The decay of the lexical boost: new evidence from noun phrase priming

Speakers tend to reuse recently encountered syntactic structures, a phenomenon termed structural priming (Mahowald et al., 2016; Pickering & Ferreira, 2008). Priming is enhanced when prime and target sentences share the head of the phrase that is primed, e.g., the verb (Pickering & Branigan, 2008). This is referred to as the lexical boost. An important question has been whether abstract structural priming and the lexical boost differ in their temporal properties; that is, whether they decay at the same rate. If their time course differs, it suggests that they have different origins. So far, research has found that whereas abstract priming is long-lasting, the lexical boost decays fast: it disappears after just one intervening sentence (Branigan & McLean, 2016; Hartsuiker et.al., 2008; Mahowald et.al., 2016). However, in previous research that introduced intervening sentences (Branigan & McLean, 2016; Hartsuiker et.al., 2008), the verbs in primes and targets were often similar in meaning even when they were not repeated (e.g., give and hand; push and pull). This may have reduced the lexical boost. They were also often repeated across items, potentially interfering with the boost from a repeated verb in the prime. In addition, the primes were comprehended rather than produced, which might have decreased memory for the verbs. In our study we tested priming in noun phrases with either a prenominal modifier (AN structure, e.g. 'the underlined cat') or a relative clause as a post-nominal modifier (NRC structure, e.g. 'the cat that is underlined'). We made sure that (i) the nouns differed in meaning in the conditions where they were not repeated, (ii) no nouns were repeated across items and (iii) all noun phrases were read out loud. In Experiment 1, 48 native English speakers completed AN or NRC prime fragments out loud (e.g. 'the underlined ... ' or 'the ... that is underlined' respectively), each followed by completion of a target fragment which accompanied a picture (e.g. an image of a cat that is underlined). The targets either directly followed the primes (Lag 0) or were separated by two intervening fillers (Lag 2). The target noun was either the same as in the prime or different. The mixed effects analysis revealed an overall priming effect (p < .001). Prime structure interacted with lag (p < .001), but priming was found in both Lag 0 (p < .001) and Lag 2 (p < .001) conditions. An analysis of the conditions without noun repetition showed that structural priming was significantly stronger in Lag 0 than Lag 2 (p < .001). The prime structure x noun repetition interaction (lexical boost) was marginally significant (p = .093, Table 1), but here was no interaction between lexical boost and lag (p = .80). The absence of any interactions with noun repetition may have been due to a lack of power. We therefore ran Experiment 2 with 56 participants in the Lag 2 conditions only. The results showed a priming effect (p < .001), but no lexical boost (p = .803, Table 2), suggesting that there is no boost across two intervening utterances, but in the absence of noun repetition, structural priming remains. Overall, these findings suggest that while abstract structural priming is long-lasting, there is yet no evidence that the lexical boost persists across two intervening sentences, consistent with the idea that they have different origins (Branigan & McLean, 2016, Hartsuiker et al., 2008, Malhotra et al., 2008).