Using a nonce-word inflection task, we examine the morphosyntactic productivity of adult native speakers of Spanish who are either beginning to learn to read and write (semi-literates) or have acquired literacy in late adulthood (late-literates), as well as age-matched controls (high-literates). High-literates consistently provided the appropriate form more often than late-literates, who in turn were better than semi-literate participants. Crucially, group interacted with person, number, and conjugation, such that the between-group differences were larger for the less frequent cells in the paradigm, indicating that literacy-related differences are not merely a consequence of the high-literacy group being more engaged or test-wise. This suggests that the availability of written representations may facilitate the acquisition of certain aspects of grammar. We also observed vast individual differences in productivity with inflectional endings. These results add to the growing body of research which challenges the assumption that all native speakers converge on the same grammar early in development. High-literates consistently provided the appropriate form more often than late-literates, who in turn were better than semi-literate participants. Crucially, group interacted with person, number, and conjugation, such that the between-group differences were larger for the less frequent cells in the paradigm, indicating that literacy-related differences are not merely a consequence of the high-literacy group being more engaged or test-wise. This suggests that the availability of written representations may facilitate the acquisition of certain aspects of grammar. We also observed vast individual differences in productivity with inflectional endings. While there is some work on speech processing and metalinguistic awareness in semi- and illiterate speakers, very little is known about their morphosyntactic abilities. One recent study ( Eme et al., 2010 ) analyzed oral narratives produced by semi-illiterate French-speaking adults and age-matched controls. The researchers observed significant differences between groups in the number of morphological errors and syntactic complexity (mean length of clauses and proportion of complex structures), with effect sizes (Cohen’s d ) ranging from 0.67 to 1.1, as well as in the ability to produce coherent narratives. This is illustrated by a study by Aguado-Orea and Pine (2015) . The authors investigated verb use in the spontaneous speech of two Spanish-speaking children between the ages of 2;0 and 2;6. On average, the children produced 1.8 different inflectional forms per verb: in other words, the vast majority of verbs occurred in only one or two inflectional forms. Furthermore, there were large differences in accuracy on forms belonging to different parts of the paradigm. The children were overwhelmingly correct on third person singular (3 s) forms (over 99% correct) and the first person plural (95–97% correct). These are very frequent forms, accounting for 87% of the verb tokens produced by the children. In contrast, error rates on less frequent forms, such as the second and third person plural, were much higher (from 33 to 46%). The overall error rate was 4%. In other words, the very low error rates on the most frequent forms hide much higher error rates in low frequency cells in the paradigm, suggesting that the children had acquired only parts of inflectional system of the language. A comprehension study by Pérez-Leroux (2005) also raises doubts about the early acquisition of verb inflections. Thus, producing a past tense form of a novel verb appears to be challenging even for adult speakers, which makes it promising area for studying individual differences in linguistic knowledge, and the possible role of literacy to explaining such differences. Spanish verbal morphology In this study, we use a nonce-verb inflection task to examine the morphosyntactic productivity of adult native speakers of Spanish with different literacy levels. Spanish verbs are marked for person, number, tense, aspect, and mood. There are three classes of verbs (traditionally referred to as ‘conjugations’), distinguished by thematic vowels. First conjugation verbs are characterized by the thematic vowel a and constitute the largest class by a large margin: according to Mendoza et al. (2001) , 78% of all verbs belong to this group. The second and third conjugations are characterized by the vowels e and i , respectively, and each accounts for about 11% of all verbs ( Mendoza et al., 2001 ). As shown in Table 1 , the paradigms for these two classes share many endings, and the preterite and imperfect endings are identical. Table 1 The indicative forms of Spanish verbs. Furthermore, also in accordance with our predictions, all three groups performed better on first conjugation than second conjugation verbs, and better on third person singular inflections than on the first person plural, which in turn was easier than second person plural. Crucially, the effects of conjugation and person/number condition were more pronounced in the two low-literacy groups, and in particular in the semi-literates, than in the control participants. Given the very large individual differences in performance within groups and the relatively small sample size, it is all the more surprising that we observed significant effects of It is also worth noting that, while the control participants performed much better on the nonce-verb inflection task than the low-literate groups, they supplied the target form in only 50% of the trials. Given the complexity of the system, the number of possible incorrect responses is quite large ( cf. ‘Coding scheme’ subsection), so getting everything right 50% of the time is still quite a feat. Nevertheless, native speakers are supposed to have full command of at least the basic inflectional patterns of the language — so the reader may be forgiven for being a little skeptical.

Although phonological deficits are unanimously recognized as one of the key manifestations of developmental dyslexia, a growing body of research has reported impairments in morphological abilities. Our study aimed at casting further light on this domain by investigating the morphological awareness skills of 21 children with dyslexia (mean age 9.10 years old) and 24 children with typical development (mean age 10.3 years old). All children were monolingual speakers of Italian, which is a morphologically rich language characterized by complex inflectional and derivational paradigms. We developed an experimental protocol inspired by Berko's Wug test and composed of 11 tasks addressing inflectional and derivational processes. Participants were asked to manipulate nonwords of various lexical categories, modeled after the phonotactic structure of Italian, and manipulation involved both word formation and base retrieval. Conditions of the experiments were based on verb conjugation classes differing in frequency, productivity, regularity, and formal transparency. Results confirmed that morphological skills are impaired in dyslexic children, who performed significantly more poorly than their age-matched peers in all tasks. Children with dyslexia were especially challenged by tasks and conditions requiring advanced morphological awareness skills, such as the retrieval of infinitives of infrequent and irregular conjugation classes. The educational and clinical implications of these findings are discussed. We developed an experimental protocol inspired by Berko’s Wug test and composed of 11 tasks addressing inflectional and derivational processes. Participants were asked to manipulate nonwords of various lexical categories, modeled after the phonotactic structure of Italian, and manipulation involved both word formation and base retrieval. Conditions of the experiments were based on verb conjugation classes differing in frequency, productivity, regularity, and formal transparency. Results confirmed that morphological skills are impaired in dyslexic children, who performed significantly more poorly than their age-matched peers in all tasks. The current study aims to explore morphological awareness skills in Italian-speaking children with dyslexia and with typical development. Italian is a fusional language, characterized by complex inflectional and derivational paradigms, and research on morphological abilities in Italian dyslexic children is still scarce. Parallel to nominal declension classes, verbal conjugations entail a high number of inflectional endings and correlate with a higher or lesser amount of root/stem allomorphy [ 65 ]. Indeed, conjugation classes exhibit different levels of frequency, productivity (referring to the ratio of the number/frequency of new verb formations of one conjugation class and the total number/frequency of new verb formations), and formal regularity, as shown in Table 2 . 10.1371/journal.pone.0276643.t002 Table 2 Italian verb conjugation classes. Conj. class Theme Vowel Example Frequency Productivity Regularity Conj. I -a- amare ‘to love’ high high high Conj. Notice that the third conjugation class in - ire contains a subclass of verbs with infixed - isc -, as in finire / fin-isc-o ‘to end / I end’ (vs. dormire / dorm-o ‘to sleep / I sleep’): this subclass is more regular and stable than the other verbs in - ire . To the aims of the present study, however, the identification of this fourth subclass is irrelevant as no task directly taps the inflection features of this verbal paradigm. Conj. II (- ere , prendere ‘to take’) is also scarcely frequent, but it is totally unproductive and highly irregular, characterized by a high rate of root allomorphy, with only a few regular forms (see [ 62 , 65 , 67 – 69 ]). As for the derivation of nouns from nominal bases, we tested the production of evaluative nouns eliciting the suffixes -ino , - one , and -accio (respectively diminutive, augmentative, and pejorative). Finally, as for the (de-)verbal domain, we assessed both inflectional and derivational phenomena, considering all three Italian conjugation classes; we also tested the children’s ability in base-retrieval tasks requiring them to identify the target verb form from a complex pseudoword. We will now describe each of the 11 tasks administered; for convenience, a summary of the elicitation formulas and experimental conditions, with an example of each condition is reported in Table 5 . The best fitting model showed a significant main effect of Group, with dyslexics being less accurate than controls (β = -0.95, SE = 0.19, z = -4.97, p < .0001) and a main effect of Category, with accuracy being lower with verbs than with nouns (β = -1.39, SE = 0.34, z = -4.13, p < .0001). In summary, both groups of children had more difficulties in inflecting and deriving verbs compared to nouns, and dyslexics were generally less accurate than controls. 3.4. Analysis 3 In Analysis 3, we aimed at comparing groups in derivational and inflectional tasks, while also considering the role of the three conjugation classes. As shown in Table 8 , the performance of both groups was generally higher in inflection than derivation, with difficulties being particularly marked with verbs in - ere ; moreover, dyslexics were generally less accurate than controls. 10.1371/journal.pone.0276643.t008 Table 8 Mean accuracy (SDs) of the two groups in inflectional and derivational tasks in the three conjugation classes. These deficits affect inflectional and derivational morphology, span over lexical categories and declension/conjugation classes, and hamper the dyslexic children’s ability to extract and apply patterns of inflection and word formation. Therefore, our results confirm the characterization of dyslexia as an impairment that extends beyond phonology and interferes with the development of morphological awareness, in line with recent research on other languages.

Morphological priming and paradigm size effects have been established in single word reading studies. Morphological priming effects have also been found in some sentence-reading studies, however, attempts to find priming effects in longer texts have failed. To our knowledge, paradigmatic effects have not yet been examined in text reading. The current study made use of the MECO corpus \citep{Siegelman:etal:2022} to explore paradigmatic and morphological priming effects in Estonian and Finnish, two morphologically rich Finno-Ugric languages. Unlike prior reports on Dutch, English and Spanish text reading, the current study showed clear long-lag inflectional priming effects. We also observed inflectional paradigm size effects for Estonian during text reading, but not for Finnish. These results suggest that inflectional variants of a particular word in Estonian and Finnish get and remain activated even when text context is present. However, effects of inflectional priming and paradigm size may be task- and language-specific. That is, they may surface only as cues when the support from an inflectional paradigm is semantically most beneficial for the reader. Running head: Long-lag morphological priming and inflectional paradigm size effects in Estonian and Finnish text reading 1 Long-lag morphological priming and inflectional paradigm size effects in Estonian and Finnish text reading Kaidi L˜ oo University of Tartu, Estonia Raymond Bertram University of Turku, Finland Victor Kuperman McMaster University, Canada Word count: Version: July 2024 Corresponding author: Kaidi L˜ oo Institute of Estonian and General Linguistics, University of Tartu Jakobi 2-405, 50090 Tartu, Estonia e-mail: kaidi.loo@ut.ee Long-lag morphological priming and inflectional paradigm size effects 2 Abstract Morphological priming and paradigm size effects have been established in single word reading studies. Morphological priming effects have also been found in some sentence-reading studies, however, attempts to find priming effects in longer texts have failed. To our knowledge, paradigmatic effects have not yet been examined in text reading. The current study made use of the MECO corpus (Siegelman et al., 2022) to explore paradigmatic and morphological priming effects in Estonian and Finnish, two morphologically rich Finno- Ugric languages. Unlike prior reports on Dutch, English and Spanish text reading, the current study showed clear long-lag inflectional priming effects. We also observed inflectional paradigm size effects for Estonian during text reading, but not for Finnish. These results suggest that inflectional variants of a particular word in Estonian and Long-lag morphological priming and inflectional paradigm size effects 17 Figure 3: Partial effects of paradigm size fitted to Estonian gaze duration, total fixation duration and rereading. longer total fixation duration than shorter words.360 Priming condition and Inflectional paradigm size were both significant predictors of the eye movements.361 As shown in Table 6 and in the middle panel of Figure 2, compared to the baseline condition, words in the362 enhanced inflectional priming condition were read the fastest, followed by words in the identity priming and363 in the inflectional priming conditions. Pairwise comparisons showed that words in the enhanced inflectional364 priming condition were read significantly faster than words in the identity and in the inflectional priming365 conditions. The difference between the identity priming and the inflectional priming conditions was not366 significant.367 The middle panel of Figure 2 illustrates the effect of inflectional paradigm size on total fixation duration.368 Words with more paradigm members were read faster than words with fewer paradigm members, while the369 clearest effect is again in the middle range.370 In summary, like in the gaze duration model, Priming condition and Inflectional paradigm size were371 significant predictors of total fixation duration.372 Long-lag morphological priming and inflectional paradigm size effects 27 Raveh, M. and Rueckl, J. G. (2000). Equivalent effects of inflected and derived primes: Long-term morpholog-583 ical priming in fragment completion and lexical decision. Journal of Memory and Language, 42(1):103–119.584 Rodriguez-Fornells, A., M¨ unte, T. F., and Clahsen, H. (2002). Morphological priming in spanish verb forms:585 An erp repetition priming study. Journal of cognitive neuroscience, 14(3):443–454.586 Royle, P., Drury, J. E., Bourguignon, N., and Steinhauer, K. (2012). The temporal dynamics of inflected587 word recognition: A masked erp priming study of french verbs. Neuropsychologia, 50(14):3542–3553.588 Schmidtke, D., Van Dyke, J. A., and Kuperman, V. (2021). Complex: An eye-movement database of589 compound word reading in english. Behavior Research Methods, 53:59–77.590 Schreuder, R. and Baayen, R. H. (1997). How complex simplex words can be. Journal of Memory and591 Language, 37:118–139.592 Siegelman, N., Schroeder, S., Acart¨ urk, C., Ahn, H.-D., Alexeeva, S., Amenta, S., Bertram, R., Bonandrini,593 R., Brysbaert, M., Chernova, D., et al. (2022). Expanding horizons of cross-linguistic research on reading:594 The multilingual eye-movement corpus (meco). Behavior Research Methods, 54(6):2843–2863.595 Stanners, R. F., Neiser, J. J., Hernon, W. P., and Hall, R. (1979). Memory representation for morphologically596 related words. Journal of Verbal Learning and Verbal Behavior, 18:399–412.597 Tomaschek, F., Hendrix, P., and Baayen, R. H. (2018). Strategies for addressing collinearity in multivariate598 linguistic data. Journal of Phonetics, 71:249–267.599 Weyerts, H., M¨ unte, T. F., Smid, H. G., and Heinze, H.-J. (1996). Mental representations of morphologically600 complex words: an event-related potential study with adult humans. Neuroscience Letters, 206(2-3):125–601 128.602 Wood, S. N. (2017). Generalized additive models: an introduction with R. CRC press.603