October 9-10, Tromsø
Keynote speakers with talk titles
- Anna Papafragou, Delaware: Semantics and Cognition: The case of evidentiality
- Theresa Biberauer, Cambridge and Stellenbosch: Formal features in systems that maximise minimal means
- Daniel Harbour, Queen Mary London: Feature scaffolding via CP-DP parallelism
- Hagit Borer, Queen Mary London: Functions and Categories
AbstractI will explore the architecture of grammatical functions, as it emerges from a set of largely shared assumptions and theoretical desiderata, including the difference between major categorial classes and functional structure, the existence of a universal hierarchy for functional structure, the need for a major categorial core within Extended Projections, and the desire to eliminate from the lexicon any reference to syntactic properties, including those which may control the formation of complex words.
Within the functional domain, the picture emerging will bank on formalizing the difference between labelled syntactic nodes, and non-categorial terminals which may value them, semantically. Labelled syntactic nodes, in this account, would be populated either by empty heads (D,≪e≫) or by pure categorizers (-ation, -ize). Empty heads, in turn, would be valued, and ultimately categorized, by semantic functors (THE, PST etc.), in conjunction with the universal functional hierarchy, and the need for a major categorial core within Extended Projections.
Additional talks will be given by Isabelle Roy (CNRS-Paris 8), Bridget Copley (CNRS-Paris 8), Cristina Real Puigdollers (UPF Barcelona), Natalia Mitrafanova (UiT), Peter Svenonius (UiT), Gillian Ramchand (UiT), and Sergey Minor (UiT).
The workshop addresses features in linguistic theory, specifically formal features which are syntactically relevant and semantically interpretable. A major question is to what extent such features are universally predetermined or not; if they are not universally predetermined, then what constrains their emergence?
In syntactic theory, formal features (FFs) are properties that distinguish heads from each other syntactically (adopting the definition proposed in Adger & Svenonius 2011). Some features trigger syntactic operations such as Merge or Agree; Adger & Svenonius call these triggering properties ‘second order’ properties of features. First-order features bear labels such as [past] and [plural] which can (usually) be given truth-conditional interpretations in a standard semantics.
First order features can be thought of as restrictions on second order features. For example, a head which is a target of agreement bears a ‘probe,’ and a probe is characterized by the second-order property of uninterpretability or unvaluedness. This property functions as an instruction to copy feature values. The probe is restricted to copy feature values belonging to a particular class of first order features, for example Phi-features (person, number, and gender).
The usual consensus is that second-order properties are drawn from a small, cross-linguistically invariant set (see, e.g., Rizzi forthcoming on the format for parameters, or Minimalist Grammar work such as Collins & Stabler 2010). Inventories of first order features in different languages vary somewhat, though the limits to the variation are more striking than its range, as observed by researchers working from such different perspectives as Corbett (2012) and Cinque (1999). Certain featural oppositions are attested in many unrelated languages, such as participant-nonparticipant, speaker-addressee, singular-plural, plural-dual, past-nonpast, future-nonfuture, perfect-nonperfect, perfective-imperfective, realis-irrealis, subjunctive-indicative, interrogative-declarative, and so on.
It has been an assumption of convenience that UG provides an inventory of FFs, with individual languages selecting a subset of the inventory (e.g., Cinque 1999, Emonds 2000, Chomsky 2000). This accounts by fiat for the limits of variation, but apart from that there is little support for it. The alternative is that there is no universal inventory of FFs, and that language-specific inventories of FFs emerge in constrained ways. The question then is, what are the constraints that lead to the emergence of the language-specific inventories of FFs that do emerge?
An important first step in answering this question is to identify the actual inventories of FFs. A bound morpheme cannot immediately be assumed to represent a formal feature (unless one makes a further assumption concerning the relation of affixes to syntax; for example, that affixes can only express information encoded in FFs; see further below). The existence of a diminutive suffix on nouns in a given language, or of an iterative suffix on verbs, for example, is not sufficient evidence for a syntactic feature [diminutive] or [iterative] in that language. According to the definition adopted here, a formal feature distinguishes heads syntactically. If a diminutive-marked noun, for example, behaves syntactically just like any other noun (or, perhaps, any other neuter count noun), then there is no motivation for a FF [diminutive]. If an iterative-marked verb behaves syntactically like any other verb (or like any dynamic verb, for example), then there is no evidence for a FF [iterative].
On the other hand, [finite] is clearly a syntactic feature, in a language like English, because finite clauses have a different syntactic distribution from nonfinite clauses. Similarly, [plural] is clearly visible to syntax, because it systematically affects agreement (a syntactic phenomenon) and the choice of determiners. The distribution of past participles is distinct from that of progressive participles, and the syntax of comparatives is distinct from that of superlatives, so those oppositions clearly involve FFs.
A ‘lexical’ model of morphology like DM (‘lexical’ in Stump’s 2001 sense, meaning that morphemes correspond to lexical entries, as opposed to word-formation rules), combined with a theory of functional heads (e.g., Abney 1986) organized in a system of Extended Projections (Grimshaw 2005), may require many or even most inflectional oppositions to correspond to FFs, because each affix implies the existence of some functional head, and each functional head selects the next one down in the extended projection. Taking the example of the iterative mentioned above, if iterative morphology cooccurs with tense and voice morphology, then that might be taken as evidence for some head between Voice and Tense, for example Aspect, formally distinct from both Voice and Tense.
Similarly, the distributional differences among demonstratives, numerals, adjectives, and nouns discussed by Cinque (2005) require FFs to distinguish them from each other. For example the domain for attachment of numerals is higher than the domain for the attachment of adjectives, and if the domains are defined by heads, then a feature distinguishes them.
In general, these FFs can be assigned semantic interpretations in some context or another. See Harbour 2007, for example, on the formal interpretation of gender features in Kiowa.
Given these definitions of features, and these background assumptions, we can now turn back to the question, what are the constraints that lead to the emergence of the specific sets of FFs that are found in each language? We propose to address this question in a two-day workshop to be held in Tromsø in the fall of 2017. Papers will be presented by a group of invited speakers,