Object Clitic Omission: Two Language Types

Abstract

The literature generally assumes that object clitic omission is equally allowed in all child languages. In this paper we challenge this claim by means of an elicitation experiment carried out with children acquiring two closely related languages, Catalan and Spanish. Our results show that while omission is high in young Catalan-speaking children, it is very low in Spanish-speaking children. We argue that this difference can be attributed to a property of their respective grammars (the presence or otherwise of past participle agreement when objects cliticize) under the Unique Checking Constraint of Wexler (1998). In a second experiment, we confirm the robustness of early sensitivity to past participle agreement; through a grammaticality judgment task we find a statistically significant difference between the two languages. We show that the parametric approach postulated can be extended, for mandatory, overt objects, beyond Catalan and Spanish, broadening the empirical scope of the Unique Checking Constraint.

Notes

¹The external argument is merged in Spec,vP. The associate of the clitic, pro, moves through an outer Specifier of vP (as in Chomsky 2001).

²An anonymous reviewer points out that we could equally adopt D'Alessandro & Roberts' (2008) rephrasing of past participle agreement in the terms of phase theory; their analysis of participle agreement in Italian would carry over to Catalan and French (since in these varieties there is no agreement with a full DP object); agreement would take place by Agree within a phase, but double checking would still take place.

³We take it that ClP instantiates definiteness in the case of definite clitics; if so, this semantic property is an interface requirement, just like tense is for a clause.

⁴An earlier version of this experiment, carried out with a subset of the children, appeared in Wexler, Gavarró & Torrens (2004).

⁵Of course, the Spanish-speaking children never produced an agreeing participle in this experiment. It should be noted that Spanish participles have the morphological possibility of agreement, as can be seen, for example, in participles as predicate adjectives and other constructions. But Spanish-speaking children know that the participle in the perfect shows no agreement with clitics.

⁶The results indicate that optionality is found in individuals: of the children who produced participle agreement, produced it only some of the time, and 3 produced it in all cases; for adults, 26 produced it only occasionally, and 6 produced it all the time.

⁷Equally, a phonological analysis along the lines of Schaeffer's cannot extend to Catalan for the reasons adduced above; as for Italian, if children were overgeneralizing a contraction rule, we would expect that overgeneralization to carry over to the homophonous definite articles; no such evidence has ever been reported, to our knowledge.

⁸Bulgarian participles are inflected for gender and number, but do not agree with the object, rather they agree with the subject (Ivan Ivanov, pers. comm.).

⁹Object clitic omission is documented for Colombian Spanish, as well as many other American Spanish varieties (from Argentina, Bolivia, Ecuador, Paraguay, and Perú)—see Campos (1999) and Schwenter (2006) for surveys. According to Schwenter (2006) the omission of object clitics (in monotransitive constructions) in South American varieties is widespread with [—animate] objects. Null objects are found quite systematically in Colombian Spanish for definite DPs in clitic sequences (i), but also in isolation (ii) (example (i) cited in Campos (1999), example (ii) due to Azucena Palacios, pers. comm.). These examples are ill-formed in Peninsular Spanish (as well as in Catalan, Italian, etc.).

Cali Spanish belongs to the El Valle variety of Colombian Spanish and has been historically subject to the influence of Andean Spanish, which presents null objects in turn due to the influence of Quechua; contemporaneously, Cali has a large immigrant community from the Nariñno department, and in that variety null objects are also found (Rubén Arboleda, pers. comm.).

¹⁰Ungrammatical agreement with a postverbal object was first attested in child Italian spontaneous production (Antinucci & Miller 1976; Volterra 1976) and was considered a systematic feature of child Italian in Borer & Wexler (1992). Pirvulescu & Belzil (2008) provide the first experimental evidence for it in French.

¹¹These word order patterns correlate with other phenomena, in, e.g., binding: Spanish allows Pintará cada coche_i su propietario_i , lit. ‘Will paint each car its owner,’ but Catalan disallows it: *Pintarà cada cotxe_i el seu propietari_i (there is no consensus as to how to best treat all these facts, but see Gallego 2007). According to the data by Belletti (2004) (and contra Cardinaletti 2001), Italian behaves like Catalan.

¹²An anonymous reviewer points out that in some Southern Italian dialects participle agreement coexists with object clitic doubling; this would undoubtedly be a problem for the generalization in Tsakali & Wexler (2004), and requires further research; notice however that this fact does not weaken the UCC analysis of clitic omission.