In this task-repetition intervention study, L2 learners' reuse of linguistic constructions was analyzed to investigate to what extent recurring reliance on specific constructions during the same task repetition predicts fluency development. English-as-a-foreign-language (EFL) learners performed oral narrative tasks three times per day under two task repetition schedules: blocked (Day 1: Prompt A-A-A, Day 2: B-B-B, Day 3: C-C-C) versus interleaved (Day 1: Prompt A-B-C, Day 2: A-B-C, Day 3: A-B-C). From a usage-based perspective, their reuse of constructions across the same prompt was examined at both concrete (lexical unigram [e.g., "bicycle"] and trigram [e.g., "behind the bicycle"]) and abstract (parts of speech trigram [e.g., "preposition determiner noun"]) level. Subsequent analyses revealed that blocked practice led to higher reuse of both concrete and abstract constructions than interleaved practice. Reuse frequency was correlated with during-training and pretest-posttest fluency changes. Particularly, greater reuse of lexical and abstract trigrams during interleaved practice led to improvements in speed and breakdown fluency (i.e., shorter mean syllable duration and fewer mid-clause pauses) after the intervention, albeit with higher effort (indicated by longer mid-clause and clause-final pauses). Taken together, these findings indicate that manipulating task-repetition schedule may systematically induce reuse of linguistic constructions, which may promote proceduralization (entrenchment) of constructional knowledge at both concrete and abstract levels.