Session 6B

There is compelling empirical evidence that the linguistic distance between L1 and an additional language can influence broad acquisition outcomes; for example the linguistic distance between L1 and L3 Dutch, measured through a variety of lexical, morphological and phonological features, can predict attainment in the latter (Schepens et al 2016, 2020). Big data from assessment and teaching organisations enable empirical studies with a wide range of typologically diverse languages. One crucial question is if the predictive effect of linguistic distance on proficiency scores arises as an aggregate of individual similarities or whether broad typological features (e.g. presence vs. absence of agreement, word order) influence the acquisition of specific structures independently of similarities/differences of individual features. More generally, how can we connect findings from large scale empirical studies with insights from studies focusing on fine-grained L1-L2 variation? Answering these questions will shed light to our understanding of crosslinguistic influence and will have concrete implications for the increasingly multilingual classroom.
This workshop will address the question of linguistic distance for the acquisition of morphosytnax in additional languages and its implications for education:
Questions:
1. How do we measure linguistic distance? Can we build on syntactic typology (e.g. Ceolin et.al.2021) to obtain measures of syntactic distance that can provide testable predictions for Ln acquisition? How do syntactic distance measures compare with measures of lexical, phonological and morphological distance?
2. Can we distinguish effects of broad typological classifications from individual level similarities/differences? E.g. does the acquisition of the article depend solely on the availability of a congruent item in L1, or also on broader typological features of nominal structure (e.g. number vs. classifier) and more language general features such as word order or tense/aspect/mood (TAM)? Are L1 incongruent items acquirable at all?
3. Can we assume that L1-L2 (typological) similarity is always facilitative for L2 acquisition? Or is it possible that similarity might mislead learners, leading to misanalysis and erroneous generalizations? 
4. How does linguistic distance interact with other factors impacting on acquisition, e.g. age, education, socioeconomic background? 
To consider these questions we propose a colloquium of 4 presentations followed by one discussant and general Q&A. Presentations with questions addressed are: 
- Oksuz, Alexopoulou, Derkach and Tsimpli: The influence of L1 typology on the acquisition of the L2 English article: a large scale corpus study (Q1&2)
- Schepens, Van Hout and Van Der Slik : "Linguistic distance effects in adult additional language learning: The relevance and interpretation of linguistic features " (Q1&4) 
- Tania Ionin: "Plural marking and cross-linguistic influence" (Q2 & 3)
- Kook-Hee Gil: "Restructuring vs. development: when typological similarities do not facilitate L2 acquisition" (Q2&3)
Discussant: Professor Lydia White (we aim to have a second discussant with a focus on education/social impact). 
References:
Ceolin et.al. 2020, Frontiers in Psychology.
Schepens et.al, 2016, Language Learning, 66(1). 
Schepens et.al. 2020, Cognition, 194.

This paper argues that typological similarity does not necessarily facilitate L2 acquisition. We present two sets of data in support: one from L2 acquisition of Korean wh-quantifiers by English, Japanese and Chinese speaking learners (Gil, Marsden, in press) and the other from L2 acquisition of English experiential constructions by Chinese and Korean speaking learners (Grillo et al. in progress). Both sets of data compare different L1 groups where the L1s contrast with each other with respect to a specific L2 target property, such that it involves different learnability tasks: (i) the L1 item has a micro variant of L2 item and the acquisiton requires (micro-)restructuring and (ii) the L1, typologically different, does not have a congruent item and the acquisition requires creating a new structure in L2 interlanguage. The findings in both studies show that the learners in (ii) perform better than those in (i). 
For instance, Gil and Mardsen (in press) compare Chinese, Japanese and English speaking learners in L2 acquisition of Korean wh-quantifiers. Chinese and Japanese are typlogically closer (but not identical) to Korean with respect to the double function of wh morphemes: wh morphems are used as wh interrogatives in wh-questions, as well as as existential quantifiers (e.g., who = someone). This morphological paradigm is different in English where the wh morpheme is only used as wh-interrogative. In the test on the use of wh existential quantifiers in L2 Korean, the results showed that it was the English learner group who outperform the Chinese and Japanese learner groups. Similar findings were also shown in the study of L2 English experiential constructions. Grillo et al. (in progress) conducted a crosslinguistic survey of experiential constructions and propose three different types of languages. One type is where a language exploits an aspectual system to express experience such as English experiential perfect ('have you ever…?'). Chinese also belongs to this group and uses an aspectual particle 'guo'. Another type of language, such as Korean, uses an entirely different construction, namely, complex DPs in existentials. Grillo et al. tested Chinese and Korean speaking learners in L2 English experiential constructions. The results show that despite the closer item available and typologically similar, the Chinese group was outperformed by Korean learners whose task involves a development of new construction, entirely different from their L1. 
Both of these studies show that having a congruent item in L1 is not necessarily advantageous and that the restructuring from a micro-variant form in L1 can be more costly and harder than other L1 groups whose acquisition tasks involves a development of a new construction in L2.
Selected references
Gil and Marsden (in press) "L2 Acquisition of Korean wh quantifiers: L1 transfer at the lexical mapping, in Tsoulas, G. (ed.) MIT Working Papers in Linguistics (Proceedings of Workshop on Altaic Formal Linguistics). 
Grillo, et al. (in progress) "The grammar of experience and the acquisition of NPIs: ever and any show different paths of acquisition in Chinese and Korean L2 speakers of English". 

Our previous studies demonstrated that different measures of linguistic distance explain a large part of the variability in additional language learning proficiency (Schepens et al., 2020). Proficiency was measured as a global score for four modalities speaking, writing, listening, or reading, taken from a B2 level state exam for Dutch as an additional language across 50,000 adult immigrants with 50 different L1s who came to the Netherlands for study or work. In this talk, we embed the role of linguistic distance within a perspective on SLA that includes contextual non-linguistic factors such as the age of arrival, exposure, education, and gender (Schepens et al., resubmitted).
We tested distance measures based on lexical, morphological, and phonological features. The measures incorporated substantial numbers of linguistic features based on large-scale typological databases such as IELex, WALS, and PHOIBLE. This way of abstraction turned out to be successful across all three domains. We made use of (1) branch lengths from Indo-European phylogenetic language trees based on expert cognacy judgments as well as automatic similarity computations, (2) different selections and comparisons of structural properties (morphology in particular) as compiled from reference grammars, and (3) comparisons of sound inventories as well as the structural properties of sounds. The effectiveness of these measures can best be explained by active transfer of language knowledge and language-related processes of production and perception.
For the present talk, we therefore discuss three relevant questions raised by our outcomes. First, it is not clear what makes distance measures complementary or overlapping. Our results show how lexical, morphological, and phonological distances correlate with each other while still making independent contributions to explaining the variation in proficiency scores. The different directions in which features are abstract, and their hierarchies, need to be compared to the ways in which (hyper)parameters can be represented. Second, it must become clear how distances relate to discussions about the nature of cross-linguistic influence within the usage-based and generative perspectives on SLA. Ln learnability and Ln learning processes are constrained by the nature of distance effects, but our distance measures vary in their correlations depending on the language modality involved (Schepens et al., 2022). We have to address the question of why the four modalities differ. Distance effects as well as their non-linear interactions with age at arrival are strong across modalities, while interactions between distance and education are strong for receptive skills only. Finally, how do our distance measures distinguish between within and between language family biases, since language structures are naturally lineage-specific? 
We will present a general overview of our approach to clarify its outcomes and to spell out the context in which we are trying to find the answers to these fundamental questions in SLA.
References
Schepens, J., Van Hout, R., & Jaeger, T. F. (2020). Big data suggest strong constraints of linguistic similarity on adult language learning. Cognition, 194, 104056. https://doi.org/10.1016/j.cognition.2019.104056
Schepens, J., van Hout, R., & Van der Slik, F. (2022). Linguistic dissimilarity increases age-related decline in adult language learning. https://psyarxiv.com/b28zd

Adult second language (L2) learners are known to struggle with inflectional morphemes in their L2 (Slabakova 2008). While older morpheme order studies (e.g., Bailey et al. 1974) emphasized universality in the L2 acquisition of morphology, more recent approaches have, from different perspectives, emphasized the role of L1-transfer in this domain. On the Feature Reassembly Hypothesis (FRH, Lardiere 2009), learners transfer morpheme feature specifications from the L1 to the L2. On the Morphological Congruency Hypothesis (MCH), learners have difficulty with L2 morphemes that have no L1 equivalent, and lack automatized knowledge of such incongruent morphemes.
This talk will discuss cross-linguistic influence with plural marking, in the L2 English of L1-Korean and L1-Mandarin speakers. The talk will focus on (i) whether these learners transfer morpheme feature specifications from their L1 to their L2; and (ii) whether they are able to integrate incongruent L2 morphemes into their Interlanguage grammar.
The Korean plural marker -tul is optional in most contexts, but obligatory for expressing plurality with definites (Kim 2005). The Mandarin plural marker -men is optional with [+human] nouns and ungrammatical with [-human] ones (Li 1999). Choi & Ionin (2021, Exp.2) found that L1-Korean L2-English learners were more sensitive to missing plural marking with definites than L1-Mandarin L2-English learners, consistent with the link between definiteness and plural marking in Korean. At the same time, the two groups were equally sensitive to missing plural marking with indefinites, in contexts where no plural marker is required in their L1. These results contradict those of Jiang (2007) / Jiang et al. (2017), who found a complete lack of sensitivity to missing plural marking on the part of L1-Mandarin L2-English learners.
In a new set of experiments, we delve deeper into the question of what L1-Mandarin L2-English learners know about English plural marking. We use four different tasks to examine learners' production, comprehension, grammaticality judgments, and online processing of the plural marker -s. In order to address the contradictory findings of prior studies, which used different cues to establish plurality, we compare how learners perform when the cue is a numeral (three books) vs. a quantifier (several books) vs. a partitive (three/several of the books) vs. no additional cue to plurality (the books). Data have been collected from 24 native English speakers and 32 L1-Mandarin L2-English learners, and data analysis is ongoing. Preliminary results show that the learners, while much less accurate than the native speakers, are sensitive to errors with plural marking both offline (in grammaticality judgments and written production) and online (in an online comprehension task; results from an online self-paced reading task are currently being analyzed). Pending further analyses, we tentatively conclude that while learners do have difficulty with morphemes that have no exact equivalent in their L1, consistent with the MCH and FRH, they are also able to acquire incongruent L2 morphemes. We consider these findings in light of whether L1-transfer should be construed in terms of broader grammatical constructions.

There is strong empirical evidence that the linguistic distance between L1 and an additional language can influence broad acquisition outcomes (Schepens et al., 2016, 2020). One crucial question is whether linguistic distance also affects the acquisition of individual features like articles. Does the acquisition of such features depend solely on the availability of a congruent element in the L1 (e.g. Murakami et al., 2016), or do broader typological differences guide how learners approach the input, influencing their acquisition? Recent work suggests a differential effect of linguistic distance on lexical acquisition and complex syntax. Shatz (2021) found no effect of linguistic distance on learners' lexical diversity and use. By contrast, Authors (2021) found that preferences in the use of L2 English relativisers correlated strongly with L1-L2 linguistic distance but not with the availability of a congruent L1 relativiser (pronoun vs. subordinator). In this talk we turn to the English article. 
Research questions
1. Is learner accuracy in the use of L2 English articles linked to a) the availability of a definite article in their L1? and/or b) the linguistic distance between learners' L1s and L2?
2. Does L1-L2 linguistic distance/congruency affect the definite and indefinite article similarly? 
Method
Data were extracted from EFCAMDAT (Shatz, 2020): 34 million words; 527,758 writings ;104,541 learners; 11 native languages (Portuguese, Chinese, German, French, Italian, Japanese, Arabic, Russian, Mexican Spanish, Korean and Turkish); A1 to B2 CEFR proficiency levels. We used teacher error labels to identify omission, and substitution errors and calculated the ratio between number of correct suppliance and obligatory contexts (Target Language Use: TLU scores, Pica, 1983). 
Typological distance. Following Murakami and Alexopoulou (2016), we classified languages as article+/-, depending on the availability of a definite article. We compared this binary classification with continuous scores of lexical (Shatz, 2022) and syntactic distance (in the nominal domain, Ceolin et al., 2021). 
Results
Table 1: Accuracy scores and linguistic distance scores per L1.pdf
Table 1 shows TLU scores and linguistic distance scores for 11 L1s. Mixed-effects regression modelling revealed that Article+/- affected L2 learners' accuracy. Accuracy increased with proficiency; this effect of proficiency was stronger for the Article+ group and for the indefinite article. The linguistic distance scores showed weaker correlations.
Discussion
The results generally confirm the effect of proficiency and L1 Article+ in Murakami et al. (2016), for a different set of L1s and in a different corpus. Some language-family effects are observed, in particular within the Article+ group. The lexical measures correlated highly with the ranking within European languages but were not predictive outside this group. The theoretical literature suggests that Chinese lacks an article but has an abstract definiteness feature (Cheng et al., 2017), which can explain the high accuracy of Chinese learners. We are currently considering further measures of morpho-syntactic distance to identify those typological features that can help capture the obtained rank and, therefore, shed light on what impacts the L2 acquisition of articles, beyond the Article+/- distinction.