The Syntax of Causative, Perception and Restructuring Verbs in Romance and Latin

Description of the event

In 2022, the annual SCUP (Syntactic Conferences at the University of Palermo) will be devoted to causative, perception and restructuring verbs. Although these predicates do not form a uniform class in the Romance languages, they show a number of common properties: they can all select bare infinitives and display some transparency effects, which were traditionally interpreted as pointing to a monoclausal structure (e.g., clitic climbing, single case domain, long passivization). Perception verbs, in addition, also occur with other types of clausal complement (gerunds, prepositional infinitives, pseudorelative clauses and complement clauses).

Historically, these constructions underwent a dramatic change in the passage from Latin to Romance. On the one hand, the typical transparency effects found in the current Romance languages seem to be generally unattested in Latin; in particular, clause union effects with embedded infinitives and the use of gerundial and Pseudo-relative clauses with perception verbs are Romance innovations. On the other hand, Latin could encode causation not only syntactically, but also morphologically (Vincent 2018). In addition, causative verbs like iubere ‘order’, complement clauses with the subjunctive in causative constructions and with the present participle in perception constructions were very frequent in Latin, but they all disappeared in the corresponding Romance constructions (with the exception of subjunctive clauses in Romanian and some Spanish varieties, Simone & Cerbasi 2001).

As far as the Romance languages are concerned, the commonalities between causative, perception and restructuring verbs have often led scholars to consider all three predicate types in their works, starting with Kayne’s French Syntax (1975), shortly followed by Radford’s Italian Syntax (1977). In his work, Kayne also introduced the terms faire-inf and faire-par, a crucial distinction in the analysis of Romance causative and perception verbs. Radford (1975, 1977), on the other hand, uses the term ‘Pseudo-relatives’ to refer to a construction typically found with perception verbs. These early works paved the way for a huge amount of studies on this topic in Romance (e.g. Bordelois 1974, Rouveret 1980, Raposo 1981, Suñer 1984, Burzio 1986, Guasti 1993, Treviño 1993, Campos 1994, Gonçalves 1999, Cinque 2006, Tubino Blanco 2010, Martins 2018, among many others).

One reason for the interest in the complementation patterns of causative, perception and restructuring verbs is the huge cross-linguistic variation. While most Romance languages do have Pseudo-relatives, gerundial and infinitival clauses, their distribution and syntactic properties vary. Gerunds, for example, are not used in these contexts in Italian and European Portuguese, but the latter language uses prepositional infinitives and inflected infinitives (with or without preposition, Raposo 1989, Barbosa & Cochofel 2004, Casalicchio 2019). In addition, constructions that are superficially similar may diverge structurally: Spanish gerunds are different from French present participles, notwithstanding their common origin. Pseudorelatives have different structures in Italian and Spanish (Casalicchio & Herbeck 2021), and there is also fine-grained variation within infinitival clauses of different Romance languages (see Labelle 2017, Mensching 2017 for a recent overview).

For non-finite complements of causative and perception verbs, various analyses have been proposed: they range from Exceptional Case Marking (ECM; or accusativus cum infinitivo, a term stemming from the grammar of Latin; cf. Mensching 2017:382) to subject-to-object raising, from Small Clauses to verb incorporation. Not only do these analyses make different predictions, but they also imply different views of the general properties of the matrix verbs. An ECM-analysis, for example, treats perception and causative verbs as two-place predicates; on the other hand, analyses postulating a control structure interpret the matrix verbs as trivalent (Moore 1996). In addition, various recent studies claim that the infinitival clause selected by these verbs can have different sizes (e. g. Sheehan & Cyrino 2018, Sheehan 2020), depending on the language but also on the matrix verb used.

Among the different complement types, a noteworthy construction is the so-called Pseudo-relative clause (‘PR’), which is found in most Romance languages but absent in some other language families (Guasti 1988, 1992, Cinque 1992, Casalicchio 2016). However, even in this case, the same construction has different syntactic properties in different Romance languages. For example, PRs vary in their distribution, and they are compatible with matrix passivization in some languages (e.g., Italian), while they are not in others (Spanish). This raises the question whether different underlying structures and analyses should be applied to PRs in different Romance languages (such as a small clause, ECM, raising-to-object, or ditransitive control structure).

What all these constructions have in common is the presence, to various degrees, of transparency effects. Some tests that have been applied are clitic climbing, long passivization and se-passives. Transparency has been observed especially with restructuring verbs, and this has led some scholars to propose monoclausal analyses, in which the restructuring verb occupies a functional head (e.g. Cinque 2006); however, other scholars postulate biclausal analyses in which the complement is either reduced or an additional process of clause union or verb incorporation is assumed (e.g. Aissen & Perlmutter 1978, Manzini & Savoia 2007, 2011). Furthermore, it has been argued that (non-)restructuring is “graded” and that different ‘sizes’ of non-finite clauses are at stake (see Wurmbrand 2001, 2014, 2015).