ECO-5

Program

Location: Boylston Hall, room 103

10:00 - 10:30  Breakfast (provided)

10:30 - 11:00 Rotsuprit Saengthong (MIT)

Clause size reduction by projection feature

11:00 - 11:30 Nofar Rimon (Harvard)

Understanding Kazakh men: Comitatives revisited

11:30 - 11:45  Break

11:45 - 12:15 Jane Loney (Harvard)

The Typology of Voice in Nguni

12:15 - 12:45 Gabriel Correa (UMass) 

Latvian ditransitives: Evidence for a low applicative structure

12:45 - 1:45  Lunch (provided)

1:45 - 2:15 Hrefna Svavarsdóttir (Harvard)

The Old Icelandic what partitive construction

2:15 - 2:45 Ioannis Katochoritis (MIT)

Long-distance agreement is neither long nor from a distance: a revival of spec-head under (c)overt movement

2:45 - 3:00  Break

3:00 - 3:30 Kevin Morand (UMass)

Again as a probe into the syntax-semantics of the left periphery

3:30 - 4:00 James Morley (MIT)

Unrankedness in person animacy restrictions

4:00 - 4:15  Break

4:15 - 4:45 Thomas Truong (MIT)

Voice mismatches and ACD

5:00  Dinner and drinks near campus

Abstracts

Clause Size Reduction by Projection Feature

Rotsuprit Saengthong (MIT)

I present cases of a mismatch between overt functional morphology (e.g., C and T) and the syntactic size of non-finite embedded clauses. Diagnosed with the usual restructuring diagnostics, the presence of these morphemes does not always correspond to the expected clausal size: in some environments they yield larger structures, while in others they fail to do so. For instance, clauses containing overt C and T may nevertheless lack the prototypical properties of CPs and instead behave like smaller structures such as TPs or even vPs. I propose that such apparent clause size reduction arises when higher functional heads fail to project. Building on feature-driven Merge, I argue that non-projecting elements enter the derivation with projection features similar to those of adjuncts, whereas projecting elements bear features like those of complement-selecting heads. This proposal further suggests that labeling is determined in narrow syntax, and that selection can possibly be strictly local, without appealing to the notion of Extended Projection.

Understanding Kazakh men: Comitatives revisited

Nofar Rimon (Harvard)

Comitatives are ‘with’ phrases that include an animate individual and are typically analyzed as adding a participant to the event (e.g. Alex danced with Sam). A longstanding debate concerns whether comitatives are DP-conjuncts, DP-adjuncts or VP-adjuncts, focusing mainly on Russian data. I present novel data from Kazakh involving the comitative marker men and revisit data from Russian and English, showing that all three analyses are necessary. I further show that comitatives may have 4 readings: collective, proximate-distributive, accomplice and distal-distributive. Only DP-conjuncts allow distal-distributive readings and only VP-adjuncts allow accomplice readings. These structure-to-reading mismatches, I argue, pose a challenge to a uniform semantics.

The Typology of Voice in Nguni

Jane Loney (Harvard)

In Zimbabwean Ndebele, subjects of monoclausal sentences can surface in-situ in Spec,vP or in Spec,TP. Similarly, subjects of hyperraising configurations can surface in embedded or matrix Spec,vP or Spec,TP (Pietraszko 2025). The choice of which position to surface in bears on other syntactic processes: object agreement and dislocation are only possible when the subject has moved out of its in-situ position. Pietraszko (2023, 2025) attributes all of these facts to the properties of, and probes on, Voice. In this talk, I present novel data from Johannesburg Zulu, a closely related language, showing that, while it allows for the same subject surface positions in monoclausal sentences, in hyperraising sentences, subjects are only allowed to surface in embedded Spec,vP and Spec,TP, and matrix Spec,TP. I attribute the lack of a matrix Spec,vP intermediate position to a deficient Voice head in raising configurations, and show that this deficiency is also realized in the barring of object agreement and dislocation in these sentences. Finally, I discuss a similar paradigm in Durban Zulu (Halpert 2015, Halpert & Zeller 2015), arguing that it displays properties of a semi-deficient raising Voice.

Latvian ditransitives: Evidence for a low applicative structure

Gabriel Correa (UMass)

Canonical ditransitive verbs in Latvian (Baltic) do not display dative alternation: the Goal is always realized as a dative DP, and the Theme is always realized as an accusative DP. Latvian, however, has flexible word order and allows either object to linearly precede the other. In this talk, I will present evidence from binding and scope interactions to suggest that both linear orders are derived via movement from an underlying low applicative structure (in the sense of Pylkkänen 2002). Furthermore, I will argue that Dative case assigned to Goals is inert, rendering them invisible for further A-operations (McGinnis 1998), echoing previous results from closely-related Lithuanian (Sigurðsson et al. 2018).

The Old Icelandic what partitive construction

Hrefna Svavarsdóttir (Harvard)

The what partitive construction (WPC) refers to a construction in which a wh-element stands in a neuter singular form with an oblique complement, usually genitive, questioning either kind or quantity. Although this construction has previously been discussed in Germanic for Old English (van Gelderen 2021) and historical German (Blümel & Coniglio 2019; Hobich 2017), North-Germanic has been largely overlooked. In this talk, I present data on the WPC in Old Icelandic, showing that the construction exhibits two primary semantic interpretations: a true partitive meaning, which questions a part of a whole, and a qualitative meaning, which questions the quality or nature of the complement. Unlike the previously studied languages, Old Icelandic WPCs are not restricted to the genitive case: depending on their semantics, they also occur with the dative. In addition to proposing an analysis of these Old Icelandic WPC types, I discuss their diachronic development into Modern Icelandic, ultimately contextualizing them with the Modern Icelandic wh-determiner hvaða 'which'.

Long-distance agreement is neither long nor from a distance: a revival of spec-head under (c)overt movement

Ioannis Katochoritis (MIT)

Long-Distance Agreement (LDA), whereby an embedding verb may φ-agree with  an argument of its complement clause, has been taken to support Agree, which dissociates agreement from movement. I object to this premise by presenting a unified analysis of LDA as local spec-head agreement via (possibly covert) movement. Crucially, I show that LDA, which alternates with default agreement, is not truly optional, but is rather a reflex of independent syntactic processes that feed LF or further syntactic operations, and contrasts with uninterpretable post-syntactic agreement. I also investigate the typology of LDA systems, accounting for the observation that many of them are ergative and verb-final. The explanation I provide makes predictions on the expected typological availability of LDA, irrespective of alignment, which are borne out by covert-(hyper)raising languages. The final typology comprises four LDA systems based on two parameters: the size of the complement clause and the feature inventory on the agreeing verb.

Again as a probe into the syntax-semantics of the left periphery

Kevin Morand (UMass)

This talk will report on-going work that aims at developing a novel diagnostic for the fine structure of the left periphery of clauses by exploiting a seldom-discussed utterance-modifying use of the adverb again. We will build on two well-established, albeit so far dissociated, lines of inquiry: (i) the structural analysis of the repetitive/restitutive ambiguity with propositional-level again as a probe into both the lexical decomposition of accomplishment predicates à la Dowty (1979); Stechow (1995) and the argument structure of ditransitives à la Beck and Johnson (2004), and (ii) independent evidence that clause type is encoded in a superordinate speech act layer above CP level that articulates speaker and hearer roles [cf. Ross (1970); Speas & Tenny (2003) among many others]. It will be argued that a parallel repetitive/restitutive ambiguity arises when again is interpreted at the level of speech acts (e.g. in 'Again, that's not correct'). We leverage this ambiguity to articulate a syntax-semantics account for the speech act domain analogous to Beck and Johnson (2004)'s analysis of ditransitives. The resulting account is argued to support a non-transformational view of the relationship between declarative and interrogative clause types.

Unrankedness in person-animacy restrictions

James Morley (MIT)

This talk investigates an understudied type of person-animacy restriction: person-animacy restrictions in which there is (at least) one person/animacy category which is unranked with respect to all others. Although it is occasionally noted in the literature, usually with respect to a single language, I show that unrankedness is attested in a range of unrelated languages, involving different (sets of) unranked person categories. Crucially, however, I show that unrankedness is incompatible with a set of basic assumptions shared by almost all contemporary theories of PARs. To resolve this puzzle, I sketch an amended theory of PARs which is capable of capturing unrankedness, without significantly expanding the theory's generative power. This theory is based on Hammerly's (2020, 2024) set-based theory of Agree, combined with the novel suggestion that probes should be able to refer to complement sets under a narrow set of circumstances. I briefly contrast this with an alternative machinery - Deal & Royer's (2025) dynamic interaction - which I show is also capable of capturing unrankedness, but at the cost of significant overgeneration.

Voice mismatches and ACD

Thomas Truong (MIT)

I examine a puzzle involving the following sentence which involves a voice mismatch ellipsis and ACD:

(1) *The professor must teach the problems that should be.

This sentence is judged to be unacceptable. However, voice mismatch VP ellipsis (Merchant 2013) and ACD on their own are both acceptable forms of ellipsis. The challenge is to formulate an identity condition for ellipsis that will predict the licensing of voice mismatch VP ellipsis and ACD on their own but predict that ellipsis is not licensed in (1). I will show that this is possible with a semantic identity condition for ellipsis with some extensions.