We conducted structural priming experiments in which participants first read aloud and completed a prime fragment using pictures, followed by a target fragment. We manipulated whether the structure in the prime consisted of an adjective followed by a noun (e.g., "the underlined ..." presented with a picture of an underlined cat or crown) or a noun followed by a relative clause (e.g., "the ... that is underlined" with a picture of an underlined cat or crown), as well as whether the noun in the prime and target was the same (cat) or different (crown and cat). Participants could complete the target fragment using either an adjective-noun or noun-relative clause structure (e.g., "The ..." presented with a picture of a highlighted cat). In Experiment 1, we also manipulated whether the target immediately followed the prime or there were two unrelated intervening sentences between the prime and target. In Experiments 2 and 3, there were two intervening sentences in all conditions. Experiments 1 and 2 had a single prime, whereas Experiment 3 had two primes that both had the same structure and same noun. The first of these primes was a complete noun phrase that participants read aloud, either an adjective-noun or noun-relative clause structure, followed by a prime and target fragment with pictures. All experiments were conducted online using the experiment software Gorilla.Previous research on structural priming suggests a different time course for lexically independent priming (structural priming in the absence of word repetition between prime and target) and the lexical boost (enhanced priming when a word is repeated): Lexically independent structural priming is relatively long lasting, whereas the lexical boost decays rapidly when sentences intervene between prime and target. This research investigated either ditransitive structures (Hartsuiker et al., 2008) or actives/passives (Branigan & McLean, 2016). In contrast, we tested noun phrase structures in our experiments. Participants first produced either an adjective-noun prime structure (e.g., “the underlined crown”) or a noun-relative clause prime structure (e.g., “the crown that is underlined”) and then had to produce a target noun phrase using pictures (e.g., “the highlighted cat” or “the cat that is highlighted”). In Experiment 1, we manipulated the prime structure, whether the noun was the same in prime and target, and the number of intervening sentences between prime and target (0 or 2). Lexically independent priming was stronger without than with intervening sentences, but the lexical boost effect was not fully significant and did not interact with the number of intervening sentences. Because the large number of conditions may have resulted in low statistical power, we only tested conditions with 2 intervening sentences in Experiment 2. This showed a significant effect of lexically independent structural priming, but no lexical boost, suggesting that lexically independent priming lasts across intervening sentences, but the lexical boost does not. Experiment 3 was the same as Experiment 2, but participants produced two prime sentences. We now observed both significant lexically independent priming and a lexical boost, indicating that the lexical boost can occur across intervening sentences if there is sufficient priming input.

Behavioural experiment. Participants were recruited using Prolific and tested using Gorilla. They were all native speakers of English, residents of the UK, had no language- or literacy-related disorders and were between 18 and 35 years of age. Forty-eight participants took part in Experiment 1, and 56 in Experiments 2 and 3. The study was approved by the University of Dundee ethics committee and all participants gave informed consent to take part in the study. In all experiments, participants first completed a prime sentence fragment using pictures and then completed a target fragment with pictures.