Speech

Language control refers to the ability to correctly understand or to correctly produce a word (in the correct language) in a given situation. In a series of experiments, we demonstrate modality-specific influences on language processing and language control – specifically in settings in which the sensory modality (i.e., auditory vs. visual input) and/or the motor modality (i.e., verbal vs. manual output) could repeat or switch from one trial to the next. We will summarize our findings on modality-specific influences in two different lines of research. First, we focus on the compatibility between sensory modality and motor modality in language processing. Here, our data show higher switch costs (i.e., worse performance in switch trials compared to repetition trials) when switching between incompatible sensory-motor modality mappings (auditory/manual and visual/vocal) than when switching between compatible sensory-motor modality mappings (auditory/vocal and visual manual). Second, we discuss differences in modality-specific influences on language production and language perception tasks in language switching. In these experiments, bimodal language switching (switching between two languages in two different modalities) was compared to unimodal language switching (switching between two languages in the same modality). While bimodal compared to unimodal switching can, under specific conditions, lead to a decrease of switch costs in language production, it increases switch costs in language perception tasks.

The picture-word interference (PWI) task is frequently used in cognitive psychology to study the processes underlying speech production. In this task a speaker names target pictures whilst processing aurally or visually presented distractor words. It is a robust finding that speakers take longer to name the target whilst processing a semantically related compared to an unrelated distractor. Yet, in day-to-day conversation, interlocutors routinely produce semantically related words. To investigate whether semantic interference is also observed in social interaction we embedded a PWI task in a communicative setting: Two participants played a card game during which one named the distractor and, after a stimulus-onset-asynchrony (SOA) of either -150ms or -650ms, the other named a semantically related or unrelated target. In Exp.1 (N=32) target naming latencies did not differ between semantic conditions, at neither SOA. In Exp.2 (N=32) we embedded audio recordings of the distractor naming obtained from Exp.1 in a classic single-subject PWI setting. In this setting we observed, as expected, semantic interference at SOA -150ms. In Exp.3 (N=32) the card game encouraged a focus on the semantic relationship between distractor and target. As in Exp.1 we did not observe semantic interference at SOA -150ms. However, at SOA -650ms we observed semantic facilitation. We conclude that the processes leading to semantic interference in single-subject settings are attenuated in communication. Depending on communicative context and inter-turn interval semantic relatedness between the partner’s and the own utterances either proves irrelevant, or facilitates speech production.

The lexical representation of compounds is still under debate, and research on the topic is inconsistent. Some studies suggest that compounds have a single entry at the lemma level (e.g., Lorenz et al., 2018), while others argue for separate lemma representations for the compound’s constituents in addition to the compound lemma (Marelli et al., 2012). Our study exploits the cumulative semantic interference effect (CSIE) to investigate the lexical representation of compounds. We conducted a continuous picture naming experiment, in which participants were presented a seemingly random sequence of objects for spoken naming. Previous studies have shown that naming latencies increase with each additional member of the semantic category presented within this sequence. This CSIE is assumed to reflect semantic interference during lexical access (Howard et al., 2006). Category membership in our experiment was established through the compounds` first constituent (category animals: dog lead, zebra crossing, pony tail, mouse trap, cat litter), while the compounds themselves were not semantically related. Additionally, pictures depicting the compounds’ first constituent (dog, zebra, pony, mouse, cat) were presented as a control condition. The results reveal a CSIE in the control and the compound condition, showing that the semantic relationship between the compounds’ first constituents influenced compound production. This might be interpreted as evidence for separate lemma representations for the compounds’ constituents as suggested by the multiple-lemma representation account of compounds (Marelli et al., 2012). This and alternative explanations will be discussed in light of the complexity of the paradigm.

We investigated the production of nominal compounds (tablecloth, lipstick, sunflower) to test models of speech production and lexical representation. Young (18-35 years) and older (65 + years) speakers named pictures of objects with compound names in the presence of morphological, semantic, and unrelated distractor words in a picture-word interference paradigm. Event-related potentials (ERPs) were analysed in addition to picture-naming latencies. In both age groups, constituent distractors of compound targets (lip or stick for the target LIPSTICK) speeded compound naming, while naming was slowed by distractors from the same semantic category as the compound (powder - LIPSTICK). No effects were obtained for distractors from the same category as the compound’s first constituent (toe - LIPSTICK). Age-related effects were observed in morphological distractor conditions, but not in semantic conditions. This is in line with prolonged/ deficient morpho-phonological encoding with age. Naming latencies confirmed weaker effects of second- than first-constituent morphological distractors in the elderly, whereas young speakers showed no such difference. Furthermore, the transparency of the semantic relation between morphological distractors and targets affected ERP effects, again only for older speakers. For the elderly, stronger ERP effects were obtained with higher semantic transparency of second-constituent distractors, which is likely to reflect enhanced semantic processing. Interestingly, ERP effects of first and second-constituent distractors were present in overlapping time windows, suggesting parallel processing of morphemes in both age groups. Our data corroborate single-lemma, but multiple-morpheme representations for compounds in production, which are stable across adulthood.