Events
RoLinC: Mazzola, Sheehan, Raffy & Garside
Speakers: Giulia Mazzola, Michelle Sheehan, Clementine Raffy, and Liam Garside (Newcastle University)
Title: The syntax and semantics of visual and auditory perception in Romance languages
Date/time: 14th October, 2-3 pm
Abstract
Abstract: Wurmbrand and Lohninger (2023), building on Givón (2001),propose a principle of Minimally Required Domains, whereby in order for a syntactic structure to map to a given semantic object, it must be of a certain size or larger. Building on Ramchand & Svenonius (2014), on the basis of cross-linguistic evidence, Wurmbrand and Lohninger argue that there must be at least three kinds of complement clauses in terms of semantic denotation: (i) events (which are tenseless and therefore simultaneous with a matrix subevent); (ii) situations (which have a pre-specified tense value, often future) and (iii) propositions (which have embedded reference time and may involve speaker-oriented parameters). Their cross-linguistic study of transparency effects suggeststhat these semantic types have Minimally Required Domains. Roughly speaking, events must be at least as large as the thematic domain, situations must contain some tense/mood/aspect projection and propositions must have fully specified tense and be more ‘clausal’. This means that, across languages, events/situations can also be denoted by larger structures even as large as CP, but the reverse is not true: because of containment, nothing lacking tense can denote a proposition.
Romance perception verbs offer a unique testing ground for this proposal as, in many cases, the same verb can occur with multiple complement types which have been shown syntactically to be of different sizes (see Casalicchio & Sheehan 2025 for a recent overview). For example, the ‘see’ verb in Italian permits the following kinds of clausal complements: (1) faire–par, with the perceivee as an adjunct; (2) faire infinitive (perceive with dative marking) andExceptional Case Marking (ECM, perceive with accusative marking); (3) finite CPs:
(1) Ho vistosuonare ilpiano da Pietro
have.prs.1sgsee.pst.ptcpplay.infdet.m.sgpiano by Pietro
‘I saw the piano played by Pietro.’
(2) {L’=/gli=}hovistosuonare ilpiano.
3sg.acc.m /3sg.dat=have.1sgsee.pst.ptcpplay.infdet.m.sgpiano
‘I saw him play the piano.’ (l’=ECM, gli=faire infinitive)
(3) Ho visto che Pietro suonava il piano.
have.prs.1sgsee.pst.ptcp comp Pietro play.pst.ipfv.3sg det.m.sg piano
‘I saw that Pietro was playing the piano.’
It is also well known that there are different kinds of perception (Dik & Hengeveld 1991, Enghels 2019, Moulton 2009). Direct perception involves spatiotemporal overlap of the perceiving and perceived events, whereas indirect perception involves spatiotemporal separation of the two events and, when most indirect, cognitive inference on the part of the observer. In Wurmbrand and Lohninger’s terms, this translates into: (i) direct perception of events; (ii) indirect perception of situations (e.g., seeing into the future) and (iii) cognitive inference of propositional content. The hypothesis suggested by Minimally Required Domains is that allcomplement types larger than the thematic domain can denote events leading to direct perception, whereas only the largest clauses can denote cognitive inference. We test out this hypothesis on Italian, European Spanish, European Portuguese and French, in a parallel corpus study. Our large-scale corpus study of these languages uses the TenTen corpora(Kilgarriff et al. 2014), analysing randomly extracted occurrences of see, hear, listen and watch (n=2,000 per verb/language) which are followed within 5 words by another verb. Once false positives were discarded, we were left with 2,800 total occurrences. All examples were manually filtered, analysed and tagged for syntactic and semantic criteria. We used the same semantic classification system across alllanguages and verbs, and we identified a nuanced scaled of meanings which complements in a more granular way the binary direct/indirect perception distinction investigated so far. We classified our meanings as: (i) direct perception (intentional or unintentional (4)-(5)); (ii) hybrid exposure meanings, like in (6); (iii) epistemic meanings, either foresight (8), or belief (commitment to a proposition) (9) and (iv) cognitive inference of a proposition (10).
[Italiano] [direct: intentional]
(4) Il brasiliano […] lo= guardò alzar=si.
the Brazilian acc.3sg.m= watch.pst.3sg get-up.inf=refl.3sg
‘The Brazilian watched him get up.’ (itTenTen, galeonedeifolli.it)
[French] [direct: unintentional]
(5) si elle le= voyaitse= gaver de sucreries ainsi.
if she 3sg.acc=see.pst.ipfv.3sg refl.3sg=stuff.inf of sweet.treats thus
‘…if she saw him stuffing his face like this.’ (frTenTen, manyfics.net)
[Eur.Portuguese][hybrid: exposure]
(6) crescemos a ouvir que basta uma vez e pimba!
grow-up.prs.1pl to hear.inf comp enough indf.f time and bam
‘We grow up hearing that once is enough and bam!’ (ptTenTen, demaeparamae.pt)
[Eur. Spanish][epistemic: foresight]
(7) Zidane, muy listo viovenir
Zvery smart.m.sg see.pst.3sg come.inf
la crisis
det.f.sg crisis(f)
‘Zidane, smartly, saw the crisis coming’ (esTenTen, quijotedigital.es)
[Eur. Portuguese] [epistemic: belief]
(8) nem vemos que haja qualquer motivo para a alterar.
nor see.prs.1pl comp there.be any motive for acc.f.sg=change.inf
‘nor do we see that there is any reason to change it.’(ptTenTen, dgsi.pt)
[Italiano] [indirect: cognitive]
(9) Non si può fingere di non vedere
neg imprs=can.prs.3sg pretend.inf of neg see.inf
che ciò sarà negativo per il Paese.
comp that be.fut.3sg negative for det.m.sg country
‘You cannot pretend that you don’t see that that will be negative for the country.’ (itTenTen, camera.it)
We then analysed the data using Bayesian Hierarchical Ordinal Regression, to see whether complement size (the outcome ordinal variable) is predicted by the semantic interpretation, and whether the prediction pattern mirrors the containment model proposed by Wurmbrand and Lohninger. Preliminary results point at a strong association between complement type and interpretation across languages. Propositions are categorically predicted to be encoded by CP complements, while situations and events show strong probabilistic tendencies in the expected directions. The results support Minimally Required Domains as complement of all sizes are compatible withdirect perception but only CPs surface with cognitive readings. We also discuss less clear patterns, related to meanings of perception verbs that do not encode pure perception, and some (rare) puzzling examples of epistemic foresight readings with very small complements. All in all, we show that beyond the pure non-finite/finite and direct/indirect distinction, semantics plays a clear, predictable role in syntactic optionality for promiscuous verbs.
If you wish to attend, please register here