facts thus provide a novel instance of non-nominal complementation.

Definite expression and degrees of definiteness

This chapter studies definite determiners formerly treated as semantic expletives and challenges the view that they can be uniformly treated as such. Assuming that definiteness consists of two features, uniqueness (iota) and familiarity (Fam), and depending on the features spelled out by the determiner, it proposes that definite articles can be fully specified for definiteness spelling out both features (full definiteness), partially specified, spelling out Fam (partial definiteness), or, in the case of true expletives, not specified at all (zero definiteness). Fully definite expressions cannot be modified by other definite nominals. In contrast, partially definite expressions form predicative FamPs, which can be modified by other definite nominals. Fam can also introduce proper names and generic kind-denoting nouns. Finally, true expletives appear even in non-definite contexts. An explanatory and descriptive account is offered that provides new insights on the properties of definiteness.

Sentience-based event structure

In Blackfoot, a Plains Algonquian language spoken in Alberta, Canada, and Montana, USA, sentience, rather than telicity, is a primary determinant of argument structure. Subjects of transitive verbs, non-core objects of transitive verbs (benefactives, malefactives, sources, etc.), and primary objects of ditransitive verbs are all subject to a strict sentience requirement. This chapter follows Ritter and Wiltschko (2015) in assuming that the strict sentience requirements on argument structure are part of the grammar (i.e. part of the “narrow syntax”) of Blackfoot, and formalizes sentience as a feature that is subject to selection, a feature-checking operation, much like AGREE. This proposal correctly predicts that (a) not only agents but also causers must be sentient in Blackfoot; (b) sentient objects (not bounded ones) serve as both initiators and delimiters of events; (c) like event types, nominal types are distinguished by sentience, rather than boundedness; and (d) eventiveness is correlated with sentience, rather than dynamicity.

Restricted and reversed aspectual contrasts

In this chapter, Scottish Gaelic data are presented as evidence for the necessity of two modifications to Cowper’s (1998 and forward) feature geometry for tense and aspectual contrasts. Both modifications involve dependencies on the PRECEDENCE feature, which in Cowper’s geometry of English is responsible for the past tense marker -ed and the past participle morpheme -en/-ed. The first modification is a RESTRICTED feature. This feature is used to encode the temporal restriction between event and reference times in the near perfect and near prospective aspects found in Scottish Gaelic. The second modification is a REVERSED feature, which reverses the temporal ordering of two times. This feature is argued to mark unrestricted and restricted prospective aspects. It is claimed that certain future-like meanings in Scottish Gaelic do not involve a modal feature like IRREALIS but rather the proposed REVERSED feature.

A feature-geometric approach to verbal inflection in Onondaga

Iroquoian inflectional verbal morphology is well-documented in the descriptive literature, but has received less attention from a generative perspective, a fact this chapter sets to remedy. Most generative analyses rely on the notion of tense as a central category and assume a universal projection of a T(ense) Phrase in tensed/finite clauses. Onondaga (Northern Iroquoian), however, typically makes very little use of tense as a grammatical concept, despite the fact that it only allows for finite clauses. Instead, this language capitalizes on the notions of aspect and mood, thereby rendering the standard generative approach inappropriate. Consequently, this chapter argues that a feature geometric analysis, which does not rely on tense as a central concept, is better suited for analysing the Onondaga verbal inflectional domain where it is aspect that serves as the crucial ingredient in activating Infl.

Cross-linguistic contrasts in the structure of causatives in clausal nominalizations

It is usually assumed that the causee ‘subject’ of a causativized transitive predicate merges below the causativizer, within the causativized clause. However, gerundive nominalizations (masdars) in Georgian provide evidence for a different structure. It is proposed that, in Georgian, this causee is projected by an applicative phrase that merges outside the causative vP, below the Voice projection that introduces the causer/external argument. Key evidence involves masdars in which causative meaning can be expressed, but a causee cannot. While the causee of a causativized transitive predicate can be expressed in a verbal context, it cannot be expressed in a masdar, even though the masdar can be based on a causative of a transitive. It is proposed that the Georgian masdar involves a nominalizing head that selects a vP complement, and that both causers and causees are excluded from the masdar because they can only merge outside this causative vP.

Contrast and representations in syntax

This introductory chapter gives an overview of how views of both contrast and features developed in linguistics, and how they are situated in contemporary Minimalist syntactic theory. It observes that though features have come to play a crucial role in Minimalism as the drivers of syntactic operations and the locus of parametric variation, less attention has been paid to the connection between this and their role in definine paradigmatic contrast. Against this backdrop, the chapter relates each of the subsequent chapters to the overall themes of the volume, discussing the different ways in which they address the theme of contrast and representations.