Topics and passives in Italian-speaking children and adults

ABSTRACT

Through two elicited production experiments we investigated how preschool Italian-speaking children access the left periphery of the clause with respect to topics in Clitic Left Dislocation (ClLD) structures. Since the discourse conditions of the experiments are felicitous for the production of passives as well, we also investigated children’s production of different types of passives, and how it compares to adults’. A rich and variegated array of results indicate young children’s early access to the left peripheral topic positions—also in a nonadult manner through use of a-marked topics—and preference for ClLD over passive in contrast to adults; children’s early access to passive in the causative voice also emerged as well as use of ClLD with null generic subject as an alternative to the (copular/venire) passive. Intervention locality/relativized minimality plays a crucial role in interpreting the articulated results within the system developed in Friedmann, Belletti & Rizzi (2009) and much subsequent work.

Acknowledgment

The research presented here was funded in part by the European Research Council/ERC Advanced Grant 340297 SynCart – “Syntactic cartography and locality in adult grammars and language acquisition”.

Notes

¹ In the passive sentence, both the auxiliary essere/‘to be’ and venire/‘to come’ are felicitous in the answer; however, with actional verbs in the present tense the auxiliary venire/‘to come’ is preferred by adults and children (Belletti & Guasti 2015).

² This preference remains also for school-age children, at least until age 9, as shown in Del Puppo & Pivi (2015) and Manetti (2017). In these works, the experimental question was patient oriented, but the designs differed with respect to the representation of the agent character: In Del Puppo & Pivi’s work the agent was partially covered, whereas in Manetti (2017), the agent was fully shown on the picture. This difference did not affect children’s preference for the use of an active verb with the object clitic pronoun.

³ This detail has been added to the story to favor a fully informative answer: Notice that the child can well see what is happening inside the house and can thus provide all the information asked by the Smurf.

⁴ In both cases, the actions were performed on the patient(s) by two distinct agents in the aim of creating two independent events. Note incidentally that in a full informative answer this should lead to the production of an overt subject (see Sections 3.6.1 and 5.2). For similar and related studies on the manipulation of the number of agent referents in the eliciting question, see Serratrice (2008) for English and De Cat (2009) for French.

⁵ The importance of this aspect of our design will become apparent in the discussion of children’s answers containing a-topics (Section 5.2) that cannot thus simply be interpreted as repetitions of the a-experiencer of the question.

⁶ The result is robust, as it has been confirmed by a further group of adults with a low level of schooling who were later tested through the same elicitation task (data are not reported here). Hence, use of passive is not the reflex of educated Italian but rather the reflex of a productive syntactic construction.

⁷ Use of SVO structures in response to patient-oriented questions has been already observed in previous studies, in children and -to a lesser extent- in adults as well (Del Puppo & Pivi 2015; Manetti 2013; Volpato, Verin & Cardinaletti 2014; Manetti 2017). SVO answers allowed children and adults to convey the information on the patient, which the question asked for, through an all-new description in which the object is not singled out as a topic. Notice that children produced SVO structures more often in the two-topic condition (18 SVO in one-topic condition vs. 84 in two-topic condition), which is the one requiring the overt expression of the topic in the ClLD. Hence, use of ClLD would have to comply with an intervention configuration, a notoriously hard structure for children to compute (see sections 2.1, 5.1, and 5.2 for detailed discussion). A subgroup of children (7/36 = 20%) systematically adopted the SVO answer in this condition (SVO = 48, 86% of their answers) and never produced ClLDs; in contrast, the same group produced Pronoun structures with no overt expression of the topic in the one-topic condition (Pronoun = 46, 82% of their answers). This further reveals the great difficulty that the DP₁-DP₂-Cl-V structure with both DPs overt and lexical poses to children, which is the fundamental insight of the intervention account.

⁸ ClLDs were practically absent in adults’ answers (only three produced by one participant).

⁹ When the subject was preverbal, it either preceded (S-O-cl-V: 62%) or followed the preposed left-dislocated object (O-S-cl-V: 38%): The preference for the S-O order is not replicated in Experiment 2, where the preference is reversed (O-S 71% vs. S-O 19%; see footnote 12), for this reason we do not analyze this order issue any further.

¹⁰ Note that the discourse condition of the experiment did not favor the postverbal position of the subject, which was among the introduced characters in the pictures. Hence it did not count as a new information focus, the most typical interpretation of a postverbal subject in Italian (Belletti 2004). The realization of the subject as null –pro- in the answer is also not favored, as the question refers to two different subjects. It is then not surprising that the subject was mostly overt, lexical, and preverbal. See section 5.2 in the discussion.

¹¹ Overall, adults preferred the periphrastic passive with the auxiliary venire/‘to come’ (81%), as in Experiment 1 (see section 3.6.2.).

¹² The subject preceded the object in 29% of ClLDs (7, S-O-cl-V) and followed the object in the 71% (17, O-S-cl-V). (cf. footnote 9).

¹³ The larger number of ClLDs in Experiment 2 also reflects the larger number of children producing the structure (89%) compared to the more restricted group of children producing ClLDs in Experiment 1 (47%).

¹⁴ On the possibility of iterating topic positions in the Italian left periphery, see the original discussion in Rizzi (1997). On the possibly different discourse value of the different topic positions, see Bianchi & Frascarelli (2010) and Frascarelli & Hinterhölzl (2007). See Manetti et al. (2016) on the possible relevance of Frascarelli & Hinterhölzl’s (2007) distinction in the comprehension of these ClLD by adults. In the interest of clarity, in the following discussion we will assume that both positions iterate topics of the same type in the children’s grammar (see also footnote 19 and the following).

¹⁵ In few of their ClLDs children used a postverbal lexical subject or a null subject. In both cases the answer is not completely appropriate: in the former case because the subject should not count as new information in the answer as it is one of the introduced characters; in the latter because it should be overt as it cannot be distinguished from the other subject if it is not pronounced. The fact that the only inappropriate children’s answers concern the position (postverbal) and nature (null) of the subject suggests that the presence of the lexical preverbal subject is particularly hard to handle by children in the ClLD configuration. This conclusion is consistent with the account in the text and does not run into conflict with the general early mastery of position and nature of subjects by young children previously mentioned in the text and in the following.

¹⁶ See also Escandell-Vidal (2009) on Balearic Catalan in which a-marking only concerns left-dislocated objects in a way that closely corresponds to the distribution found in the Italian developing children, who are thus adopting a possible grammatical option as is typically the case in stages of acquisition (see Belletti Forthcoming and references cited therein for some speculation on the possible innovation that this type of children’s productions may lead to in Standard Italian, as happened in previous stages of, e.g., Spanish and Japanese).

¹⁷ a-Topics are evenly distributed in the few cases in which the subject was either a null referential third person singular pronominal or a postverbal one (cf. section 3.6.1).

¹⁸ The conclusion is based on Belletti et al.’s (2012) results on the effect of gender mismatch in enhancing the comprehension of Object relative clauses in Hebrew. The main result of that work is that gender is a feature relevant for the computation of fRM in Hebrew but not in Italian. An Intersection relation of relevant features is created in Hebrew Object relatives in a gender mismatch condition but not in Italian, enhancing the comprehension of Object relatives in the former language but not in latter. For further details on the implementation of fRM in these terms and its relevance to interpret development, see Friedmann, Belletti & Rizzi (2009), Belletti et al. (2012), Bentea (2016), and Guasti, Branchini & Arosio (2012). For an original related approach in terms of the grammatical principle fRM to account for impaired computations of object long-distance dependencies in aphasic patients, see Grillo (2008); Friedmann, Yachini & Szterman (2015) on the relevance of the approach in the production of Object relatives in the SLI population; Belletti & Rizzi (2013) on the relevance of the principle in adult processing; Friedmann, Rizzi & Belletti (2017) for recent further illustration on the (lack of) role of Case. On the role of the feature number in creating an intersection relation enhancing the comprehension of ClLDs in Italian, see Manetti et al. (2016).

¹⁹ The subject can also be left dislocated in DP₁-DP₂-Cl-V, as discussed in 5.1 and illustrated in Examples 23 and 24. If the two topic positions are treated as different (see footnote 14), with the highest one endowed with a further feature, the structure would still instantiate the Inclusion relation, the hard relation for children. See Manetti et al. (2016) for discussion along these lines, in the spirit of Frascarelli & Hinterhölzl (2007) and Bianchi & Frascarelli (2010). Alternatively, children may not differentiate the two topic positions as assumed here for the sake of simplicity (see footnote 14; see also the following structure in Example 31). If this were the case, the structures would reduce to a case of Identity (with both DPs endowed with the same Top and NP features).

²⁰ As shown in Costa et al. (2014) on PP relatives in EP and Hebrew, the categorical distinction DP vs. PP cannot be considered the relevant distinctive factor, as PP relatives are as hard for children as (direct) Object relatives, despite presence of the Preposition in the former case.

²¹ The feature +Top is indicated in parenthesis since, as discussed, with this word order the subject could be either left dislocated or fill the subject position within the clause (cf. section 5.1). The fact that the relevant feature for fRM is lexicalized through the preposition a may make the intersection relation more readily available to children. Thanks to one reviewer for suggesting this possible supplementary factor.

²² These children’s data on the production of ClLDs in Italian share close resemblance with data on children’s comprehension and production of Object relative clauses in Hebrew presented and discussed along similar lines in Friedmann, Belletti & Rizzi (2009). Hebrew, similarly to Italian, has a generic plural null subject, and use of such a generic plural null subject significantly enhanced children’s comprehension and production of the otherwise hard-headed object relatives.

²³ The likelihood of a cumulative role played by the feature number comes from the consideration of previous comprehension results (see Manetti et al. 2016, footnote 18), which reported that the number mismatch condition did significantly enhance children’s comprehension of the ClLD structures, which had both the left-dislocated direct object and the subject in the following clause realized as lexical noun phrases, hence both [+NP] (of the type: Il pinguino, i gatti lo lavano/‘the penguin the cats him-Cl wash’), in line with the fRM approach.

²⁴ See Manetti & Belletti (2015) using the priming technique adapted from Messenger et al. (2008). See also Contemori & Belletti (2013) indicating use of si-causative passive structures in Passive Object Relatives/PORs (cf. section 2.2.).

²⁵ For instance, Hebrew appears to be a language with this type of distribution (Ur Shlonsky personal communication), as well as Greek (as one reviewer has pointed out).

²⁶ The aboutness relation holds even if the subject has not been previously mentioned in discourse, even if it is not a discourse topic. This is illustrated by all new sentences, which can have a preverbal new indefinite subject:(ii)a.Q: Cosa è successo?‘What happened?’b.A: Un camion ha tamponato una macchina.‘A truck has bumped into a car.’Here the sentence is about un camion/‘a truck.’ In the passive sentence as (iii):(iii)a.A’: Una macchina è stata tamponata da un camion.‘A car has been bumped into by a truck.’The sentence is about una macchina/‘a car.’ See Rizzi (2005, 2018) for relevant discussion and Manetti (2017) for related considerations with children.

²⁷ Prosodic experimental evidence may also be of help in determining this point. This is an open question for now.