The influence of animacy on children's online processing of restrictive relative clauses.

Macdonald, R. G., Serratrice, L., Brandt, S.,Lieven, E, &  Theakston, A. The influence of animacy on children’s online processing of restrictive relative clauses. Paper presented at the 3rd annual LuCiD Language and Communication Development Conference, Lancaster, UK.


Subject relative clauses (SRCs, “the deer that is chasing the cow”) are typically processed more easily than object relative clauses (ORCs, “the deer that the cow is chasing”), but this difference is diminished by the presence of an inanimate head-noun in ORCs (“the tractor that the cow is chasing”) (Mak, Vonk, & Schriefers, 2002). We investigated the influence of animacy on children’s online processing of SRC and ORC sentences. Forty-eight English-speaking children (aged 4;5–6;5) and 32 adults listened to sentences that varied in the animacy of the head-noun (Animate/Inanimate) and the type of relative clause (RC) used (SRC/ORC) (Table 1). Concurrently, participants saw two images depicting the same two agents, carrying out reversed actions (e.g. deer chasing cow/cow chasing deer, Figure 1) and were asked to choose the picture matching the sentence using a game-pad. We hypothesised that children would find ORC sentences more difficult in the animate condition than the inanimate condition. Participants’ eye-movements were monitored to investigate online processing as a RC unfolds. Specifically, we focussed on anticipatory fixations after the onset of the RC (“that…”). We predicted more anticipatory looks to the picture matching a SRC-sentence in the animate condition compared to the inanimate condition.

Both child and adult participants were quicker to respond to SRC sentences and children were more accurate with SRCs (adult performance reached ceiling). As expected, children were significantly more accurate with ORCs with an inanimate head-noun rather than an animate head-noun, but animacy had no effect on the response time for ORCs.

Surprisingly, for SRCs, after the onset of the RC (“that…”) children made more looks more quickly to the target in the inanimate rather than animate condition (Figure 2), suggesting greater anticipation for SRCs with inanimate head-nouns. Adults showed no preference for SRCs in the animate condition but they did in the inanimate condition, although this preference emerged earlier than it did with the children. These results appear counter-intuitive given children’s performance in the forced-choice selection task. Children were more accurate with ORC sentences in the inanimate condition, yet during the RC they looked at the target less in this condition. The seemingly increased anticipation for SRCs in the inanimate condition may be due to surprisal at inanimate objects acting on animates, resulting in the inanimate SRC-image (e.g., the tractor chasing the cow) capturing more attention during the RC (or earlier in the case of adult participants). Alternatively, it may be due to the inanimate objects being more distinct from their animate competitors, making the SRC-image easier to identify more quickly, leading to earlier looks to this image. We are currently investigating these possibilities by repeating our experiment with animate-inanimate and inanimate-inanimate pairs. If surprisal at inanimate agents led to our eye movement results, we would again expect more looks to the inanimate SRC-image, however if the effects were driven by the animate-inanimate contrast we would expect more looks to the animate SRC-image. Regardless of the cause, our results show children’s anticipatory fixations at RC-onset do not consistently predict performance.