Italian grammatical gender stays incompletely mastered by L2 learners and heritage speakers alike, and a gendered first language helps only conditionally.¶

Claim · All claims
Evidence · 1 study · associational r? · q2 · n=not reported here
Currency · newest study 2025 · not yet checked against its sources · not reviewed by a person

Having gender in your first language does not transfer as a rule. The advantage it confers is specific to particular item configurations, and it does not appear in every task.

Subclaims¶

q2 i? Across every configuration tested, gender knowledge on third-person singular and plural object clitics was incomplete for both L2 learners and heritage speakers. → Bäck Romano et al. 2025

q2 i? Typological proximity helped only conditionally: Spanish–Italian groups were advantaged where the gender-assigning nouns were non-cognate and gender-incongruent, Swedish–Italian groups where items were gender-incongruent cognates. → Bäck Romano et al. 2025

q2 i? The advantages were moderated by intrinsic differences between heritage and L2 speakers, and appeared in production only. → Bäck Romano et al. 2025

Evidence¶

Bäck Romano et al. 2025¶

Bäck Romano, S., Guijarro-Fuentes, P., Calpe-Álvarez, V., & Rivera, J. (2025). Gender in L2 and heritage Italian by Swedish and Spanish dominant speakers. Italiano LinguaDue, 17(2), 27–48. doi:10.54103/2037-3597/30422

q2 · between-groups comparison across two language pairings and two speaker types · i? · direction of advantage reported, error percentages not given · n=not reported here · associational · r?

Spanish–Italian and Swedish–Italian bilinguals — L2 learners and heritage speakers in each pairing — were compared on gender assignment to third-person singular and plural object clitics, crossing cognate status against gender congruency. The paper reports which configurations favoured which group and that the effects were production-only; it does not report error rates, so the study establishes that the difficulty persists rather than how large it is.

Discussion¶

Gender does not generalise from a rule, even for learners whose first language has gender. It behaves like item knowledge — a fact about a word rather than a procedure applied to it — which argues for holding it per lexeme rather than as one grammar objective averaging hundreds of independent facts.

Heritage and L2 speakers are different populations, not points on one scale. A heritage speaker may assign gender correctly by ear while unable to state a rule, and an L2 learner the reverse. A single mastery number over both channels describes neither. The wiki's existing treatment of heritage language is about its value rather than its structure — see Preserving a bilingual student's heritage language supports rather than hinders English acquisition [+M] — and this claim is the complement: same population, a place where the two channels come apart.

Same shape as the phonology finding, one level up. In A learner's first language predicts L2 phoneme perception more strongly than their proficiency level does [+S] the L1 helps when it encodes the relevant distinction. Here it helps only in some item configurations, which is a weaker and more awkward result to design around: you cannot read the advantage off the learner's L1 alone.

Limits. Object clitics specifically, two language pairings, production-only effects, no error percentages reported.

Related Claims¶

• A learner's first language predicts L2 phoneme perception more strongly than their proficiency level does — the same conditional-transfer shape in phonology, with effect sizes
• Preserving a bilingual student's heritage language supports rather than hinders English acquisition — the other claim in this wiki about heritage speakers
• Fluent bilingualism is linked to metalinguistic awareness of how the two languages relate — bilingualism as an asset, where this is a place it does not straightforwardly help
• Form-focused instruction on noun endings significantly increases French immersion students' ability to correctly assign grammatical gender, with gains maintained on delayed posttests — related