The Mental Lexicon - Volume 16, Issue 2-3, 2021

To explore how processing lexicality may change with aging and in the presence of Alzheimer’s disease (AD), we conducted two experiments investigating lexicality judgements using an on-line behavioural psycholinguistic methodology and electrophysiological/event-related potential (ERP) methods; oddball lexical decision tasks. Results from these lexical decision tasks showed that while those with AD show similar rates of accuracy for their lexical decision as compared older adults (OA), they are particularly slowed when making judgements for pseudowords. Our results from the ERP tasks also showed that the two groups behaved differently with regard to elicitation of the P3 ERP response, which indicates differences in how these two groups form lexical categories. The pattern of ERP responses suggests that older adults are sensitive to the orthography/phonology of the stimuli during the course of lexical processing as compared to participants with AD who show less sensitivity to orthographic/phonological cues. Additionally, the ERP P3 amplitude results suggest further linguistically related differences between healthy older adults and those with AD, and highlight the importance and usefulness of combining behavioural psycholinguistic and ERP methodologies.

The role of morphophonology during Chinese visual word recognition was investigated in three masked priming lexical decision experiments. The primes and targets were Cantonese Chinese bimorphemic words. In Chinese, most characters correspond to morphemes, but sometimes the mapping between character and morpheme is not one-to-one. Specifically, some characters are heteronymic, which had one visual form associated with multiple pronunciations and meanings (e.g., “長” is pronounced as /coeng4/ and /zoeng2/ in Cantonese, which means “long” and “senior”, respectively). In Experiment 1, facilitative priming was found when the primes and targets shared heteronymic characters of identical (e.g., “長遠-long term” /coeng4jyun5/ and “長短-length” /coeng4dyun2/), but not a different, pronunciation (e.g., “長官-senior official” /zoeng2gun1/). Sharing word-level phonology only (e.g., “場景-scene” /coeng4ging2/) had no effects. Experiment 2 replicated Experiment 1, and further indicated that the effects could not be attributed to sharing word-level meanings (e.g., “即時-immediate” /zik1si4/). Experiment 3 compared the priming effects produced by the two alternative pronunciations of the heteronymic characters. The results showed that the strength of priming was statistically comparable in the two pronunciation-congruent conditions. Together, this study provided evidence that morphophonology was activated to facilitate the ambiguity resolution of heteronymic characters. The lemma model was modified to accommodate the results.

Valid and reliable measurements of response latency are crucial in testing empirical predictions across fields of psychology. In research utilizing verbal responses, acoustic latency is the typical measure of response latency, but its validity has been questioned. We describe a simple and affordable alternative – articulatory latency based on tracking lip position. Using this method, we measured the acoustic and articulatory latencies of syllables beginning with various simple and complex onsets and ending with “uh” using the speeded naming task, where participants were instructed to have their mouths either closed or open before articulating. The initial oral configuration, place of articulation, and voicing all had significant effects on this measure of articulatory latency across segments, factors that researchers must consider in designing experiments and selecting stimuli.

The present study explored the conditions under which phonological similarity effects arise without orthographic confounds by testing languages with true cognates but divergent scripts. We investigated the similarities and differences between within- and cross-script processing patterns by providing data from an understudied language pair, Korean and Cantonese, which have many cognates but bear no orthographic resemblance. In two word-naming and translation tasks, beginning and intermediate Cantonese-speaking learners of Korean (N = 112) were tested for the processing speed of Sino-Korean words. The results of the word-naming experiments showed that phonologically similar words were processed faster than dissimilar ones, regardless of L2 fluency, especially when the logographic L1 characters were used as primes. However, facilitation by shared phonology was not observed in the translation experiments in either direction. L1-to-L2 forward translation was much faster than L2-to-L1 backward translation, indicating conceptual memory being used as a primary processing pathway. The characteristics of cross-script processing patterns were discussed in terms of the structure of bilingual memory.

Collocations are words associated because of their frequent co-occurrence, which makes them predictable and leads to facilitated processing. While there have been suggestions that collocations are stored as unanalysed chunks, other researchers disagree. One of the arguments against holistic storage is the fact that collocations are not fixed phrases, for example, their word order can vary. To explore whether reversed collocations retain the processing advantage that they have in their canonical form, we conducted two primed lexical decision experiments: Experiment 1 in English, and Experiment 2 in Lithuanian, an understudied language. We presented both forward and backward collocations and compared them to matched control phrases. We also explored which collocational measure (phrasal frequency, MI, t-score, or ΔP) worked as the best predictor of processing speed. We found a clear priming effect for both languages when collocations were presented in their forward form, which is in line with previous research. There was no priming for the backward condition in English, but a priming effect for it in Lithuanian, where the reversed word order is acceptable albeit marked. These results are not easily explained by holistic storage. As far as collocational measures are concerned, they all seem to perform reasonably well, with none of them being clearly better than the others.

Learning multimorphemic words involves the simultaneous learning of two hierarchically organized categories. In such words, sub-lexical units usually encode superordinate categories, whereas whole words encode exemplars of these categories. Complex, non-linear word structure is common in Semitic languages and can be used to probe the learning of multiple form-meaning associations. The aim of this study was to investigate how well Hebrew-speaking adults learn the dual form-meaning relationships that reflect different categorical levels following a few exposures to novel Hebrew-like words. Twenty-four native Hebrew-speakers were exposed to novel words through an interactive video story. Following a few exposures to the words, the learning of the exemplars was tested in a three-alternative-forced-choice identification test. The learning of the sub-lexical morphemes and the categories they encode were tested in generalization tests. The results show that a few exposures to novel, morphologically and conceptually complex words are sufficient to allow unsupervised simultaneous learning of two hierarchical categories even though the superordinate was not explicitly represented in the input.

The main aims of the present study are, first, to extend the current cognitive research on affective word processing in bilinguals by analyzing affective words in minimal linguistic contexts and, second, to explore the potential impact of the affective valence of prime nouns on the affective valence of target adjectives. To fulfill these aims, a semantic decision task was employed in which the Persian-English bilinguals saw a pair of words one after another, and were asked to decide whether or not the target word, which was an adjective loaded with positive or negative valence, was related in meaning to the preceding word, which was a noun. Mixed factorial repeated measure ANOVA was run on reaction times and error rates data. The results showed that bilinguals’ responses were slower and less accurate to negative target adjectives in comparison to positive target adjectives. The data further revealed that bilinguals were faster but less accurate when they were responding to related target adjectives compared to unrelated target adjectives. The results provide evidence for a dynamic interaction between cognitive and affective language processing in bilinguals.