Inflectional morphology errors have been found to be prevalent in adult second language production, even by learners with high overall proficiency and extensive L2 exposure. Particular difficulties are observed with acquisition of complex inflectional systems by learners whose first languages are less morphologically rich. Thus, L1 English learners acquiring the Russian case and gender systems can provide a testing ground for competing approaches to L2 morphological acquisition. This study utilized the Russian Learner Corpus of Academic Writing, consisting of texts by advanced learners in Portland State University’s Russian Language Flagship program, to compare frequencies of case and gender-marking errors in timed versus untimed compositions by a sample of heritage and traditional L2 learners in order to gauge the putative effect of processing pressure on such errors.

It was predicted that significantly more frequent errors in timed compositions would support the position that inflectional morphology errors by advanced learners largely reflect processing difficulties under time pressure. However, results for the heritage group showed that while descriptive tendencies appeared to point toward processing difficulties, the difference between timed and untimed error rates did not reach statistical significance. In the L2 group, error rates were higher in students’ untimed texts than in their timed texts, although this difference also did not reach significance. The lack of reliable differences between timed and untimed error rates could be interpreted as demonstrating representational deficits in learners’ interlanguage grammar, particularly in the L2 group. However, the greater complexity (measured by words per T-unit) of the untimed essays provides an alternative explanation for the unexpected finding of a higher untimed error rate among the L2 learners, for whom the correlation between differences in complexity and error rates for timed and untimed texts approached significance. For the heritage learners, error rates appeared to be affected more by time pressure than text complexity. In addition, the heritage group had lower case-marking and significantly lower gender-marking error rates than the L2 group. This finding suggests that heritage learners are less likely than traditional L2 learners to show evidence of possible representational deficits of nominal functional features in their interlanguage grammar.