Programme

Organisation: Tom Leu et Heather Newell


Programme (PDF)


Jeudi/Thursday


Vendredi/Friday (info Zoom) à la Salle des boiseries (J-2805)

Distinguishing zero from zero: the morphosyntax & exponence of PRO vs. pro

(handout)

A cross-linguistic look at lexical availability and mood choice

(slideshow)

The Source of Words

(handout)

Tlingit verb morphology is syntax

(slideshow)


Samedi/Saturday (info Zoom)  à la Salle des boiseries (J-2805)

Untangling nominalization, number, noun class, and agreement properties in Chichewa (Bantu)

(handout)

Bantu Noun Class as Gender: exploring a little n approach

(handout)

MaS versus Metasyncretism: The Case of Latin Noun Declension

(slideshow)

(slideshow)

(à venir)

Résumés / Abstracts


A cross-linguistic look at lexical availability and mood choice 

Léna Baunaz, Université Côte-d’Azur, et Eric Lander, Uppsala Universitet

The data. Mood is morphologically realized as suffixes on verbs in Romance and Hungarian, but on the complementizer in Balkan (as a free morpheme) and West Slavic languages (as a suffix). The difference between indicative and subjunctive mood marking is indeed morphologically encoded on verbs in Romance and Hungarian (showing different verb forms) but on the complementizer in Balkan and West Slavic (see Sočanać 2017 for Slavic languages and Giannakidou & Mari 2021 for Modern Greek). The connection between this special complementizer and subjunctive mood, as well as the crosslinguistic distribution of mood marking as ‘verbal’ vs ‘clausal’, has been extensively debated (see Giannakidou 1998, 2009; Krapova 1998; Roussou 2000, 2009, 2010; Sočanać 2017; Giannakidou & Mari 2021; a.o). 

These langugages differ, however, on what type of predicates triggers the subjunctive mood: in Romance, subjunctive mood on the embedded verb is triggered by a subset of non-factive predicates (e.g. ‘want’), and a subset of factive predicates (= emotive factives, e.g. ‘regret, be happy’), in the matrix clause. In Balkan only non-factive verbs like ‘want’ can select for the subjunctive complementizer, whereas emotive factives like ‘be sad’ require a specific (non- subjunctive) comp instead (Giannakidou 2009, 2015; Sočanać 2017 a.o). Hungarian, like French, has verbal mood (see Baunaz & Puskás 2022). Subjunctive mood is realized on the embedded verb under a volitional matrix predicate. As far as the classes of matrix predicates triggering subjunctive, however, Hungarian is like Balkan in that factive verbs like ‘regret’ and ‘be happy’ do not trigger subjunctive in the embedded clause. Finally, Polish is like Balkan in showing subjunctive marking on the complementizer, yet in the form of a suffix -by on the complementizer że (and not as an entirely separate form as in MG na vs. oti or pu). It is also like Balkan and Hungarian in that verbs like ‘regret’ and ‘be happy’ do not trigger embedded subjunctive. How can we make sense of these data? 

The four patterns represented by French, Greek, Hungarian, and Polish will be our main focus in this talk. 

Framework. Our talk is couched within the nanosyntactic approach to grammar (Starke 2009, 2011; Caha 2009) and develops a previous work of ours (Baunaz and Lander to appear). Nanosyntax is an approach where morphemes are internally complex and composed of syntactico-semantic features which are hierarchically ordered according to an invariable functional sequence (fseq). When the syntactic component generates a structure, it proceeds step by step, merging one feature after another according to the functional sequence. After every step of merge, the syntactic structure must be properly matched to a lexical structure in the lexicon. Crosslinguistic variation in nansoyntax is understood essentially in terms of how a single universal functional sequence is lexicalized in different ways depending on the idiosyncrasies of the lexicon. While we may all have the same syntax and the same derivational options available to us, we do not have the lexical entries available for lexicalizing these options. 

Claim. We apply a comparative approach to the problem of subjunctive mood selection by focusing on the patterns in the four languages described above. Mood selection is a complex issue, partly due to the high degree of crosslinguistic variation observed, but we hope to show that this kind of perspective is crucial to understanding the universal mechanisms underlying the selection of subjunctive mood in the embedded clause. The variation observed can, we argue, be derived from relatively minor language-specific idiosyncrasies of the lexical entries available in each language. 



Bantu noun class as gender: exploring a little n approach.

Vicki Carstens, University of Connecticut, Storrs

There are many strands of semantic coherence in the noun classes of Bantu languages: a cluster of human-denoting nouns is found in one singular/plural pair of classes, animal names in another, language names in a third; some Bantu languages have dedicated classes for diminutives and for locatives, and so forth, though arbitrary class assignments predominate. This talk will argue on the basis of agreement with conjoined DPs that the noun class system has exactly two interpretable flavors: [+/-human]. The phenomena provide diagnostics for interpretability and for assessing several recent approaches to variability in agreement with conjuncts.


Tlingit verb morphology is syntax

James Crippen, McGill University

The Tlingit language, like other Na-Dene (= Dene-Eyak-Tlingit) languages such as Navajo, is famed for its complex verbal morphology. I argue based on an extremely broad range of empirical evidence (essentially the entire documented lexicon) that the verb word in Tlingit is actually a complex syntactic structure minimally encompassing AspP. Each morpheme is a separate syntactic terminal instantiating things like argument structure, aspect, mood, and modality, all built upon an acategorial root. Ordering of terminals is a conseqeuence of straightforward syntactic structures formed by Merge, conditioned by selection, agreement, and phonological constraints. The syntax in Tlingit verbs requires no extraordinary theoretical mechanisms beyond vanilla Minimalism. Supposedly discontinuous morphology does not exist: long distance dependencies between syntactic terminals are instances of phase-based movement and agreement together with compositional semantics. The appearance of a verb as a single phonological unit is a consequence of a regular mapping of syntactic structure to domains in a recursive phonological word, conditioned by a system of metrical and segmental constraints. Nouns and verbs appear to have parallel syntactic structures and the phonological systems of each also seem to be parallel, suggesting a unified organization for all referential expressions in the language.


Untangling nominalisation, number, noun class, and agreement properties in Chichewa (Bantu) 

Peter Msaka, University of Malawi and Stellenbosch University 

This presentation explores the complexities of nominalisation, number, noun class, and agreement in Chichewa and other Bantu languages. Putative analyses of Bantu morphosyntax, particularly concerning noun classification, hypothesises that the Bantu noun class system is predominately dependent on so-called noun class prefixes, which also serve as nominalisation and number affixes. Recent empirical evidence, however, contradicts this notion, demonstrating that the traditional prefix-based method does not account for the full spectrum of available data. To address these data-related challenges comprehensively, this study assembles a wide-ranging dataset from diverse sources, including the Chichewa monolingual electronic dictionary, the Chichewa general corpus, fieldwork data, and the researcher's introspection. The properties of nominalisation, number, noun class, and agreement are independently investigated before exploring their relationships and potential dependencies. The findings demonstrate that each of these nominal grammatical aspects exhibits a diverse array of characteristics, necessitating a paradigm shift in how Bantu nominalisation, number, noun class, and agreement marking are understood. Notably, Chichewa exhibits a more extensive range of nominalisation affixes than previously considered. Furthermore, contrary to the prevailing assumption, the marking of singular number in Chichewa is not as pervasive as commonly believed, as it applies to only a limited subset of nouns. In fact, it is noteworthy that in a particular dialect of Chichewa, the use of singular prefixes is absent. The commonly assumed fixed relationship between number and noun classes is applicable to only a limited subset of nouns. Taking a Distributed Morphology perspective, what emerges in Chichewa's nominal morphosyntax is that derivational affixes appear lower in the nominal spine. It is also within this lower domain that noun class features reside. Above this, we have number features and above the spine we have speaker-hearer perspectives in the form of augmentatives, diminutives, and honorific affixes. Agreement on words in construction with the noun is influenced by two crucial factors: semantic features and phonological features. Specifically, animate or agentive nouns constitute one class, while the remaining nouns are further categorised based on their word-initial phonological characteristics. These results shed new light on the complexities of Chichewa and Bantu languages' morphosyntax and provide valuable insights into the mentioned linguistic systems. 


MaS versus Metasyncretism: The Case of Latin Noun Declension

Neil Myler, Boston University

I have developed a fragment of a grammar for a substantial (albeit sanitized) portion of the Latin Noun Declension system, hewing closely to the Morphology-as-Syntax (MaS) approach described in Collins and Kayne (2023).  The fragment makes crucial use of proposals regarding irregular plurals in Collins (2018), and of the approach to case contiguity in Collins (2020).  My talk will mostly be an overview of this fragment, with special emphasis on what it teaches us about MaS approaches to metasyncretism.  I will also discuss what the fragment reveals about how MaS might approach such important morphological issues as secondary exponence and class-sensitive allomorphy.   At the end I will engage in some editorializing, which will include pointing out at least the following: (i) MaS can eschew Late Insertion only if it embraces Late Filtering—there is no escape from Late-ness; (ii) MaS commits one to a non-uniform approach to metasyncretism,  setting up an interesting theoretical contest with realizational approaches to morphology, which tend to deal with metasyncretisms in the same way regardless of what corner of the morphology of a language they turn up in; (iii) the fragment as a whole strikes its creator, and might well strike its audience, as “bloated” as compared with imaginable realizational alternatives.  One should consider carefully what the sources of this bloat are, and what weight this bloat should carry in comparing MaS with rival frameworks.


Distinguishing zero from zero: The morphosyntax & exponence of PRO vs. pro

Sandhya Sundaresan, Stony Brook University

This talk concerns itself with the morphosyntax of the silent pro-forms labelled "PRO" and "pro". Though they cannot be distinguished morphologically, being both silent, it has been proposed (cf. Chomsky 1981, et seq.) that they denote distinct underlying elements, with PRO being an obligatorily bound variable that always and only occurs as a non-finite subject and with pro exponing a pronominal object or subject in a finite clause which is interpreted as a free variable. But the robustness of this bifurcation is challenged by the existence of languages which allow overt pronominal or R-expression subjects in non-finite clauses (cf. Sundaresan & McFadden 2009) and, furthermore, allow subject pro-drop. In such cases, we can no longer automatically assume that a silent non-finite subject is PRO; if anything, we expect it to be pro. In Sundaresan (2014), I showed that, contrary to expectation, the silent non-finite subject of a non-finite clause in such languages always behaves like PRO, not pro, wrt. syntactic and semantic diagnostics for bound-variable behavior. 


The goal of my talk is to explain this unexpected complementarity in the distribution of pro and PRO, in the absence of case- or finiteness-related motivations.  I argue that such complementarity is automatically explained if we assume that pro and PRO start out as the same element underlyingly -- a single underspecified pro-form that I label "UPro". The interpretive and distributional distinctions lurking behind the labels "pro" and "PRO" will be seen to derive from how UPro interacts with its structural environment and from language-specific rules of morpho-phonological realization. Specifically, OC PRO labels a rather specific interpretation that arises in embedding contexts where a syntactic OC relationship with an antecedent can be established. Different types of pro and NOC PRO, on the other hand, involve ‘control’ by (typically) silent representations of discourse-contextual elements in the clausal left periphery. 


The Source of Words

Peter Svenonius, University of Tromsø – The Arctic University of Norway

Morphology is generally understood as concerning the structure of words, while syntax concerns the arrangement of words into phrases and sentences. Startlingly, there is no consensus in linguistics concerning what words are. It has been suggested several times that linguistic theory can make do without a notion of word, but to date no satisfying alternative has been provided. The alternative “output of a word-formation component” (WFC) is problematic, for reasons explicated at length in DM work, and the alternative “maximal X0” is problematic, for reasons detailed by various critiques of head movement (HM). I develop a theory of words which requires neither a WFC nor HM. It makes use of span-based spell-out and linearization. A syntactic word is a syntactic object containing a single @-node, a node with the property @ (pronounced “at”), which it acquires as part of the labeling operation which accompanies Merge. Nodes without @ will spell out as parts of a syntactic word including an @-node. Only syntactic objects containing an @-node can undergo syntactic movement (internal Merge), so the inseparability of words is derived.