• Abstracts

Pavel Caha (Masarykova Univerzita, Brno)

Lexical items and syntactic categories: Some lessons from Shona and Luganda locatives

In generative syntax, it is a virtually unquestioned assumption that a lexical item has a unique category, one out of a list of universal categories (N, V, D, P, ...). Sometimes, it has to be admitted that an item is ambiguous (for instance, "front" may either be a preposition or a noun). But it is quite rare to say that an expression is both a noun and a prepositon at the same time.

However, this unusual scenario is something that is predicted by theories with late phrasal spell-out. If phrasal spell-out exists, a single lexical item may be the pronunciation of several terminals, where each terminal has a distinct categorial label. As a consequence, the approach predicts the existence of expressions whose behavior  corresponds to a mixture of several categories. In the talk, I present
a joint work with Marina Pantcheva, where we look at the puzzling behavior of Bantu locative class markers. We show how the tradition struggles to assign them to one category or another (N or P), and we show that the struggle dissolves if they are analyzed as an item that has both of the categories (N and P) at the same time.

Berthold Crysmann (CNRS, Universite Paris Diderot)

Deconstructing exuberant exponence

Kleanthes Grohmann (University of Cyprus)

Localities: Building blocks and other pieces of the puzzle

By going over some developments in minimalist building blocks over the past 20 years, I will try to couch some issues concerning locality measurements in the larger context of the IGRA research theme: do certain operations (counter)bleed or -feed others, do particular assumptions stand in some kind of competition or cooperation relation to one another, and how can we tell? Particular emphasis will be put on the formulation of derivational valuation points for licensing local and not so local relations, and some of the proposals to this end that have been floating around.

Karen Jesney (University of Southern California)

Complex process interaction: Weighted constraints vs. ranked constraints

This talk argues that the cumulative constraint interaction of Harmonic Grammar (HG; Legendre, Miyata & Smolensky 1990, Smolensky & Legendre 2006) allows certain complex patterns of process interaction to be modeled using only a few basic constraints. This simplicity of analysis stands in contrast with rule-based frameworks and with Optimality Theory, both of which must make recourse to extrinsic ordering or more complex building blocks in order to account for the patterns in question. I illustrate these claims with a case study from Ponapean (Rehg & Sohl 1981) and consider the limits of HG's ability to simplify phonological analyses through cumulative constraint interaction.

Björn Köhnlein (Universiteit Leiden)

Rules vs. constraints in studies on tonal accent

This talk addresses a classical issue of sentence-level phonology: the association of intonational tones to phonological strings. In this context, so-called tone accent languages can be of particular interest, because in such languages the assignment of intonational tones can can affect the lexical meaning of words (similar to ‘real’ tone languages), as we find it in Swedish, Norwegian, or Franconian (spoken in Belgium, Germany, the Netherlands). Since, as I claim, the tonal contrasts derive from differences in the metrical organization of words, tone accent systems can help us a better understanding of how intonational tones associate with phonological units at the word level. The emerging question is, then, whether these interactions are better described in terms of rules (e.g. as association conventions for intonational tones) or whether they should rather be formulated in terms of output constraints (as In Optimality Theory). As we shall see, while some principles of association can be formulated in terms of rules or constraints, other generalizations can be expressed more easily with constraints.

Olaf Koeneman (Radboud Universiteit Nijmegen)

Do-support and the syntax of finiteness

In English, only a subset of verbs can appear in a position external to the verb phrase, indicated by their position to the left of a sentential negation. In the absence of any such verb, a finite form of do must be inserted if the clause contains a negation. Standard accounts of this paradigm (Chomsky 1957; Bobaljik 1995, etc.) make use of two assumptions that are in all likelihood wrong: (i) the assumption that modal verbs can be basegenerated in a verb phrase external position and (ii) the assumption that negation not is a syntactic head. We will propose an analysis of the data without these assumptions and argue that all verbs start out in the verb phrase. A verb can move if two prerequisites are met: (i) it does not assign an internal theta role and (ii) it is irregularly inflected. We propose that these prerequisites, which do not form a uniform class, highlight the role of little v in the verbal syntax of English.

Laura McPherson (Dartmouth College)

Phrasal phonology, restructured: The place of Dogon tonosyntax

Tonosyntax in the Dogon languages is a unique system of phrasal grammatical tone in which words have their tone completely overwritten by tonal overlays in particular syntactic contexts. The heavy involvement of syntactic information (both category and structure) indicates that this is no longer pure phrasal phonology, though this is its likely origin. However, the exact place in the grammar for this restructured system is not self-evident. In this talk, I present the empirical data on Dogon tonosyntax then sketch out three possible analyses, suggesting finally that tonosyntax is best understood under a constructional view of morphology/the lexicon.

Katya Pertsova (University of North Carolina, Chapel Hill)

Conflicting pressures can lead to ungrammaticality

In this talk I will focus on the problem of paradigm gaps in inflectional morphology, a phenomenon that presents a challenge for many models of grammar.  In particular, I will discuss new corpus and experimental evidence suggesting that the famous 1p.sg. gaps in Russian verbs are not synchronically arbitrary as assumed before, but are due to a conflict between the regular alternation rules and the pressure from Paradigm Uniformity not to alternate.  This conflict leads to gaps only in those cases in which the expected alternation does not independently exist in other morphologically related forms. I will discuss implications of this finding for the theories of morpho-phonology, and propose a tentative way of modeling these facts.

Michelle Sheehan (University of Cambridge)

Ergative alignment in Romance causatives: A parameter hierarchy approach

The Romance faire-infinitif (FI) poses interesting challenges for parametric theory. On the one hand FI shares many properties across those varieties in which it is attested (various varieties of Spanish, Italian, Portuguese, Catalan and French). On the other hand, several kinds of micro-parametric variation are attested concerning the optional/obligatory nature of object clitic climbing, past participle agreement, passivisation, the possibility of se clitics, the case of intransitive subjects etc.. In this talk, I propose that this micro-parametric variation can be modelled by a parameter hierarchy with the same shape as that which regulates clausal (ergative/accusative) alignment. The format of these hierarchies, it will be proposed, is emergent and determined by (i) general acquisition strategies and (ii) the need to create convergent derivations.

Francesc Torres-Tamarit (Vrije Universiteit Amsterdam)

Voicing-dependent vowel lengthening and final devoicing in Milanese: A transparent or an opaque interaction?

In Milanese stressed vowels preceding word-final devoiced obstruents lengthen, but remain short preceding word-final voiceless obstruents. This opaque interaction between vowel lengthening and final devoicing adds a further complication. Final devoicing in Milanese has been reported as being subject to free variation, for some authors, or being incomplete, for other authors. My aim in this talk is to argue that only a constraint-based serial analysis framed within Serial Variation (Kimper 2011) accounts for vowel lengthening as a process triggered by the need to comply with final devoicing. This analysis thus gives a unitary account of the two phenomena. The Serial Variation analysis also implies that final devoicing is a process subject to free variation. I also show that an alternative parallel analysis framed within Turbidity Theory (Goldrick 2001, van Oostendorp 2008), which is able to express representationally the difference between devoiced and voiceless obstruents, runs into problems in accounting for vowel lengthening as naturally following from final devoicing. Formal phonology is still an explicit theory of the sound patterns of language, but solid data on this dialect is missing, which makes difficult to evaluate the two different proposals.

Luis Vincente (Potsdam University)

A remnant-correlate identity condition on ellipsis

(joint work with Matt Barros, Rutgers)

Existing formulations of the identity condition on ellipsis take "identity" to be a relation that holds between the E(llipsis)-site and the A(ntecedent)-phrase (for  example, Sag 1976 requires that the E-site and the A-phrase have identical LFs; Merchant 2001 requires mutual entailment between the E-site and the A-phrase, modulo focus closure and existential-type shifting; AnderBois requires mutual entailment at the inquisitive level; etc). Call this family of conditions "E-site conditions": our claim is that E-site conditions alone are insufficient to account for the distribution of possible and impossible ellipses. Rather, they need to be supplemented with a morphosyntactic identity condition between the remnant of ellipsis and its correlate in the antecedent.

(1) Remnant-correlate identity condition: the remnant of ellipsis and its correlate must have the same categorial features.

We support this claim with two case studies of sluicing and fragment answers, the first one involving symmetric predicates ("make out with", "be related to"), and the second one involving cleft/copular antecedents (we present English data, but our argument can be replicated in other languages too). E-site conditions incorrectly predict these cases to be grammatical, but (1) correctly rules them out.

(2) Symmetric predicates.
    * Someone was making out with Jack, but I don't know with who.
    [cf. "...but I don't know who Jack was making out with"]

(3) Cleft/copular antecedents.
    * The guy that Sally talked to was someone from Accounting, but I don't know to who.
    [cf. "...but I don't know who (from Accounting) Sally talked to"]

In addition, we show that (1) can be also extended to account for the impossibility of voice and argument structure alternations under sluicing, as well as for the impossibility of P-stranding effects under sprouting, making morphosyntactic identity conditions on argument-introducing functional heads (Chung 2006, 2013; Merchant 2013) superfluous.

(4) No voice alternations under sluicing.
    * Someone assassinated JFK, but I don't know who by.
    [cf. "...but I don't know who JFB was assassinated by"]

(5) No argument structure alternation under sluicing.
    * Sally embroidered something with peace signs, but I don't know what on.
    [cf. "...but I don't know what Sally embroidered peace signs on"]

(6) No P-stranding under sprouting.
    * Sally is jealous, but I don't know who.
    [cf. "...but I don't know who Sally is jealous of"]

Geraldine Walther (Université Lyon)

The realisational lexicon

Realisational approaches to inflectional morphology are firmly embedded in a lexicalist approach to grammar. Yet existing realisational approaches mainly focus on the specification of a particular rule architecture, mapping morphosyntactic property sets to sets of word forms, but leaving the definition of the interface between inflectional paradigms and the nature of the lexicon largely, if not completely, inexplicit.

In particular, although inflectional, and more specifically paradigmatic, irregularity has been an important focus of recent realisational approaches, little work so far has specifically addressed the nature of a realisational lexicon and the representation of those irregularities within inflectional lexical entries.

We present a realisational model which aims to address this deficiency by providing an explicit definition of realisational lexical entries. This model focuses on the relationship between paradigmatic structures in a language and the properties of a language’s individual lexical entries. In particular, it highlights the origin of lexeme specific inflectional irregularities, and offers a precise, lexically integrated, formalised, and quantifiable definition of fundamental inflectional notions, such as inflectional classes, inflectional categories, morphosyntactic features, and inflectional regularity.

letzte Änderung: 08.07.2015