ABSTRACT
In this chapter, I analyze data from a sentence-completion priming experiment on the dative alternation involving advanced German learners of English and triangulate them with verb-specific constructional preferences (as determined by a distinctive collexeme analysis) from corpus data of native speakers of English. In a first step, the experimental data—data originally collected by, but very much underanalyzed—are analyzed with a generalized linear mixed-effects model to determine whether the learners exhibit priming effects and, if so, what their nature (direction and effect size) might be. In a second step, verb-specific constructional preferences from a previous collostructional analysis are newly computed using an association measure more useful for a comparison with sentence-completion data and then related to the results of the experimental data. The results show that the learners do exhibit priming effects and that their sentence completions are highly correlated with the constructional preferences of the dative-alternation verbs in native speaker data. In addition, the analysis returns results that raise a variety of caveats regarding experiment-only analyses of priming; examples of such caveats discussed here include within-experiment learning effects, input distributions that are biased with regard to verb-construction co-occurrence and prime-target similarity, and the absence of surprisal effects.
