Are Uninterpretable Features Unlearnable? Revisiting the Interpretability Hypothesis in Second Language English Acquisition

The Interpretability Hypothesis (IH) predicts that uninterpretable features absent in the L1 become inaccessible in adulthood, whereas Full Access theories maintain that Universal Grammar (UG) remains available. Despite extensive research, it remains unclear whether elevated resumption acceptance in L2 learners reflects a representational deficit or processing-based strategies, and no study has examined the role of working memory in this distinction for Saudi learners. This study tested whether Saudi learners of English show L1 transfer in accepting resumptive pronouns (RPs), grammatical in Arabic but ungrammatical in English, and whether this reflects representational deficits or processing limits. A total of 164 Saudi learners and 45 native English speakers completed a grammaticality judgment task employing a 2 × 3 × 2 factorial design (35 English relative clauses (RC) manipulating gap vs resumption, RC type, and relativizer) plus a working memory (WM) task. Saudi learners accepted ungrammatical RPs more often than natives (29% vs 3%) but still rejected them on most trials (71%), which is inconsistent with categorical IH predictions. Learners performed similarly on that- and who-relatives (77% vs 76%), indicating acquisition of the uninterpretable [wh] feature, and showed a universal subject advantage without selective difficulty in object-of-preposition relatives. WM predicted overall accuracy, but group differences in RP acceptance remained, suggesting language-specific transfer rather than domain-general cognitive limitations. Overall, the results support Full Transfer/Full Access models and challenge strong IH claims.