How robust are exemplar effects in word comprehension?

Abstract

This paper studies the robustness of exemplar effects in word comprehension by means of four long-term priming experiments with lexical decision tasks in Dutch. A prime and target represented the same word type and were presented with the same or different degree of reduction. In Experiment 1, participants heard only a small number of trials, a large proportion of repeated words, and stimuli produced by only one speaker. They recognized targets more quickly if these represented the same degree of reduction as their primes, which forms additional evidence for the exemplar effects reported in the literature. Similar effects were found for two speakers who differ in their pronunciations. In Experiment 2, with a smaller proportion of repeated words and more trials between prime and target, participants recognized targets preceded by primes with the same or a different degree of reduction equally quickly. Also, in Experiments 3 and 4, in which listeners were not exposed to one but two types of pronunciation variation (reduction degree and speaker voice), no exemplar effects arose. We conclude that the role of exemplars in speech comprehension during natural conversations, which typically involve several speakers and few repeated content words, may be smaller than previously assumed.Keywords: speech comprehension, exemplar effects, pronunciation variation, acoustic reduction 3 Several models of speech comprehension assume that the mental lexicon stores the pronunciation of a word with two types of representations, namely abstract representations and exemplars (e.g., Goldinger, 2007;McLennan, Luce, and Charles-Luce, 2003). Abstract representations are strings of sound symbols like phonemes or phonological features, which only contain information about acoustic properties that distinguish between these symbols. In contrast, clouds of exemplars represent many occurrences of words that the language user has uttered or heard. Each exemplar is a detailed representation corresponding to the speech signal of one occurrence and thus contains subtle acoustic information, for example about the word's exact pronunciation or the speaker's voice. Many articles in the literature point to a role of exemplars in word comprehension. This study investigates the robustness of these exemplar effects.Exemplar effects have been established in several priming experiments (e.g., Bradlow, Nygaard, and Pisoni, 1999;Craik and Kirsner, 1974;Goh, 2005;Goldinger, 1996;Janse, 2008;Mattys and Liss, 2008;McLennan et al., 2003;McLennan and Luce, 2005;Palmeri, Goldinger, and Pisoni, 1993).These experiments contained repeated words and the comprehension of the second occurrence of a word (the target) is expected to be facilitated by the first occurrence (the prime). Primes and targets were completely identical, that is the same token, or they differed in speech rate, time-compression, the realization of a certain segment (e.g., intervocalic /t,d/ produced as [t,d] or as a flap in American English), or the speaker's voice. Most experiments showed...

Combining Data-Oriented and Process-Oriented Approaches to Modeling Reaction Time Data

“…In this simulation, we first investigated the impact of the detrending parameter α on the average correlation between the observed RT sequences across all pairs of subjects, as obtained in [10]. The results are given in Figure 2.…”

“…The horizontal axis displays the range of α; the vertical axis shows the average correlation. The rightmost point α = 1 coincides with the choice in the lmer model in [10], presented in the previous section. The figure shows that, in general, the average correlation increases with decreasing value of α.…”

“…The lexical decision experiment that we take as basis for simulation is experiment 1 in [10]. Following [15], they investigated the robustness of exemplar effects in priming experiments.…”

“…In practice, all non-linear terms are discarded, so that we end up with a linear mixed effects model with subjects and word as random variables. This resulted in an lmer model with speaker (male, female), word frequency, the prefix, the prime-target match/mismatch, the duration of the stimulus, and the reaction time given to the prime as fixed effects, and subject and word as random effects [10]. All nonsignificant effects and random slopes were excluded: model1 = lmer(logRT˜speaker + logwordfreqcgn + prefix + variant_match + logdurms + logRTprime + logRTprev + (1|subject) + (1 + logRTprev|word))…”

“…The behavioral data under study in this paper come from a priming experiment using auditory stimuli in which subjects provide lexicality judgment and reaction times using prime-target pairs ( [10]). The goal of that experiment was to establish under what conditions exemplar effects become visible in the reaction times (e.g., [11,12]).…”