Research

Word order and word order change

Different languages exhibit different word orders. While this may seem obvious, it raises the question: why is this the case? Grammar rules function as the DNA of a language. My enduring research interest lies in documenting and analysing fundamental word order patterns and word changes throughout the history of the Romance languages — and exploring their implications for our understanding of language variation.

Clitic placement

Object pronouns in Romance languages are particularly interesting because they occupy a different syntactic position than the nouns they replace and are often fused onto the verb: Jean aime Marie vs. Jean l'aime. These pronouns are known as clitics. Clitic placement has garnered significant attention in the comparative, diachronic, and formal linguistic literature — domains that my work spans.

One area of my research investigates why French (le faire) and Italian (farlo) exhibit different word orders with clitic placement in infinitival constructions. To explore this, I traced the phenomenon back to Old French. My findings indicate that early French initially patterned like Italian, but underwent a syntactic change around 1300. Interestingly, this change was not in clitic placement itself, but in verb movement: the infinitive used to move to the left of the clitic. I argue that this movement ceased due to a phonological shift. More technically, I propose that the weakening of non-finite T led to the loss of V-to-T movement, which in turn gave rise to proclisis.

I have also investigated clitic climbing (e.g., je le veux faire) and its decline in French. My research shows that clitic climbing was obligatory until the 17th century, after which it gradually weakened — first with me, te, se, then with le, la, les, and eventually with en and y.

Auxiliary selection

French, like other Romance languages, forms its compound tenses using avoir or être (‘have’ or ‘be’) plus a past participle. Predicting which auxiliary is selected has long posed a challenge in the formal literature, particularly in relation to reflexive verbs. While much of the existing research attributes auxiliary selection to argument structure — specifically, whether a verb is transitive or not — I argue against this hypothesis because indirect reflexives are clearly transitive yet select être (e.g. je me suis lavé les cheveux). Instead, I demonstrate, within a minimalist framework, that auxiliary selection is determined by the syntactic operation Agree. More precisely, I propose that the choice of auxiliary reflects the interaction of person feature identity among elements in the structure.

The framework I have developed also accounts for auxiliary switch in restructuring contexts — a phenomenon where the auxiliary appears to be selected by the embedded infinitive rather than the matrix verb (e.g., sono voluto partire instead of ho voluto partire in Italian). To better understand this under-documented phenomenon, I have analysed data from the diachrony of French, Sardinian, Occitan, and Italian. My findings indicate that reflexive clitics consistently trigger the use of être in these contexts, whereas locative and partitive clitics (y and en) do not, which further supports my analysis based on person feature identity.

Restructuring clauses

Restructuring clauses involve matrix verbs that behave like auxiliaries — namely, modal, aspectual, and motion verbs — and select an infinitival complement. This syntactic environment is particularly intriguing because, despite the presence of two verbs, the structure often appears to be monoclausal. This is evidenced by phenomena such as clitic climbing and auxiliary switch, both of which have been central to my research. In addition to these, I have investigated the structural properties of restructuring clauses themselves.

Drawing on data from a range of Romance languages — including Catalan, Romanian, Old French, and Old Italian — I explored the role of intervening complementizers, which can signal biclausality. The findings of this research show that clause size varies depending on both the matrix predicate and the language in question. This cross-Romance variation reveals that restructuring cannot be adequately captured by a one-size-fits-all analysis.

As part of this broader inquiry, I also examined infinitive fronting, a phenomenon attested in Medieval Romance. I argue that such fronting is only possible when the matrix predicate is a restructuring verb. This insight has prompted a re-evaluation of transparency effects — their nature, distribution, and typology across Romance — and has contributed to refining our understanding of how many distinct types of transparency effects there actually are.