The effects of teaching parsing strategies on the processing and learning of L2 English relative clauses

The effectiveness of explicit training on the second language (L2) offline comprehension and production has been demonstrated by numerous studies. Yet only a few studies have investigated training effects on online processing, and the findings to date are inconclusive. The current study investigated the effects of teaching explicit knowledge with input-based practice on L2 English relative clauses learning. In addition, how native English speakers process and produce relative clauses was also investigated. 79 Chinese-speaking L2 learners and 21 native speakers were involved in the study. The L2 learners were randomly assigned to either a parsing (n=27), input flood (n=26), or test-only (n=26) group. The tests measured offline (aural sentence-picture matching) and online (self-paced reading and visual-world eye-tracking) comprehension, oral production (picture description), and metalinguistic knowledge (written sentence-picture matching with explanations), administrated in a pre-, one-week post- and three-week delayed post-test design. The L2 learners attended all test phases, and the native speakers only attended the pre-tests. Between the pre- and post-test, the parsing (parsing strategies with explicit information and practice) and the input flood (exposure to the target structures) group received two 30-minute training sessions. The results from the native speakers demonstrated the expected asymmetry between subject relative clauses (SRC) and object relative clauses (ORC). The results from the L2 learners suggested that the explicit training developed offline comprehension, production and metalinguistic knowledge. Very small improvements were also found in one online measure, but only for SRCs. The input flood group did not have significant gains in any measures, relative to the test-only group. The findings contribute to our understanding about the effects of explicit training on online and offline processing, production and knowledge of syntax. Following the small amount of practice, effects of explicit training seemed to be reliably observable on offline measures, potentially observable on online measures for a structure that is easier to process (SRCs), and not likely to be observed on online measures for a structure that is more difficult to process (ORCs)