Research

• [publication] [agreement handout] [syntax-prosody handout]

• Concordant agreement, SVC structures (adjunction, compounding, and complementation), syntax-prosody mapping

• Amarasi, Äiwoo, Formosan [Austronesian]

The East Timoric language Tetun Terik displays two notable typologically oddities: 1) there are a vanishingly small number of vowel-initial verbs (despite plentiful vowel-initial nouns and adjectives), and 2) subject agreement is found on all verbs except this handful of vowel-initial predicates. 

This paper presents a historical account of these facts, showing that verbs that should have been etymologically #V have instead prothesised an unexpected initial /h/. On the basis of comparative evidence, I argue that this /h/ originated from the antepenultimate syncope of a causative prefix *pa > *ha- > h-, before being reinterpreted as a vowel hiatus repair strategy at morpheme boundaries and eventually spreading analogically to verb roots in their non-prefixed form. I further propose that this second step, involving the reinterpretation of morphological irregularity as a regular phonological repair, constitutes a form of phonological exaptation. 

• [handout] [recording]

• Epenthesis, prothesis, exaptation, reinforcing morphology

• Tetun Terik, Habun, Welaun, Mambae [Austronesian]

Languages within the Austronesian family provide robust evidence for reconstructing a stative prefix *ma-, used most often in adjectival contexts. This project investigates its apparently double functions in modern languages like Amarasi, Balinese, Indonesian, and Tagalog, where 'nominal' ma- combines with inalienable nouns to indicate possession, while 'verbal' ma- combines with verbs to produce mediopassive/middle-like semantics.

I propose that ma- instantiates an athematic/raising applicative, and that the difference between these two uses is entirely predictable based on whether the verb ma- that applicativises is a null copula or lexical verb. This analysis also provides a typology of Austronesian voice along two parameters: i) agentive vs. non-agentive VoiceP; and ii) extraction marking across the Voice head.

• [handout]

• ApplP, noun incorporation, possessor extraction/raising, VoiceP, medio-passives & 'middle' voice  

• Amarasi, Balinese, Indonesian, Tagalog [Austronesian], Sanskrit and Ancient Greek [Indo-European]

The West Timorese language Amarasi displays two surface-similar constructions with full pronoun doubling: in copular clauses and with quantified pronominal arguments. I argue that these two functions involve distinction underlying morphosyntactic structures. In copular contexts, I argue that Amarasi possesses a pronominal copula like Arabic, Hebrew, and Polish, which uniquely requires full phi-agreement. In quantified contexts, Amarasi appears to have adnominal pronoun constructions in which the predicate itself is a low pronoun, merged in nP. These distributions expand the list of potential competitors under Dependent Case Theory (Baker 2015) and shed light on the diahronic development of pronoun > pronominal copula > predicative agreement (Akkuş 2016; Abramovitz 2020).

• [publication] [handout] [recording] 

• PredP, copulas, pronominal/nP structure, adnominal pronoun constructions, dependent case assignment

• Amarasi [Austronesian], Arabic, Hebrew [Semitic], Polish [Indo-European] 

Vedic Sanskrit has a set of dvandva compounds in which both elements bear their own dual suffix, denoting pairs of individuals. These compounds show seemingly contradictory behaviour when it comes to the application of ruki, vocative accent assignment, derivational/inflectional suffixing, and meter. On one hand, they behave like single non-decomposable lexical units; on the other, they act like compound phrases comprising two independent words. This paper derives these facts through a Match Theory (Selkirk 2011) account of constrained mismatches in Prosody-Syntax mapping by proposing a revision of Match Head to Match Stem. The primary proposal is that prosodic words are only mapped onto categorised stems and not bare roots. This account also models how these dvandvas underwent syntactic (and prosodic) reanalysis from Vedic to Classical Sanskrit.

• [publication] [poster] [handout] 

• Compounding, nP structure, Prosody-Syntax mapping, prosodic recursion, Optimality Theory, syntactic reanalysis

• Vedic and Classical Sanskrit [Indo-European]

Turkish has a construction in which the passive morpheme is doubled, resulting in an impersonal passive interpretation. In Sakha, this doubled morpheme has been fossilised as the simple passive and reanalysed as monomorphemic, with homophony between the impersonal and passive uses. We investigate several morphosyntactic diagnostics to tease apart these functions, as well as the diachronic development of this reanalysis. Sakha thus provides additional support for a syntactically-projected impersonal pronoun (Legate et al. 2020) and the possibility of iterated VoicePs.

• [handout] [publication]

• VoiceP, passives vs. impersonals, anticausatives, case, syntactic reanalysis

• Sakha and Turkish [Turkic]

We explore a preliminary typology of cross-linguistic Case (mis)matching in Free Relative constructions, focusing on contexts where the Case requirements imposed on a wh-item by the embedded relative clause and the matrix clause differ. We identify four main types of Case (mis)matching scenarios, where languages differ based on parametric variation in the characterisation of Agree and feature valuation. We propose two novel parameters: Subset Checking and Feature Addition, which together derive the full typology of attested (mis)match patterns.

• [slides]

• (Multiple) Agree, Case containment, feature gluttony

• German, Polish, Gothic, Ancient and Modern Greek [Indo-European]

I investigate the semantics and pragmatics of the Singapore English (Singlish) sentence-final particle what. I propose that its denotation involves assertion of a given proposition as already existing within the Common Ground. This relies on the notion that the Common Ground is not an objective, single shared entity between interlocutors, but rather a projection held by each speaker. This work also explores cross-linguistic correlates including German doch, Mandarin/Hokkien ma33, and Cantonese wo21, and the contact scenario in which borrowing of this element into Singlish occurred.

• [slides] [publication]

• Update Semantics, Common Ground, Speech Act operators, contact-induced grammaticalisation

• Singapore English, German [Indo-European], Mandarin, Hokkien, Cantonese [Sino-Tibetan]

This paper explores the Singlish corollary of verb-copying resultatives in Mandarin, in which sentences containing both a direct object and post-verbal result/manner clause require two instantiations of the main verb. I propose an analysis involving VP-fronting for topicalisation, where the higher copy of the verb is rendered non-distinct from the tail of the chain due to the process of morphological fusion (Nunes 2004), resulting in Multiple Copy Spell-Out.

• [manuscript] [squib]

• Multiple Copy Spell-Out, PredP, resultatives, argument structure (the Post-Verbal Constraint)

• Singapore English [Indo-European], Mandarin and Cantonese [Sino-Tibetan]