L1 Mandarin children’s acquisition of the “One + Classifier” structure: empirical evidence and insights from special language domain theory

IntroductionThis study examines early Mandarin acquisition of the yi “one” + classifier construction in monolingual Taiwanese Mandarin-speaking children and asks whether numeral-classifier development is holistic or staged.MethodsUsing the CHILDES-TCCM corpus, we analyzed 35,864 child utterances from 126 transcript files produced by ten children. We isolated 554 screened yi-target records, compared them with a matched caregiver-input baseline, and retained compressed adult headline materials as a qualitative contrast.ResultsOvert yi + classifier forms overwhelmingly dominated the child data: 543 of the 554 screened records were target-like. Only four clear omissions and two classifier-form errors were identified, and five additional tokens were excluded as ambiguous. Clear omissions were rare but concentrated in the 2;00, 3;00 age range and distributed across three children. Individual classifiers dominated throughout, and child output was even more strongly skewed toward individual classifiers than caregiver input. The classifier ge functioned as an early broad-use form, but post hoc review showed that only a small subset of yi ge + N tokens constituted clear overgeneralization.DiscussionThe comparison with headline data is retained as a qualitative contrast. It does not motivate a shared derivational mechanism, but shows that under strong discourse pressure, surface Num + N strings remain interpretable in Mandarin. Overall, the findings support a staged and non-holistic view of numeral-classifier development while treating the evidence for omission cautiously because of the small number of clear omission tokens.