[SynSem] Edoardo Cavirani (KU Leuven)

Illusions in the nP domain -- Syntax-phonology mismatches in Italo-Romance gender and number systems

UPS Pouchet salle 124 & zoom
Meeting ID: 936 6928 0888        Passcode: hAi3kf


retour Séminaire Syntaxe et Sémantique

Edoardo Cavirani (KU Leuven)

Illusions in the nP domain
Syntax-phonology mismatches in Italo-Romance gender and number systems (abstract pdf)

This talks focuses on some unexpected patterns displayed by the DPs of (i) a set of Northern- Italian varieties spoken in Lunigiana (MS), and (ii) Standard Italian. In both cases, unexpectedness concerns (i) the linear ordering of the PL marker, which surfaces closer to the root than the F marker, as in peg@r-jPL -aPL ‘sheep.PL’, and (ii) its distribution within the DP, the PL marker surfacing on almost all logically possible subsets of DP-internal constituents (Art, Dem, Q, Poss, A).

Interestingly, the observed linear ordering does not comply with the Mirror Principle, which, in this case, would predict a √-GEN-NUM ordering (*pegr-aF-jPL, cfr. Sp. ovej-aF-sPL ‘sheep.PL’). Building on strict CV, I will argue that the observed ordering is a phonological ‘artifact’, in the sense that it follows from the phonological shape of the involved markers. This allows for keeping the morphosyntactic derivation as standard and as in line with other Romance varieties as possible.
As for the variable realization of the PL marker in the wider DP domain, the patterns provided by Lunigiana varieties represent a challenge for current analyses of partial concord (e.g. Cyrino and Espinal 2020, Pescarini 2021), for structures can be found where concord ‘skips’ a constituent, as in (1). Furthermore, in most cases, the realization of the PL marker is optional, which seems to point to an adjunct/modifier formal status of PL (Cyrino and Espinal 2020, Pescarini 2021, Cabredo-Hoffher 2021, Wiltschko 2021). However, as shown in (2), even if not realized, the verb displays PL morphology, which suggests that there is actually a PL head controlling agreement
(Wiltschko 2021).

(1)  tut-j-a l-a nOstr-ja niece-F
       all-PL-F the-F our-PL-F nvOd-a
      ‘All the our nieces’
(2) l-a pegr-a l en be.3PL l-a nOstr-a
     the-F sheep-F  SCL EN the-F our-F
    ‘the sheep are ours’

Thus, the patterns just mentioned raise some questions, e.g. where exactly is NUM (low, high, both ...), what is its formal status (head, adjunct ...), what is concord domain, why there is no partial concord with M, who controls agreement when the PL marker is not realized, how to formalize PL unrealization, how to make sense of ‘holes’, why there is no partial concord with M? In the talk, I will address these questions, and I will sketch some possible answers.

Time permitting, I will also discuss a pattern that represents a sort of mirror image of the agreement-without-realized-controller case in (2). The pattern involves expressive expressions containing forms such as cazzo lit. ‘dick’. As shown in (3-4), whereas expressive NPs such  as merda ‘shit’ must NUM-agree with the referential NP, and must be agreed with by D, expressive NPs of the cazzo-type (5-6) do not NUM-agree with the referential NPs (in fact, they cannot pluralize), and can be skipped by D, which can agree with the lower referential NP (5).

(3) un-a merd-a di pecoron-e
a-F.SG shit-F.SG of ram-M.SG
‘a shitty ram’
(4) *un-∅ merd-a di pecoron-e
a-M.SG shit-F.SG of ram-M.SG

(5) un-∅ cazzo di pecor-a
a-M.SG dick-M.SG of sheep-F.SG
‘a fucking sheep’

(6) un-a cazzo di pecor-a
a-F.SG dick-M.SG of sheep-F.SG

Another striking property of cazzo-type expressive NPs is that they seem to be higher than other expressive NPs. As show below, cazzo can combine with other expressive NPs, such as merda, and they can both occur either precede, or follow the referential NP. Interestingly, in both cases, merda needs to be closer to the referential NP. This suggests that cazzo, i.e. ‘pure’ expressive NP, are merged higher than other expressive NPs.

(7) un-∅ cazz-o di merd-a di pecor-a
a-M.SG dick-M.SG of shit-F.SG of sheep-F.SG
‘a FUCKING sheep’

(8) *un-a merd-a di cazz-o di pecor-a
a-F.SG shit-F.SG of  dick-M.SG of sheep-F.SG

(9) ?un-a pecor-a di merd-a del cazz-o
a-F.SG sheep-F.SG of  shit-F.SG of.the  dick-M.SG
‘a FUCKING sheep’

(10) *un-a pecor-a del cazz-o di merd-a
a-F.SG sheep-F.SG of.the  dick-M.SG of  shit-F.SG

I will tentatively argue that this state of affairs follows from a difference in grammaticalization between cazzo (functional) and merda (semifunctional), along the lines of Cavirani-Pots (2020), Klochmann (2017). As such, I will argue that, despite the (phonological) appearance, cazzo is not morphologically complex, and lacks GEN and NUM altogether.


Pas d'interprétation en LSF