BRIEF RESEARCH REPORT published: 07 February 2020 doi: 10.3389/fpsyg.2020.00105 Child Relativized Minimality and Grammaticality Judgement Anna Gavarró* Departament de Filologia Catalana, Centre de Lingüística Teòrica, Universitat Autònoma de Barcelona, Barcelona, Spain Grammaticality judgements are the fundamental experimental source of generative linguistic theory. They may be difficult to elicit, especially in some populations, but generally they inform us neatly about what the grammar licenses or, on the contrary, bans. On the other hand, acceptability is multifactorial and therefore, unlike grammaticality judgement, can be quantified. In this paper I consider a particular empirical domain, that of Relativized Minimality (RM) in acquisition, and its relation to the dichotomy between grammaticality and acceptability. Friedmann et al. (2009) argued that children hold a stricter version of RM than adults. In particular, children would require a disjoint feature specification, not just a distinct feature specification, between target and intervener. The literature shows asymmetries in comprehension of subject and object relative clauses in various languages which fulfill the predictions of child RM. Variation between adults and children might be expected not only in production and comprehension, but also in grammaticality judgement. If so, children would be predicted to reject object relatives Edited by: as well as the classic RM violations. Alternatively, if child RM is a processing effect, the Urtzi Etxeberria, prediction is that children would be able to tease apart object relative clauses from RM Centre National de la Recherche violations under favorable processing conditions. The question I address is: do children Scientifique (CNRS), France assimilate RM violations and object relative clauses? Grammaticality judgement should Reviewed by: Cristiano Chesi, provide an answer to this question. In this paper I present an experiment targeting University Institute of Higher Studies in grammaticality judgement for object relatives and RM violations and report preliminary Pavia, Italy Evelina Leivada, results for a group of 6-year-old Catalan-speaking children showing that object relatives University of Rovira i Virgili, Spain and RM violations are not judged in a parallel fashion, since RM violations are rejected *Correspondence: more often than object relatives. Anna Gavarró anna.gavarro@uab.cat Keywords: grammaticality judgement, processing, child grammar, object relatives, Relativized Minimality violations, Catalan Specialty section: This article was submitted to Language Sciences, INTRODUCTION a section of the journal Frontiers in Psychology The literature on language acquisition has attested an asymmetry in the comprehension and production of relative clauses, object relative clauses lagging behind subject relative clauses (see, Received: 31 July 2019 Accepted: 15 January 2020 for English, Brown, 1971; De Villiers et al., 1979; for French, Labelle, 1990; for Portuguese, Corrêa, Published: 07 February 2020 1995; for Spanish, Pérez-Leroux, 1995; for German, Adani et al., 2013; for Italian, Contemori Citation: and Belletti, 2014, etc.). Friedmann et al. (2009) proposed a new analysis for this well-known Gavarró A (2020) Child Relativized asymmetry: subject and object relative clauses differ in the position from which the wh- constituent Minimality and Grammaticality moves, and they argued that children apply a stricter constraint on A’ movement than adults that Judgement. Front. Psychol. 11:105. renders object relatives (under specific conditions) difficult for them. In this report I explore a doi: 10.3389/fpsyg.2020.00105 prediction of Friedmann et al.’s hypothesis if one assumes that this stricter constraint on movement Frontiers in Psychology | www.frontiersin.org 1 February 2020 | Volume 11 | Article 105 Gavarró Child RM and Grammaticality Judgement constitutes a truly grammatical constraint (as opposed (2009, p. 85) words, “a configuration [like that in (6)] is to the result of a processing limitation): grammaticality disallowed as it violates the ‘strong’ RM requiring featural judgement, then, should yield the same pattern as production disjointness.” If object relatives2 are assimilated to RM violations and comprehension. in child grammar, the prediction is then that children will judge The report is organized as follows: in this section I detail them as equally ill-formed in a grammaticality judgement task. Friedmann et al.’s hypothesis and state the prediction to test. This prediction is put to test in the experiment described in the In section A Grammaticality Judgement Task, I motivate the next section. experimental design and give details about the experimental (5) does not exhaust all possible configurations. items, procedure and participants of the pilot study. In section In later work on featural RM effects in weak island Results, I present the results, and in section Discussion, I consider environments, Villata et al. (2016) consider the configurations them against the background literature. in (7). Following the approach of Relativized Minimality (RM) (7) a. [+Q] [+Q] < +Q> (bare identity) (Rizzi, 1990) to constraint movement, in a configuration like b. [+Q] [+Q, +N] < +Q> (inverse inclusion) (1), X and Y fail to relate if Z, the intervener, which is c. [+Q, +N] [+Q] < +Q, +N> (inclusion) structurally closer to X, can act as its antecedent because of its d. [+Q, +N] [+Q, +N] < +Q, +N> (complex identity) featural configuration. In (5) inverse inclusion (7b) was not considered, and bare (1) . . . X . . . Z . . . Y identity (7a) and complex identity (7d) fell under identity. Inverse The effect of RM can be illustrated with a classic example such inclusion is exemplified in (8a), complex identity in (8b), both as (2), in which movement of how is blocked by the intervening of them examples with intervention (taken from Villata et al., interrogative who. 2016, p. 81). (2) ∗ How do you wonder who behaved? how (8) a.Qu’est-ce quej tu te demandes quel étudianti —i a [+Q] [+Q] [+Q] résolu —j ? what is that you cl-2s wonder which student There is no RM violation in a subject relative clause (3), nor in an has solved object relative clause (4). ‘What do you wonder what student solved?’ (3) the boy that the boy hugs the monkey b.Quel problèmej te demandes-tu quel étudianti —i a résolu —j ? (4) the chicken that the cow kisses the chicken which problem 2s.cl wonder you which student Both are well-formed for adults; however, in Friedmann et al.’s has solved (2009) analysis, for children the subject the cow acts as an ‘Which problem do you wonder which student solved?’ intervener in (4); there is no possible intervener in (3)1 . This is In this paper we focus on the configurations initially considered argued to be the source of children’s delay with object relatives. in Friedmann et al. (2009) and the subsequent research on The configuration in (1) can be instantiated as in (5) [(29) in language acquisition. (Friedmann et al., 2009), p. 84]. (5) a. +A ... +A ... (identity) b. +A,+B ... +A ... < +A,+B> (inclusion) A GRAMMATICALITY JUDGEMENT TASK c. +A ... +B ... < +A> (disjunction) The experiment designed is a grammaticality judgement The example in (2) falls under the case of (5a) and is therefore task3 . Young children experience some difficulty in producing ill-formed for children and adults alike. When B is featurally grammaticality judgements, possibly because of the inability of distinct from A, as in (5c), the resulting sentence is licensed in the experimenter to transmit what the task is about, and because both child and adult grammar. Differences only emerge with the task requires some metalinguistic awareness. For that reason, (5b), where the potential intervener, +A, is characterized by a the children recruited were in the age range of 5–7 years and featural configuration that is a subset of the featural configuration not younger. of the antecedent +A+B. This corresponds to the configuration underlying object relatives like (4): 2 To be accurate, not all object relatives are problematic for children (for (6) [+R, +NP] ... +NP ... < +R, +NP> example, in Hebrew object relatives with null pronominal subjects with arbitrary interpretation, as well as free object relatives, do not give rise to intervention Adult grammar licenses (6), but child grammar treats it as a effects), and this follows from different featural specifications, i.e., they would not violation of (a stronger version of) RM. In Friedmann et al.’s fall under the configuration (5b)—see Friedmann et al. (2009). 3 The term I use is grammaticality judgement, as is customary in the generative 1 Interventionis defined structurally (in terms of c-command), not linearly. This framework, to refer to the task that a speaker performs when asked about the well- was shown clearly in the case of subject and object relative clauses in Chinese, formedness or ill-formedness of a sentence (ill-formedness being represented by which follow the asymmetry outlined above in a language in which relative clauses an asterisk diacritic); acceptability would refer to well-formedness with respect to are prenominal and therefore linear and structural intervention do not concur (see a given discourse/pragmatic context, which is not at stake. A sentence is standardly Hu, 2014). assumed to be grammatical when its derivation converges. Frontiers in Psychology | www.frontiersin.org 2 February 2020 | Volume 11 | Article 105 Gavarró Child RM and Grammaticality Judgement Materials and Methods Each of the three experimental conditions was exemplified Three sentence types were tested in Catalan: (i) object relative by 6 items, and so the total number of test items was 18 (a clauses, (ii) long distance wh- questions, and (iii) ungrammatical complete list appears in the Annex). Items were between 7 and wh- questions involving RM violations. It is worth stressing that 11 syllables long and were presented in pseudorandom order. Of the RM violations in the experiment were ill-formed and not just the 18 items, only 6 were ungrammatical for the adult speakers; degraded, as some weak island violations may be—see examples should children find object relatives ungrammatical, then 12 out (9), taken from Villata et al. (2016) and the examples in Rizzi of the 18 items would be rejected. (1990), as well as the discussion of gradations of acceptability also If children assimilate identity configurations (5a) and in relation to RM in Rizzi (2018). inclusion configurations (5b), the prediction is that they will perform equally with the two. This is what the literature on (9) a. ??Which problem do you wonder whether John could child RM has argued: children fail with object relatives when the solve (in this way)? configuration is that seen in (5b); subject relatives do not give b. ?Which problem do you wonder how to solve? rise to such a configuration, and the subject/object asymmetry The objective relative in (10) instantiates (5b), the long- follows4 . A second prediction, not stated by Friedmann et al., is distance wh- question in (11) is an instance of (5c), and the that, if the assimilation of (5b) to (5a) is operative, children will wh- question involving a RM violation in (12) instantiates judge instances of (5b) as bad as instances of (5a). This is the (5a). The featural configuration in (10) is such that the head rationale of the experiment. of the relative clause bears the features [+R,+N], and the An anonymous reviewer points out that the comparison intervening DP the feature [+N] (as assumed in Friedmann between object relative clauses and wh- extraction is far from et al., 2009). The featural configuration of the wh- questions perfect, since these two constructions have been shown to be exemplified in (11) is assumed to be [+Q] for the wh- elements quite different, so that, for example, in English, Preposition involved. Likewise in the ungrammatical question exemplified stranding is favored in indirect object wh- questions, but pied- in (12). piping is preferred in indirect object relative clauses (Bianchi and (10) Veig el gos que la nena buscava el gos. see-1s the dog that the girl looked-for ‘I see the dog that the girl was looking for.’ [+R,+N] [+N] < +R,+N> (11) Com dius que ha vingut com? how say-2s that have-3s come ‘How are you saying he came?’ [+Q] < +Q> (12) Què penses qui arreglarà què? what think-2s who repair-fut-3s ‘What do you think who will repair?’ [+Q] [+Q] < +Q> In the well-formed wh- questions, two of the experimental Chesi, 2015); in a cross-linguistic study, Sprouse et al. (2016) sentences contained dir “say” as verb selecting the embedded show that island effects are different between relative clauses and clause and six contained the verb pensar “think/wonder”; the wh- dependencies (in English, adjunct relative clauses do not same verbs (and in the same proportion) were used in the show island effects, while adjunct wh- dependencies do; Italian ungrammatical RM items. The wh- words used were all bare does not exhibit subject island effects in relative clauses, but it wh- elements, including què “what,” qui “who,” com “how” and does in wh- extraction). The reviewer suggests that a better design quan “when.” Since sentences were produced out of context would therefore include only wh- questions; this remains for and, furthermore, no complex wh- phrase was used, the effect future research. of D-linking was excluded. In the wh- questions, the wh- element corresponded to an argument or an adjunct of the embedded clause, either because it was a direct object of the Participants embedded verb, or because, as adjuncts, quan “when” and The children who participated in the study were native speakers com “how” would more naturally modify the embedded clause of Central Catalan from the extended metropolitan area of (as in When do you think you will go?). The same is true Barcelona. Twenty-five children were tested, but three were of the RM items: què “what” could only be an argument excluded because they failed to understand the task. The of the embedded clause; in the remaining cases with com remaining 22 children were in the age range of 5;05,20 to 7;04,27 “how” the adjunct would most naturally modify the embedded (mean age: 6;05). Five adults took part in the experiment as a verb, venir “come over” or portar-se “behave,” and in this control group. last case it was selected by the verb. In the wh- questions The guidelines of the Declaration of Helsinki on human and RM items no overt DP intervened between the wh- experimentation were enforced during the whole procedure and elements (subjects were null pronouns except in one case in which the overt subject was postverbal and, therefore, lower in 4 The analysis of relative clauses assumed in this literature (and here) is a raising the structure). analysis (see Bianchi, 2002a,b). Frontiers in Psychology | www.frontiersin.org 3 February 2020 | Volume 11 | Article 105 Gavarró Child RM and Grammaticality Judgement the experiment was approved by the ethics committee of the UAB The answers of all participants were recorded on an answer (CEEAH evaluation number 4,856). sheet by the second experimenter and then transcribed into an RStudio file. Procedure RESULTS The experiment was carried out individually in a quiet classroom of the children’s school. It involved two experimenters, one Adults performed as expected: they rejected all RM violations manipulating a dog puppet and uttering the target sentences, the and accepted all grammatical long- distance interrogatives and other introducing the task and questioning the child. The child object relatives. was told that the puppet was learning to speak but sometimes The total number of answers provided by the children was said things that didn’t sound right and so the child would be 396 (18 × 22), 132 per condition. The data set is freely available asked if the sentences s/he heard uttered by the puppet sounded at https://ddd.uab.cat/record/215041. Children performed as right. The experimental phase was preceded by a training phase shown in Figure 1 and Table 1, representing mean acceptance consisting of at least two items, one grammatical, another rate, standard deviation and the five number summary (order ungrammatical (Tinc molta gana “I am very hungry” vs. ∗ Molta statistics) Minimum, Q1, Median, Q3, and Maximum. tinc gana “Very I am hungry”); if necessary, the training phase If we turn to individual results, all the children rejected at least included more items. Positive feedback was given to the child one RM violation, while 10 children accepted all object relative in the experimental phase irrespective of his/her answers. The clauses. Two children judged these two sentence types identically; task took around 15 min. Adults were tested individually on the three more children judged RM violations better than object university campus. relatives. The remaining 17 children accepted object relatives more often than they accepted RM violations, tending toward the adult pattern. Individual results appear in Figure 2. Even though few children took part in the experiment, and it would be desirable to run it with more participants, some statistical analysis was undertaken. A Generalized Linear Mixed Model was used to model the number of acceptances by sentence type as a binomial response, taking into account repeated measures from each participant. The statistical analysis was obtained using R (R Core Team, 2019). Statistically significant differences were found as an effect of Sentence Type (F-Value = 63.19; p_value < 0.0001). For the RM (ungrammatical) items, the percentage of estimated acceptance responses was 47.48% (CI95% = [36.8%, 58.4%]). For the object relative items, the percentage of estimated acceptance responses was 84.22% (CI95% = [75.59%, 90.19%]). For wh- questions, the percentage of estimated acceptance responses was 93.47% (CI95% =[87.4%, 96.72%]). These results are represented in Figure 3. Pairwise comparisons of the three sentence types were all significant. There were statistically significant differences between object relatives and RM (z-ratio = 5.8; p_value < 0.0001), with higher acceptance of object relatives than RM violations (OR = 5.9, i.e., the odds ratio of acceptance of object relatives was 5.9 times the odds of acceptance of RM violations). There were marginal statistically significant differences between object relatives and wh- questions (z-ratio = −2.4; p_value = 0.0424), with higher acceptance of wh- questions (OR = 0.37, i.e., FIGURE 1 | Acceptance rate of the three sentence types, children. the odds ratio of acceptance of wh- questions was 1/0.37≈2.68 times the odds for object relatives). Finally, there were statistically TABLE 1 | Acceptance of the three sentence types, Mean, SD, and order statistics. Stype Data Mean SD Minimum Q1 Median Q3 Maximum Relative 22 0.826 0.215 0.167 0.667 0.833 1.000 1.000 RM 22 0.477 0.243 0.000 0.333 0.417 0.667 0.833 Wh- 22 0.924 0.143 0.000 0.875 1.000 1.000 1.000 Frontiers in Psychology | www.frontiersin.org 4 February 2020 | Volume 11 | Article 105 Gavarró Child RM and Grammaticality Judgement FIGURE 2 | Individual results. significant differences between RM violations and wh- questions as in adult judgements. However, there is a difference in the (z-ratio = −7.05; p_value < 0.0001), with wh- questions being acceptance rates of object relatives and wh- questions in the accepted more often than RM violations (OR = 0.0632, i.e., the judgements of children that is not found in the judgements odds ratio of acceptance of wh- questions was 1/0.0632≈15.82 of adults, albeit the difference is smaller than between any of times the odds for the RM items). these two grammatical sentence types and the ungrammatical The results so far show that object relatives and RM violations RM sentences. did not pattern alike for children: children accepted object These results are tentative; however, with the sample here the relatives at much higher rates than RM violations. Rather, object hypothesis that object relatives and RM violations are judged in relatives tended to pattern with long-distance wh- questions, the same way by children cannot be upheld. Frontiers in Psychology | www.frontiersin.org 5 February 2020 | Volume 11 | Article 105 Gavarró Child RM and Grammaticality Judgement TABLE 2 | Subject and object relative clause comprehension and production, Catalan (Gavarró et al., 2012, p. 194). Subject Object relatives relativesa Comprehension 4;06–5;06 (Mean 4;11,06) 64/66 97% 53/121 43% >5;06 (Mean 6;0,12) 60/60 100% 63/110 57% Total 124/126 98% 116/231 50.2% Production 5 (Mean 5;05,15) 98% 62.5% a The object relatives here include relative clauses with pre- and post-verbal subjects. It is beyond the scope of this report to sum up the literature FIGURE 3 | Estimated acceptance of the three sentence types. that has been carried out on relative clauses and related constructions over the years, which has led to the development of experiments manipulating Case, number, and gender features (Guasti et al., 2008; Adani et al., 2010; Belletti et al., 2012; Bentea et al., 2016; Friedmann et al., 2017), all relevant to the RM DISCUSSION hypothesis5 . Although Friedmann et al. (2009, p. 71) assert that In this section, I discuss the results in two respects: first, “the difficulty with object relatives is overcome at around the age I consider age and performance in other tasks which, by of 6 (Friedmann and Novogrodsky, 2004),” the literature on the hypothesis, relate to the one here; second, I go back to the acquisition of Romance shows that object relatives are not target- question that motivated this study, namely, does child RM stem like at age 7 (and even beyond) and so, if all of these results are from a property of child grammar, defining grammaticality, or to receive a unified account (a desirable outcome), then we can does it stem from processing limitations? assume that child RM is operative at age 7, the oldest age group First let us consider the results with respect to age. In future in this study. research more children and from a wider age range should be To my knowledge, no study so far has considered the tested; with the current sample, the five children who could child version of RM with grammaticality judgement. The be considered to conform to the parallel performance in RM general expectation is that grammaticality judgment should violations and object relatives were not amongst the youngest, align with production and comprehension, in absence of and performance appears to bear no relation to age (within the any indication to the contrary. While dissociations between limited age span here). e.g. production and comprehension are attested in language Notice that the children in this study were slightly older than acquisition, they call for an explanation. If the path of the Hebrew-speaking children in Friedmann et al. (2009), who language development is grammar-driven, the prediction is were in the age range of 3;07 to 5;0. Other studies, however, that production, comprehension and judgement will develop show that delay in the comprehension of object relatives extends in parallel. This is the assumption underlying the experiment beyond age 5;0. In a study of the acquisition of relative clauses in this report. Even though children are known to often fail in Catalan, Gavarró et al. (2012), on the basis of a picture in their production and comprehension of object relatives in identification task, found that the comprehension of object Catalan, and this is attributed to child RM in the literature, relatives was delayed when compared to the comprehension they do not judge object relatives in the same way as they of subject relative clauses. Production (elicitation based on judge RM violations. This argues against an assimilation of Novogrodsky and Friedmann’s, 2006 method) yielded very object relatives and RM violations in child grammar (that similar results [see Table 2, which summarizes the results of the is, against the grammatical assimilation of the identity and two experiments, administered to 21 children (comprehension) inclusion conditions). and 20 children (production)]. The results here are exploratory; let us suppose that children Similar results have been obtained for other languages, such do not judge RM violations in the same way they judge object as Italian. In Arosio et al. (2009), which involved 139 Italian- relatives at an age at which the child strict version of RM is speaking children of ages 5–11, object relatives with post-verbal operative, as the results so far suggest. In that case, what could subjects were miscomprehended at ages 7 and 9 (with adult the explanation be? Friedmann et al. (2009) do not discard the performance below 50%) and only at 11 was comprehension idea that child RM is the result of a processing limitation. The adult-like (see also Adani, 2010). Parallel results for object wh- 5 Thereis also work disputing the claims of Friedmann et al. (2009) (see, for interrogatives (also subsumed by Friedmann et al.’s account) example, Goodluck, 2010), and some results that the hypothesis cannot readily showed that 8- to 9-year-olds had not yet achieved adult encompass, especially from studies on topicalization (e.g., Hu et al., 2018 on performance (De Vincenzi et al., 1999; Guasti et al., 2012). Chinese)—but this is not discussed in this paper. Frontiers in Psychology | www.frontiersin.org 6 February 2020 | Volume 11 | Article 105 Gavarró Child RM and Grammaticality Judgement fact that other populations (language impaired children, patients DATA AVAILABILITY STATEMENT with aphasia) also perform differently with subject and object relatives, and healthy adults under certain circumstances may All datasets generated for this study are freely available at: https:// also display the same asymmetry (Cohen and Mehler, 1996, ddd.uab.cat/record/215041. and much subsequent work; Warren and Gibson, 2002; see Grillo, 2008) would seem to favor a processing account. In ETHICS STATEMENT Friedmann et al.’s words (2009, p. 84–85), “It may be tempting to speculate that the ban against [(5b)] in early systems may The studies involving human participants were reviewed and relate to a limitation in the operative syntactic memory: clearly, approved by the Comissió d’Ètica en l’Experimentació Animal disjointness is easier to determine, as it can be calculated feature i Humana (CEEAH), Universitat Autònoma de Barcelona. by feature, whereas calculating a subset-superset relation requires Written informed consent to participate in this study was holding in operative memory and comparing the whole featural provided by the participants’ legal guardian/next of kin. specifications associated with different positions, an operation that may exceed the capacity of the early systems.” In adults, AUTHOR CONTRIBUTIONS on the other hand, “a partial overlap of features giving rise to a configuration like [(5b)] is grammatical, but determines AG designed the experiment reported and ran it with the ‘complexity effects’ detectable in experimental work.” Under collaboration of Míriam Muntané. AG analyzed the data with the such a processing account, one could speculate that the source help of Ester Boixadera and wrote the paper. of the difference between the results here and the results in the literature on relative clause comprehension and production FUNDING are related to the experimental method. If grammaticality judgement is less costly than comprehension/production in terms The author acknowledges the following funding sources: 2017 of processing (to the extent that the interpretation of the sentence SGR 634 from the Generalitat de Catalunya, and project FFI2017- may not need to be fully accessed) then one would predict 87699-P from the Ministerio de Economía y Competitividad. that object relatives and RM violations would not be judged homogeneously by children, even under the assumption that ACKNOWLEDGMENTS child RM is operative. In addition, there is a further difference between object As well as the funding agencies mentioned above, the author relatives and RM violations, even for children: while children do wishes to thank Míriam Muntané for acting as experimenter; the produce (to varying degrees) object relatives, RM violations of Escola Maria Borés in La Pobla de Claramunt, for allowing the the kind exemplified in (2) and (12) are not attested. This may be participation of their children in the study; and the children who an indication that the configuration underlying object relatives is took part in the experiment for their enthusiastic collaboration. part of child grammar, while RM violations are ungrammatical The author thanks the two reviewers and J.M. Brucart for their for children. Grammaticality judgement can therefore provide suggestions, and Ester Boixadera, from the Servei d’Estadística a new source of evidence to characterize child RM as either a Aplicada of the UAB, for conducting the statistical analysis; any grammatical or a processing phenomenon. remaining errors are the author’s sole responsibility. REFERENCES Bentea, A., Durrleman, S., and Rizzi, L. (2016). Refining intervention: the acquisition of featural relations in object A-bar dependencies. Lingua 169, Adani, F. (2010). Rethinking the acquisition of relative clauses in Italian: 21–41. doi: 10.1016/j.lingua.2015.10.001 towards a grammaticality based account. J. Child Lang. 38, 141–165. Bianchi, V. (2002a). Headed relative clauses in generative syntax. Part I. Glot. Int. doi: 10.1017/S0305000909990250 6, 197–204. Adani, F., Sehm, M., and Zukowski, A. (2013). “How do German children and Bianchi, V. (2002b). Headed relative clauses in generative syntax. Part II. Glot Int. adults deal with their relatives,” in Advance of Language Acquisition, eds 6, 1–13. S. Stavrakaki, M. Lalioti, and P. Konstantinopolou (Newcastle: Cambridge Bianchi, V., and Chesi, C. (2015). “On a PP/DP asymmetry in extraction,” in Scholars Publishing), 14–22. Structures, Strategies and Beyond. Studies in Honour of Adriana Belletti, eds E. Adani, F., Van Der Lely, H. K. J., Forgiarini, M., and Guasti, M. T. Di Domenico, C. Hamann, and S. Matteini (Amsterdam; Philadelphia, PA: John (2010). Grammatical feature dissimilarities make relative clauses easier: Benjamins), 47–66. doi: 10.1075/la.223.03bia a comprehension study with Italian children. Lingua 120, 2148–2166. Brown, H. D. (1971). Children’s comprehension of relativized English sentences. doi: 10.1016/j.lingua.2010.03.018 Child Dev. 42, 1923–1936. Arosio, F., Adani, F., and Guasti, M. T. (2009). “Grammatical features in the Cohen, L., and Mehler, J. (1996). Click monitoring revisited: an on-line study of comprehension of Italian relative clauses by children,” in Merging Features: sentence comprehension. Mem. Cogn. 24, 94–102. Computation, Interpretation, and Acquisition, eds M. José Brucart, A. Gavarró, Contemori, C., and Belletti, A. (2014). Relatives and passive object and J.Solà (Oxford; New York, NY: Oxford University Press), 138–155. relatives in Italian-speaking children and adults: intervention in Belletti, A., Friedmann, N., Brunato, D., and Rizzi, L. (2012). Does gender make a production and comprehension. Appl. Psychol. 35, 1021–1053. difference? Comparing the effect of gender on children’s comprehension doi: 10.1017/S0142716412000689 of relative clauses in Hebrew and Italian. Lingua 122, 1053–1069. Corrêa, L. (1995). An alternative assessment of children’s comprehension of doi: 10.1016/j.lingua.2012.02.007 relative clauses. J. Psychol. Res. 24, 183–203. doi: 10.1007/BF02145355 Frontiers in Psychology | www.frontiersin.org 7 February 2020 | Volume 11 | Article 105 Gavarró Child RM and Grammaticality Judgement De Villiers, J., Tager Flusberg, H. B., Hakuta, K., and Cohen, M. (1979). Children’s study. J. Psycholinguist. Res. 47, 1279–1300. doi: 10.1007/s10936-018- comprehension of relative clauses. J. Psychol. Res. 17, 57–64. 9575-6 De Vincenzi, M., Arduino, L. S., Ciccarelli, L., and Job, R. (1999). “Parsing strate- Labelle, M. (1990). Predication, wh-movement and the development of relative gies in children comprehension of interrogative sentences,” in Proceedings of clauses. Lang. Acquisit. 1, 95–119. European Conference on Cognitive Science, ed S. Bagnara (Rome: Istituto di Novogrodsky, R., and Friedmann, N. (2006). The production of relative clauses Psicologia del CNR), 301–308. in syntactic SLI: a window to the nature of the impairment. Adv. Speech-Lang. Friedmann, N., Belletti, A., and Rizzi, L. (2009). Relativized relatives: types of Pathol. 8, 364–375. doi: 10.1080/14417040600919496 intervention in the acquisition of A-bar dependencies. Lingua 119, 67–88. Pérez-Leroux, A.-T. (1995). Resumptives in the acquisition of relative clauses. doi: 10.1016/j.lingua.2008.09.002 Lang. Acquisit. 4, 105–138. doi: 10.1080/10489223.1995.9671661 Friedmann, N., and Novogrodsky, R. (2004). The acquisition of relative clause R Core Team (2019). R: A Language and Environment for Statistical Computing. comprehension in Hebrew: a study of SLI and normal development. J. Child Vienna: R Foundation for Statistical Computing. Available online at: https:// Lang. 31, 661–681. doi: 10.1017/S0305000904006269 www.R-project.org/ Friedmann, N., Rizzi, L., and Belletti, A. (2017). No case for case in locality: case Rizzi, L. (1990). Relativized Minimality. Cambridge, MA: The MIT Press. does not help interpretation when intervention blocks A-bar chains. Glossa J. Rizzi, L. (2018). Intervention effects in grammar and language acquisition. Probus Gen. Linguist. 2:33. doi: 10.5334/gjgl.165 30, 339–367. doi: 10.1515/probus-2018-0006 Gavarró, A., Cunill, A., Muntané, M., and Reguant, M. (2012). The acquisition Sprouse, J., Caponigro, I., Greco, C., and Cecchetto, C. (2016). Experimental syntax of catalan relatives: structure and processing. Revue Roumaine de Linguist. and the variation of island effects in English and Italian. Nat. Lang. Linguist. 57, 183–201. Theory 34, 307–344. doi: 10.1007/s11049-015-9286-8 Goodluck, H. (2010). Object extraction is not subject to child relativized Villata, S., Rizzi, L., and Franck, J. (2016). Intervention effects and relativized minimality. Lingua 120, 1516–1521. doi: 10.1016/j.lingua.2009.10.005 minimality. new experimental evidence from graded judgments. Lingua 179, Grillo, N. (2008). Generalized minimality: syntactic underspecification in Broca’s. 76–96. doi: 10.1016/j.lingua.2016.03.004 (Doctoral dissertation). LOT, Universiteit Utrecht, Utrecht. Warren, T., and Gibson, E. (2002). The influence of referential Guasti, M. T., Branchini, C., and Arosio, F. (2012). Interference in the production processing on sentence complexity. Cognition 85, 79–112. of Italian subject and object wh-questions. Appl. Psycholinguist. 33, 185–223. doi: 10.1016/s0010-0277(02)00087-2 doi: 10.1017/S0142716411000324 Guasti, M. T., Stavrakaki, S., and Arosio, F. (2008). “Number and case Conflict of Interest: The author declares that the research was conducted in the in the comprehension of relative clauses: evidence from Italian and absence of any commercial or financial relationships that could be construed as a Greek,” in Language Acquisition and Development. Proceedings of GALA potential conflict of interest. 2007, eds A. Gavarró and M. J. Freitas (Newcastle: Cambridge Scholars Publishing), 230–240. Copyright © 2020 Gavarró. This is an open-access article distributed under the Hu, S. (2014). Intervention effects and the acquisition of relativization and terms of the Creative Commons Attribution License (CC BY). The use, distribution topicalization in Chinese. (Doctoral dissertation). Universitat Autònoma de or reproduction in other forums is permitted, provided the original author(s) and Barcelona and Università degli Studi di Milano-Bicocca, Milan. the copyright owner(s) are credited and that the original publication in this journal Hu, S., Guasti, M. T., and Gavarró, A. (2018). Chinese children’s knowledge is cited, in accordance with accepted academic practice. No use, distribution or of topicalization: experimental evidence from a comprehension reproduction is permitted which does not comply with these terms. Frontiers in Psychology | www.frontiersin.org 8 February 2020 | Volume 11 | Article 105 Gavarró Child RM and Grammaticality Judgement ANNEX: EXPERIMENTAL ITEMS Object relatives within transitive clauses (1–6), RM violations (7–12), long-distance interrogatives (13–18): 1.– Veig la nena que la mestra ha renyat. 2.– Veig el gos que la nena buscava. 3.– Tinc el pinzell que els nens buscaven. 4.– Veig el gat que el gos ha mossegat. 5.– Veig els pollets que la gallina buscava. 6.– Tinc el conte que els nens llegiran. 7.– Com dius qui es porta? 8.– Com dius qui vindrà? 9.– Com penses qui vindrà? 10.– Què penses com farà? 11.– Què penses qui arreglarà? 12.– Què penses qui llegirà? 13.– Com dius que ha vingut? 14.– Com dius que va a casa seva? 15.– Què penses que farem avui? 16.– Què penses que farà la senyoreta? 17.– Quan penses que farem vacances? 18.– Quan penses que aniràs a casa? Frontiers in Psychology | www.frontiersin.org 9 February 2020 | Volume 11 | Article 105