Considering the whole paradigm: Preschoolers’ comprehension of agreement is not uniformly late

ABSTRACT

Many languages encode phi-features via overt morphology, yet children’s use of this morphology in comprehension tasks varies widely. Here, we use a picture-selection task to test comprehension of Spanish verbal agreement and clitics, comparing performance across and within each paradigm to examine the effect of two factors: (i) phonological salience, and (ii) semantic (under)specification. Both paradigms encode the same person and number features, but clitics may be easier to comprehend than agreement because they carry more phonological material. Within each paradigm, first- and second-person morphology may be easier to comprehend than third-person because they carry an explicit person feature. We find limited support for phonological salience and stronger support for semantic (under)specification. However, we also find evidence for a third factor affecting interpretation of third-person morphology: discourse prominence. Both adults and children permit third-person agreement and clitics to refer to the speaker and/or addressee if they have been mentioned in the immediately preceding context.

Acknowledgment

We would like to thank the members of the Michigan State University Language Acquisition Lab, in particular Alan Munn, Jessica Gamache, Ni La Le, and Camila Alfonso. Many thanks to our adult participants at MSU and to the teachers, parents, and children at Servicios Educativos del Desarrollo Infantil, Mexico City, D.F., as well as Beti López Juárez and Patricia de la Fuente Zuno. Thanks to the audience at BUCLD39 for insightful feedback.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 This study excludes 2nd person plural agreement and clitics because they are subject to a rather complex pattern of regional variation, with some dialects differentiating between formal and informal register and others not (see Lipski 1994 for details). In Mexico City Spanish, the variety studied here, there is no formality distinction in the 2nd person plural, and this creates syncretism with the 3rd person plural. In subject position, the pronominal subject ustedes triggers the same agreement as 3rd person plural (–n). In object position, the 2nd person plural accusative clitic is syncretic with 3rd person plural los (masculine) or las (feminine). To avoid the ambiguity caused by this syncretism, we do not include a 2nd person plural condition.

2 In the third person, clitics additionally encode gender.

3 These authors assume that preverbal subject clitics have the status of agreement markers. Although not all authors agree with this analysis (e.g., de Cat 2002), there seems to be evidence for it applying to modern spoken French (see Culbertson 2010; Legendre et al. 2010b). For ease of exposition we will describe results as if these forms are agreement markers, as Legendre et al. (2014) assume to be the case.

4 Note that Childers et al. (2001) find the opposite asymmetry in Chilean Spanish for comprehension of present progressives está Xndo versus están Xndo ‘is Xing’ vs. ‘are Xing.’

5 More precisely, these features introduce presuppositions that restrict the denotation of whatever referring expression they accompany, returning a value if their requirements are satisfied and returning nothing if they are violated (e.g., Sauerland 2003; Charnavel 2019, and many others).

6 A mixed-effects logistic regression model failed to find a significant effect of item order on children’s performance within either the agreement block (β = 0.01, SE = 0.01, z = 0.49, p = .62) or the clitic block (β = −0.01, SE = 0.01, z = −0.96, p = .34). We therefore assume that any differences between blocks are due to differences between agreement and clitics rather than to the order in which the blocks were presented.

7 Random effects for item were not included, since agreement and clitics were tested using different items.

8 It is our assumption that the referent has the highest status “in focus” as opposed to the next-highest status “activated.” Gundel, Hedberg and Zacharski (1993) claim that mentioning a referent places it in short-term memory (elevating it to “activated” status), but they do not explicitly discuss pointing as a way to bring it into the current focus of attention (further elevating it to “in focus”). Nevertheless the important distinction is that this referent has a cognitive status that is equal to or higher than that of the target referent—which crucially has not yet been mentioned at the point when the clitic or agreement marker is uttered.

9 These data are taken from the 14 experimental trials that were immediately preceded by a filler and excludes the 14 trials preceded by a distractor and the two trials appearing at the beginning of each block (see section 4.2.)

10 We thank two anonymous reviewers for thoughtful comments leading to a substantial reframing of these results.

11 Although both this study and these Principle B studies both suggest that discourse-pragmatic considerations can override more strictly linguistic restrictions, it is important to acknowledge the limits of this suggestion. There is still a lot we don’t know about how children (and adults) rank different discourse and syntactic properties when choosing referents or how this ranking differs across languages. What we know so far from a variety of processing studies is that the choice of an antecedent for a pronoun depends on (i) properties of the potential antecedents, (ii) properties of the pronominal element itself (null pronoun vs. overt pronouns vs. demonstrative, etc.), and (iii) their respective syntactic positions, among other considerations (see Miltsakaki 2002; de la Fuente 2015; Forsythe 2018 for reviews of this literature).

Additional information

Funding

This research was funded in part by a Tinker Field Research Grant from the MSU Center for Latin American and Caribbean Studies to Hannah Forsythe, a NSF Postdoctoral research grant [#SPRF1810159] to Hannah Forsythe, and a NSF Grant [#BCS-1656133] to Cristina Schmitt.