Research

Based on evidence from word order and the licensing of PGs, I argued that (1b) is not in situ, but rather has moved to a dedicated whP in the in the Low Left Periphery (LLP) at the edge of vP (Belletti, 2004; Bonan, 2019). 

The next question that arises is whether (1a-b) differ structurally or whether they are derivationally equivalent. In the first case, (1a) would have the derivation in (2a) and (1b) that in (2b). 

(2) a. [whP quan [TP te va [whP quan [vP [VP quan va en vacanse ]

b. [whP [TP te va [whP quan [vP [VP quan va en vacanse ]

In the second scenario, both (1a) and (1b) would have the exact same derivation: one wh-chain with the possibility to spell out a different copy, as in (3).

(3) a. [whP quan [TP [whP quan [vP [VP quan ] Higher copy

b. [whP quan [TP [whP quan [vP [VP quan ] Lower copy

To answer this question, I am looking at several diagnostics: inverse scope, binding, parasitic gaps, and islands.

Exclusives differ from exceptives in that they express, indeed, exclusion rather than an exception to a generalization or domain subtraction. In (2) some students who aren't Maria came, but the speaker is not saying anything about whether Maria came or not. 

In this project, I am arguing that structures like (1) and (2) differ both syntactically and semantically. Eccetto + universal quantifiers constructions (1) are ‘true exceptives’: they carry all three inferences. Furthermore, they have an underlying clausal structure, which involves ellipsis. On the other hand, eccetto + existential quantifiers, numerals, and wh-words constructions (2) are phrasal. Furthermore, the latter do not carry any of the inferences mentioned above. In fact, they subtract an entity, Yuri, from the common ground, in an ‘anti-topic’ behavior. 

P-mismatches do not, however, freely occur (Table) and are limited to certain PPs PG, arguments of verbs like litigare 'to argue' and innamorarsi 'to fall in love'.

These verbs belong to the class of together reciprocal (Levin, 1993), which undergo the alternation illustrated for English in (3-4).

3. John and Mary argued. - John argued with Mary. 

4. John and Mary fell in love. - John fell in love with Mary.

These data points are not only very interesting but provide a good argument in favor of unification of PGs and ATB extraction.

 Wanna know more? Email me for a draft of our paper!

In Italian, like in most romance languages, there are two kinds of reflexives: derived and inherent. The former are derived from transitive (Gianni si lava ‘Gianni washes himself’) and ergative verbs (Gianni si ferisce ‘Gianni wounds himself’). Inherent reflexives, on the other hand, only exist in this form. While derived reflexives can be formed with any transitive verb and thus basically form an open category, inherent ones are not very common and are not productive. To this category belong both stative and action verbs, with an animate subject. Examples are ammalarsi ‘to fall ill’, pentirsi ‘to regret’, and arrabbiarsi ‘to get mad’.

Interestingly, derived and inherent reflexives behave differently in certain syntactic context. First, inherent reflexives can be made causative (Faccio ammalare Lucia ‘I make Lucia fall ill.’), while if made causative derived ones lose their reflexive meaning (Faccio lavare Lucia ‘I make Lucia be washed’). Furthermore, these two categories behave differently wrt to ne cliticization. Such structure has been used for decades as a diagnostics of unaccusativity, as only verbs with an internal subject seem to allow ne cliticization. Derived reflexives do not allow ne cliticization (??Se ne sono lavati tre ‘Three have been washed’), but interestingly inherent ones do (Se ne sono ammalati tre ‘Three have fallen ill’). This is crucial evidence, as for decades the discussion on the nature of reflexives has considered the ne restriction as proof for an analysis of reflexives as intransitive.

This asymmetry suggests that derived and inherent reflexives are structurally different, a hypothesis that has not yet been formally investigated.

If we lived in a utopian world, languages would present a perfect one-to-one correspondence between meaning and form. In reality, however, languages display a very uneven correspondence between the concept and meaning levels. There is no need to dig deep into the realm of morphosyntax to realize this. Look at “Don’t eat any chips!” or “Many cats“: in both sentences a unit on the meaning level, negation and plurality, has been represented twice. This phenomenon is called Redundancy and, together with Discontinuity (the famous French ne…pas), Fusion (e.g. Cumulation of gender and number features in Spanish for instance), and Form-based-form (e.g. Expletives), it is one of the ways in which languages violate transparency. The present is a very adaptive framework, ideal for typological research, especially in contact situations.

I have carried out two studies in this framework, both as part of the broader project Transparency in Language at the University of Amsterdam. The first has been on Haitian Creole, and its sub- and superstrate languages Fongbe and French. Creoles have been described for decades as “simpler” than older idioms, thus in my research, I investigated whether they were in fact more transparent. Turns out it is so, which suggests that languages are born with a very transparent nature and over time become more opaque.

The second investigation, on the other hand, dwelled on the origin of the Balkan Sprachbund, in which I looked at Bulgarian, Greek, Italian, Romanian, and Russian. This study confronted two models: convergence (aka contact) and diglossia. While the first model has been the most acclaimed one in the past century, my results point in the opposite direction. The presence of a prestige language used by the clergy seems to have influenced the development of the languages of the Balkan Sprachbund.

More on Transparency can be found here.