Structural diversity does not affect the acquisition of recursion: The case of possession in German

ABSTRACT

Languages vary according to which morphosyntactic forms of embedding are present in the grammar as well as to which of these forms allow recursive embedding. The present study examines how German-speaking children discover which forms of embedding are recursive. In German, possessive modifiers are expressed by several structural options (i.e., genitive case, possessive -s, relative clauses, and von-prepositional phrases, placed to the left or the right of the possessum), of which only some are recursive. In contrast, other forms of phrasal noun modification are more homogeneously realized as two basic structures (right branching PPs or relative clauses), both recursive. We examine whether recursive possessives are delayed in German L1 acquisition compared to other forms of recursive modification. A referential elicitation task tested 5-year-olds’ (n = 21) and adults’ (n = 22) production of recursive modification of possessives, comitatives, locatives, and relational structures. Overall, production of recursive possessives is not inhibited by structural diversity, relative to the other conditions. Children’s target responses to the possessive condition differed from adults’ in that children reduced the inventory of structural types and relied more commonly on certain forms that adults used less frequently. These results indicate that structural diversity does not delay children’s mastery of recursive expressions in a given domain and that structural complexity can determine the overall timing of the onset of recursive modification, but this fails to help explain performance across domains or the actual options children select.

Acknowledgments

We are grateful to Tom Roeper, Charles Yang, Helmut Weiß, Merle Weicker, Helen Engemann, Ailis Cournane, Olivia Marasco, and two anonymous reviewers as well as Kriszta Szendroi for their comments and suggestions.

Notes

¹ See also Weicker (2019) for a discussion of complexity in the acquisition of semantics.

² At the same time, superficially different constructions may well be acquired at the same time if the underlying requirements are the same. Paired and triple wh-questions, for example, differ in length and number of wh-words (Who is eating what? vs. Who is giving what to whom?) but involve the same underlying properties (mapping and exhaustivity, see Schulz & Roeper 2011). Triple wh-questions are acquired simultaneously with paired wh-questions (Schulz 2015).

³ As we focus on phrasal recursion, we will not discuss these lexical forms further.

⁴ The difference is visible for feminine nouns. Compare the Saxon -s structure Mutters Auto ‘mother’s car’ with the genitive structure das Auto der Mutter (the-NOM car-NOM the-GEN mother-GEN) ‘the car of the mother,’ which does not contain an -s on Mutter (Rauth 2014).

⁵ These morphological constraints on -s are not well understood. H. Weiß (2008) suggests that only kinship terms with -r stems can take the -s, which could explain the pattern in (3). Here we leave open whether Saxon -s leads to a reading as a proper name but note that it is also attested with relational nouns such as Nachbars ‘neighbor’s’ in (i), which is distinct from the genitive case form Nachbarn. And neighbor does not refer to a specific person in (i):

(i) Nachbars Bäume sind tabu. ‘neighbor’s trees are forbidden’

(https://www.welt.de/welt_print/article3917547/Nachbars-Baeume-sind-tabu.html)

⁶ Note that in principle the -s in Peters could also mark genitive case.

⁷ The possessor DP dem Mann refers to the possessor; the possessive pronoun sein marks the possessive relation (Weiß 2008).

⁸ A reviewer suggested that the ungrammaticality of (5a) is due to neighbor not being a kinship term. However, the recursive variant seems to be acceptable for some speakers with kinship terms, see (i) from Haider (1988:56). Also see footnote 5.

(i) der Mann, dessen Mutters Schwesters Kind ich kenne, …

‘The man whose mother‘s sister’s child I know’

⁹ Note that ihm sein Buch (him-DAT his-NOM book-NOM) ‘his book’ is grammatical. In this case the structure is not recursive; it is the same as Dem Mann sein Buch ‘the-DAT Man his book’ (Table 2, example d), ‘the man’ doubles the pronoun ihm ‘him’ (Weiß 2008).

¹⁰ Wer ‘who’ and was ‘what’ can be used as relative pronouns, but their use is restricted to certain contexts and dialects (e.g., Fleischer 2004).

¹¹ Note that in three of the categories involving relative clauses (i.e., Possessum-RC-PP, Possessum-RC-von, and Possessum-RC-s) the second modifier is embedded within the relative clause; in all other categories the two instances of embedding appear adjacent to each other.

¹² Recall that -s is basically restricted to proper nouns (see Section 3.1.1., example 3). Four out of seven possessive trials targeted a proper noun.

¹³ Except for cases such as das Buch Peters (the book Peter’s), see Table 1, variant e.

¹⁴ Japanese uses the same markers in recursive structures (the particle -no and relative clauses) to express our four notional domains, yet it also exhibits the POSS,COM > REL,LOC generalization; Pérez-Leroux et al. (to appear).

Additional information

Funding

This research was supported by a grant from the Social Sciences and Humanities Research Council of Canada to A. T. Pérez-Leroux and Y. Roberge (SSHRC 435-2014-2000 Development of NP Complexity in Children). Active exchanges between the Toronto and Frankfurt research teams were made possible thanks to funding from the Germany Europe Fund and Goethe Partnership at the University of Toronto.