Abstract In this paper, we explore whether the cognitive system is sensitive to distributional information when reading novel compounds. In two lexical decision experiments, we investigate the role of positional constituent frequency and interpretability in processing novel noun-noun compounds in Italian. Experiment 1 presents a classical factorial design, manipulating position-specificity and position of occurrence of constituent morphemes. Experiment 2 is a more exploratory regression approach investigating a greater range of positional frequency values at two different presentation durations. Results suggest a weak effect of position-specific frequency and a strong effect of compound interpretability. Post-hoc analyses reveal an effect of constituent position on interpretability ratings, suggesting that position-specific frequency enters compound processing through the backdoor of interpretability. The findings underscore the dynamic nature and the role of distributional properties in compound processing.
Predictive language processing enables listeners to anticipate upcoming words using available linguistic cues. In Greek, adjectives provide both semantic information and grammatical-gender agreement, offering a unique window into how these cues jointly guide real-time comprehension. The present study examined how semantic and grammatical cues in adjectives influence prediction in adulthood. Thirty-three native speakers of Greek, divided into three age groups (19-30, 48-60, and 61-80 years), completed a webcam-based eye-tracking task using the Gorilla platform. Participants viewed a 2 & times;2 visual array while listening to spoken sentences containing adjectives that varied in predictive informativeness across four conditions: No Cue, Semantic Only, Gender Only, and Semantic + Gender. Results showed that all age groups engaged in predictive processing, with the highest anticipatory fixations occurring when semantic and gender cues co-occurred. Younger adults demonstrated the earliest and strongest cue-based predictions, whereas older adults were slower overall, and relied more heavily on semantic information. These findings show the added value of examining morphology in the study of prediction and provide new evidence of age-related changes in the integration of linguistic cues during comprehension.
Abstract Stroke represents one of the leading causes of acquired cognitive and communicative impairments, limiting the ability to use language appropriately and profoundly impacting quality of life and social participation. Post-stroke aphasia offers a crucial pathway for the understanding of the neuropsychological correlates of language processing and, despite advances in acute stroke management, linguistic deficits continue to represent a significant challenge for rehabilitation and long-term care. Additionally, one of the major challenges in investigating post-stroke language deficits lies in the marked variability across patients. In this pilot study, we investigated whether stimulus design may enhance the detection of linguistic processing deficits in stroke-affected individuals, targeting lexical class distinctions and morphosyntax/lexical semantics interface-level processing. Results from twelve Italian-speaking patients with unilateral stroke (6 left-hemisphere, 6 right-hemisphere) revealed differential effects of lesion lateralisation and lexical class on task performance. Task accuracy was shaped by individual and lesion-related factors, revealing task-specific dissociations between left- and right-hemisphere patients. These findings underscore the value of using targeted linguistic stimuli to detect subtle deficits that may go undetected by standard screening tools, contributing to a more comprehensive understanding of post-stroke aphasia.
The study examines emotional responses to words representing a wide range of psychological valence and focuses on gender-related differences. We aimed to find out whether men and women differ in their emotional responses, and whether they can take the perspective of another gender. We used the slider paradigm (Warriner et al., 2017): participants saw a humanoid manikin, a scale and a word on the screen and were instructed to place the manikin as close to, or as far away from, the word as they believed the person represented by the manikin would prefer to be. A change in the shape of the manikin (we used the bathroom figures of a man and a woman) signaled the perspective that the participant was asked to adopt. To assess the cross-linguistic validity of the findings, we collected data from English and Russian.We found that women showed a wider range of emotional responses, while men displayed a flatter affect. Participants changed their response strategy in the right direction when estimating words for the opposite gender. The study showed a degree of universality in perspective-taking, demonstrating that both men and women are aware of the emotional preferences of the opposite gender.
Stroke represents one of the leading causes of acquired cognitive and communicative impairments, limiting the ability to use language appropriately and profoundly impacting quality of life and social participation. Post-stroke aphasia offers a crucial pathway for the understanding of the neuropsychological correlates of language processing and, despite advances in acute stroke management, linguistic deficits continue to represent a significant challenge for rehabilitation and longterm care. Additionally, one of the major challenges in investigating post-stroke language deficits lies in the marked variability across patients. In this pilot study, we investigated whether stimulus design may enhance the detection of linguistic processing deficits in stroke-affected individuals, targeting lexical class distinctions and morphosyntax/lexical semantics interface-level processing. Results from twelve Italian-speaking patients with unilateral stroke (6 left-hemisphere, 6 right-hemisphere) revealed differential effects of lesion lateralisation and lexical class on task performance. Task accuracy was shaped by individual and lesion-related factors, revealing task-specific dissociations between left-and right-hemisphere patients. These findings underscore the value of using targeted linguistic stimuli to detect subtle deficits that may go undetected by standard screening tools, contributing to a more comprehensive understanding of post-stroke aphasia.
Reading aloud of morphologically complex words has been consistently found to be impaired in acquired dyslexia. This morphological disorder is characterized by the production of larger numbers of errors with complex than with simple words and by the production of so-called morphological errors, i.e. errors that reflect the internal morphological structure of words and are dissociable from purely visual errors. However, most previous research has focused on suffixed words (e.g., payment [pay][ment]), while much less is known about how prefixed words (e.g., prepay [pre][pay]) are affected. In the present study, we report three experiments involving a German speaker with acquired phonological dyslexia. We aimed to provide a characterization of the impairment of prefixed words, as well as to test whether there are systematic differences in how prefixes and suffixes are impaired. Our findings suggest that prefixed words show the same general pattern of morphological impairment previously reported for other types of complex words; however, prefixes tend to be more severely impaired than suffixes. This aligns with accounts proposing that prefixes are acquired and processed less easily than suffixes, and more vulnerable in language loss.
In this paper, we explore whether the cognitive system is sensitive to distributional information when reading novel compounds. In two lexical decision experiments, we investigate the role of positional constituent frequency and interpretability in processing novel noun-noun compounds in Italian. Experiment 1 presents a classical factorial design, manipulating position-specificity and position of occurrence of constituent morphemes. Experiment 2 is a more exploratory regression approach investigating a greater range of positional frequency values at two different presentation durations. Results suggest a weak effect of position-specific frequency and a strong effect of compound interpretability. Post-hoc analyses reveal an effect of constituent position on interpretability ratings, suggesting that position-specific frequency enters compound processing through the backdoor of interpretability. The findings underscore the dynamic nature and the role of distributional properties in compound processing.
Many studies on different languages analyzed how spelling errors are produced and detected. Recently, a new generalization was made for several languages: frequently misspelled words are read more slowly, even when they are written correctly and one knows how to spell them. This is explained by the lower quality of their lexical representations diluted by the exposure to recurring errors. In this study, we confirm this generalization for Russian and report several novel findings. We conducted four experiments with different participants: two lexical decision tasks and two spelling error detection tasks, to compare a task that consciously focuses on spelling to the one that does not. Firstly, the accuracy rate in the error detection task was a better predictor of response times in the lexical decision task than other factors including spelling entropy and word frequency. This further confirms the low lexical quality hypothesis. Secondly, although Russian orthography is in the middle of the transparency scale, Russian patterned with languages having transparent orthographies in these experiments - this shows which properties may be relevant. Thirdly, we tested errors of different types and showed that this factor was important for the error detection task, but not for the lexical decision task, in which only the frequencies of different spellings matter.
Walking has been the focus of much of the existing work on multitasking given its complexity as a cognitive process and importance in daily life. This complexity is evidenced by gait variation observed in dual-task contexts. However, an open question concerns how walking effects concurrently performed cognitive tasks. Thus, we use virtual reality to investigate how walking and visual distraction modulate language processing in an ecologically valid, yet controlled manner. In this novel experimental paradigm, we gradually increase the cognitive burden on participants' lexical decision responses by adding visual distractors and concurrent walking demands. Accordingly, healthy, young participants performed a lexical decision task as either (1) a single-task+, while seated and with randomly appearing visual distractors, or as (2) a multitask, while walking on a self-paced treadmill through a VR city scape including visual distractors. Participants generally made faster lexical decisions while walking. However, in the single-task+ condition, participants made more errors when a distractor was present. These effects were somewhat modulated by individual differences in visual processing. Crucially, no clear dual-task cost was observed; rather, behavior adapted to increased demands within a specific domain. Overall, these findings suggest an interplay of both task-related and individual characteristics determining multitask performance.
This study investigated the degree to which cognitive mechanisms support word recognition and word inflection in aging and how this changes in Alzheimer’s disease (AD). We tested competing hypotheses regarding the functional organization of language within the broader cognitive system. One set of hypotheses, derived from dual-system theories like the Declarative/Procedural (DP) model, predicts a functional architecture segregated by linguistic function. An alternative set of hypotheses posits a more integrated architecture, organized by task demands and resource availability. We analyzed participants’ performance on a lexical decision task and a word inflection task, alongside neuropsychological tests, using both behavioral and network analyses. In healthy controls (HC), the network analysis revealed a highly integrated architecture where language tasks were clustered by functional demands (e.g., speed vs. accuracy) rather than segregated along a strict lexicon/grammar divide. In the AD group, behavioral results showed a classic dissociation, with disproportionate impairment on irregular word inflection — a pattern traditionally seen as evidence for a modular memory failure. However, our network analysis revealed a different underlying mechanism. We observed a dramatic network reorganization where a core declarative memory module became functionally isolated, causing language tasks to form new, compensatory alliances with remaining frontal-executive resources. This provides clear evidence of a shift where executive functions are recruited to support language abilities when dedicated memory systems decline. These findings suggest that the cognitive substrate for language is not static but adapts dynamically in neurodegeneration, shifting its reliance from failing declarative memory systems to domain-general executive control pathways.
Cortical memory circuits that represent words in the brain form the foundation of our mental lexicon; yet, their architecture remains poorly understood. A valuable approach to probing these representations is measuring the Mismatch Negativity (MMN), an electrophysiological brain response sensitive to various psycholinguistic variables; however, its specificity to lexical processing remains debated. To scrutinise its properties as a neural index of word-specific memory trace activity, we adapted the classical passive auditory oddball design to fMRI and recorded the BOLD (blood oxygenation level-dependent) correlate of the MMN elicited by words of different lexical frequencies and by phonologically matched control pseudowords. The results showed significant BOLD-MMN responses in bilateral superior-temporal and middle-temporal cortices. Crucially, these activations were more expressed for meaningful words than meaningless pseudowords, indicating BOLD-MMN sensitivity to the stimulus' lexicality. We also found the left temporal activity to be more pronounced for high- than low-frequency words, the effect not found for their pseudoword analogues, further confirming the lexical nature of these responses. This pattern of results is best explained by the automatic activation of long-term memory traces for real words formed in the process of previous linguistic experience whose intensity determines the strength of connectivity within these circuits and thus the magnitude of their activation, which thereby reflects the respective stimuli's status in the brain's mental lexicon.
This study explores gender assignment strategies in Russian-Hebrew code-mixed adjective-noun phrases. Russian features a three-gender system (masculine, feminine, neuter), while Hebrew uses a two-gender system (masculine, feminine). Despite these differences, both languages share transparent gender assignment cues: nouns ending in -a are typically feminine, while those ending in consonants are generally masculine. Both languages also feature opaque nouns. Eighty Russian-Hebrew speakers participated, divided into heritage language (HL) speakers (age of bilingualism onset [AoB]: 0–6 years, dominant in Hebrew) and immigrant (IMM) speakers (AoB: 9+ years, dominant in Russian). Participants rated the acceptability of code-mixed sentences featuring Russian nouns within Hebrew matrix sentences. Results showed a preference for strategies combining shape-based and insertion approaches with transparent nouns, reflecting overlapping linguistic cues in both languages. The same strategies were preferred for opaque congruent nouns, where no gender conflict existed. For opaque gender-incongruent nouns, strategy use was shaped by the degree of overlap between the languages’ gender systems. Although differences between HL and IMM speakers were expected, no group variation in strategy use was found. These findings advance our understanding of gender assignment in code-switching contexts and shed light on how bilinguals represent and process gender when two linguistic gender systems are involved.
Morpho-phonological alternations in inflectional paradigms are commonly analyzed as purely formal phenomena, in which the mapping of phonological structure and morpho-syntactic categories is organized without recourse to semantic properties of the words involved. The present paper explores the role of semantics using the Discriminative Lexicon approach (Baayen, et al. 2019). The test case explored in this paper is German nominal number, a system involving complex morpho-phonological variation (e.g. K & ouml;pcke et al., 2021; Heitmeier et al., 2021; Plag et al., 2024; McCurdy, 2024). Using word2vec vectors as semantic representations, and triphones as form representations, we created two-layer linear discriminative learning (LDL) networks that map form representations directly onto semantic representations (modeling comprehension), and semantic representations onto form representations (modeling production). The LDL mappings successfully predict the forms and the meanings of the singular and plural nouns taken from a pertinent study (Domahs et al., 2017). A number of semantic and phonological measures derived from the LDL network also very successfully distinguished between singular and plural forms. Our results demonstrate that semantics, in addition to formal and grammatical properties, may play a decisive role in the representation and processing of German nominal number. The system of German nominal number can be understood as emerging from the distributional properties of words on the one hand, and basic principles of discriminative human learning on the other.
Morphological effects are consistently reported in written word recognition but their theoretical explanation is still debated. The paper aims to disentangle the specific role of morphological information from orthography during written word recognition. We exploit the “morphemic ambiguity” of stem-homographs, words that share formally identical but morphologically and semantically unrelated stems (e.g., mor -a , “blackberry” and mor -ire , “to die”). Two unmasked visual priming lexical decision experiments are described. The results replicate the inhibitory priming effects elicited by stem-homographs, extend it to their allomorphic variants in Italian (e.g., mor -a , and muor -e , “he/she dies”, verb form deriving from an allomorph ( muor -) of the homographic stem mor - of morire) and differentiate the stem-homographs and the stem-allomorphs inhibition from the positional effects of orthographic overlap in prime-target pairs. The pattern of data confirms the presence of mechanisms of morphological segmentation in word recognition. On the other hand, the effects fit with hypotheses of graded morphology where word stems are supposed to affect the processing of words on the basis of multiple connections between regular and irregular forms included in their inflectional paradigms and derivational families.
This study investigates whether the recognition advantage of proper names (PN) over common nouns (CN) — reported in several languages — is also observed in Spanish, and whether different types of PN (e.g., personal vs. geographical) are affected to the same extent. Drawing on semantic theories that assign different presuppositional meanings to subcategories of PN, we designed four experiments to examine PN processing patterns: a lexical decision task, two categorization tasks, and a semantic priming task. To explore which semantic factors account for variability within each subcategory, we also conducted a series of regression analyses. The results confirm a cross-linguistic recognition advantage for PN in categorization tasks; however, this effect is limited to personal PN — even after controlling for affective factors and familiarity — while the advantage for geographical PN appears to be language-specific. Participants’ behavioral responses to geographical PN resembled those elicited by CN. These findings suggest that PN is a semantically heterogeneous category, and that their recognition lies on a continuum with that of CN.
Pseudo-compound words (e.g., hippie ) are words that look like compound words (e.g., snowman ) but, in fact, do not have the morphemic structure of a compound word. For instance, the pseudo-compound word hippie has hip and pie embedded in it, but they do not function as morphemes. Pseudo-compound words vary in terms of phonological transparency. Some, such as pumpkin , are phonologically transparent because the pronunciations of pump and kin are maintained when these pseudo-constituents become part of pumpkin . In contrast, the pseudo-compound word carrot is phonologically opaque because the pronunciations of car and rot change when they are embedded in carrot . Previous studies have demonstrated that compound words go through morphological decomposition and attempts at meaning construction during written production tasks. For instance, compound words are not output as single units during typing tasks, but rather are typed in chunks based on their morphology (e.g., snowball is typed in two parts: first as snow and then as ball ), which results in an increase in typing latencies at the morpheme boundary (i.e., between the last letter of the first constituent and first letter of the second constituent). The same is true for pseudo-compound words, even though these words do not have the morphemic structure of a compound word. Given that previous research has shown that the morphological decomposition of compound words during typing tasks looks different depending on the semantic transparency of the compound word’s constituents (i.e., the degree to which the meaning of each constituent of a compound word contributes to the word’s overall meaning), we wanted to examine whether the level of phonological transparency of the pseudo-constituents of a pseudo-compound word influences typing latencies at the pseudo-morpheme boundary.
According to the affix-chunking hypothesis, a letter search should be more difficult for a letter embedded in an affix compared with a non-affixed letter sequence because affixes have a functional significance. On the other hand, the decomposition hypothesis claims that derived (e.g., hunter ) and pseudo-derived words (e.g., corner ) are processed similarly, with lexical access being driven by affix stripping followed by the activation of the remaining stem to reach the mental lexicon. We conducted a letter-search task to test these hypotheses using both prefixed (e.g., détour ‘detour’), suffixed (e.g., acteur ‘actor’) words, compared with matched pseudo-prefixed (e.g., décor ‘decor’), pseudo-suffixed (e.g., fleur ‘flower’) words. Decision latencies on letter targets were compared to non-affixed words for each type of affix (e.g., drogue ‘drug’ for détour, décor and tâche ‘task’ for acteur, fleur ). Our results revealed an asymmetry in the processing of suffixed versus prefixed words. While a significant facilitation effect was found for suffixed words relative to pseudo-suffixed words, no similar advantage was observed for prefixed over pseudo-prefixed words. The asymmetry in identifying letters in prefixes and suffixes is interpreted in terms of the differing functional salience of affixes in French.
Through a consideration of the noun inflections of Finnish, we examine Butterworth’s (1983) full-listing hypothesis (FLH), that whole-word forms are stored in the mental lexicons of language learners/users. While previous studies have consistently confirmed that complex derivational constructions are stored/retrieved this way in Finnish, the current consensus is that the inflectional forms are not, primarily because of the extremely large number that Finnish grammar projects. We challenge this reasoning in favor of a new theory that posits a dual lexicon for inflections, one for the full forms of stems and a parallel one for inflectional affix strings. To test this, we performed a segmentation experiment assessing the relative salience of morpheme and syllable boundaries in Finnish nouns, and also examined a large corpus of quasi-natural spoken dialogues for further evidence. Together, our findings demonstrated that the salience of the boundaries we examined was consistent with our expectations, that the number of inflected forms used in our corpus was only a tiny fraction of those predicted by rule, and that their productivity was highly correlated with their frequencies. While this evidence does not fully confirm the dual lexicon hypothesis, it does strongly indicate that no online grammatical rules are involved.
