Pied-piping in embedded contexts in the acquisition of English: A qualitative study of comprehension

The alternation between pied-piping and preposition-stranding offers a window into the interaction between phrase and clause structure and movement (Hornstein and Weinberg 1981, Chomsky 1995, Suner 1998). As Cowper (1987) notes, pied-piping may interact differently with main and embedded clause structure. In English, the alternation is also subject to the subtle variation across embedded clause types seen in (1)-(2), where wh-forms, prepositions, and verbal inflection all interact. (1) Tensed relative clause a. Eeyore pushes the boat [ behind which Pooh runs t ] b. Eeyore touches the boat [ Op Pooh runs behind t ] c. Eeyore touches the boat [ which Pooh runs behind t ] (2) Infinitival relative clause a. Pooh picks the blanket [ under which to rest t ] b. Pooh picks the blanket [Op to rest under t ] c. *Pooh picks the blanket [which to rest under t ] In the pied-piped examples in (1-2a), the preposition accompanies the wh-form in clause-initial position. Both tensed and infinitival relative clauses permit a null operator in preposition-stranding structures, as in (1-2b), but the two clause types are distinguished in whether they permit an overt operator with preposition-stranding, as in (1c), or not, as in (2c). Inflection of the embedded verb thus interacts with clause structure and operator syntax in the domain of pied-piping. The results reported in this paper are part of a larger project investigating the acquisition of piedpiping and preposition-stranding across embedded contexts. In the production component of the larger project (Foley 1998, Foley and Fugett-Fuller 2002), 80 children acquiring English (5;6-9;5) and 20 adults were tested in an elicited imitation experiment including tensed and infinitival relative clauses and embedded questions. The experiment tested the hypothesis that if pied-piping and preposition-stranding truly interact with both verbal inflection and clause structure, children’s production might distinguish pied-piping across embedded clause types. Pied-piping was imitated with less success overall than preposition-stranding, consistent with earlier work (French 1984, McDaniel and McKee 1996, McDaniel, McKee and Bernstein 1998). However, both the quantitative results and qualitative analysis of errors revealed that children distinguish pied-piping across the three embedded types. For example, the difference in percentage of “correct” imitations across pied-piping and preposition-stranding was much larger for infinitival relative clauses than for the two relative clause types (quantitative), and the specific types of structural changes children made differed across embedded clause types (qualitative). The experiment reported here explores the comprehension of both pied-piping and preposition stranding in language acquisition for two of the embedded clause types tested in the production study: tensed and infinitival relative clauses. (Because we used an act-out method, discussed below, the design could not easily include embedded questions.) This initial, exploratory study included a fine-grained qualitative analysis of a set of responses from ten children and ten adult control subjects.