Linguistic Approaches to Bilingualism - Volume 16, Issue 3, 2026

Variability at the group and individual level can inform understanding in second language research. For example, examining by-group and individual differences can play an important role in teasing apart competing theoretical accounts of first and second language processing. In this paper, I review existing research examining variability in second language sentence processing. Focusing on relative clause attachment, filler-gap dependency resolution and subject-verb agreement, I examine these three phenomena as case studies for how examining variability can inform key debates in second language processing research. I review variability at the group and individual level in both cognitive and linguistic abilities, language experience and proficiency, and in the linguistic environment. I also discuss methodological issues in generalizing findings across studies and in using psycholinguistic tasks to examine individual variation in language processing, which pose important challenges that need to be addressed if the field is to move towards an individual differences perspective of second language processing. Although the review focuses on three linguistic phenomena in second language sentence processing, the issues discussed are relevant to the examination of variability in bi-/multilingual language acquisition and processing more broadly.

The Timing Hypothesis predicts that age of onset and L2 input in bilingual acquisition interact with timing in monolingual acquisition: early-acquired phenomena, mastered before age four by monolingual children, are subject to age of onset effects, whereas late-acquired phenomena, mastered after age four by monolinguals, are not affected by age of onset but by L2 input. The current study evaluates the prediction regarding late-acquired phenomena in the area of sentential semantics by examining how bilingual children understand sentences with temporal connectives in their L2 as a function of age of onset and length of L2 exposure. A group of six- to twelve-year-old children with L1 Greek and L2 German, varying widely in age of onset and length of L2 exposure, was tested on their comprehension of sentences with the connectives before and after in iconic and non-iconic order, using a picture-sequence selection task. Baseline data from monolingual German-speaking children showed that temporal connectives are a late-acquired phenomenon, still not mastered by age seven. Bilingual children’s L2 performance was not influenced by age of onset, despite its wide range, but by length of L2 exposure. This finding provides novel support for the Timing Hypothesis from temporal semantics.

This study explores how learning a Scandinavian language as an L3 affects word order in advanced L2 English, depending on the learner’s knowledge of the relevant structure in L3. We target verb-particle constructions, which cross-linguistically employ dissimilar word order patterns. In English and Norwegian, the particle can both precede and follow the object, and the preference is modulated by length of the NP. Conversely, the particle is obligatorily pre-posed in Swedish and always post-posed in Danish. We report acceptability judgement and self-paced reading data from instructed, non-immersed L1 Polish L2 English learners acquiring either L3 Norwegian (n = 65), Swedish (n = 52), or Danish (n = 46) from beginning to advanced levels, with the Norwegian group serving as a control. All learners are matched for L2 English proficiency. An analysis using linear mixed-effects models reveals regressive cross-linguistic influence in participants’ judgements and reading times in L2 English, reflected mainly in the post-posed particle being preferred by L3 Danish learners and dispreferred by L3 Swedish learners, particularly when target-like knowledge of the structure is in place. Our study is the first to show that L3 impacts word order in L2, thereby extending the observation of regressive cross-linguistic influence to syntax.

A developmental account of how bilinguals organise syntactic knowledge is crucial to understanding their mental representations. While adult studies suggest that syntactic representations can be shared across languages, evidence from child and adolescent heritage speakers remains limited and mixed. We conducted two syntactic priming experiments with adolescent heritage speakers of Polish in the UK (N = 35, mean age = 15;3) to test whether they would produce (1) relative clauses (RCs) instead of adjectival phrases for attributive relationships, and (2) possessor-second structures for possessive relationships with referential possessors, following cross-linguistic priming. A third experiment tested first-generation Polish-speaking adult immigrants in the UK (N = 32) on the same tasks. Adolescents were resistant to priming for RCs, whereas adults showed bidirectional priming. Both groups were primed to produce possessor-second structures only in Polish, where this is the canonical word order. Results indicate that increased proficiency and language experience facilitate priming for less frequent, complex structures like RCs, supporting shared syntax only in adults. For highly frequent constructions like possessives, where Polish and English differ in word order, priming occurred only when consistent with the language-specific preference. This underscores the role of frequency, canonicity, and complexity in shaping bilingual syntactic representations across development.

This study investigates whether heritage speakers of European Portuguese (EP) show a diachronically advanced behaviour in their comprehension of null object constructions. Based on a comprehension experiment, we compared heritage speakers to homeland speakers of EP, on the one hand and homeland speakers of Brazilian Portuguese (BP) on the other, the latter representing a variety in which null objects have diachronically spread. Our results confirm significant differences between EP homeland speakers and the two other groups but not between heritage speakers and BP homeland speakers, indicating an acceleration of diachronic change in the heritage speaker group. In addition, our study confirms that null objects in islands are available to homeland speakers of EP, challenging earlier analyses of null objects in this variety.