MLU counts

[Dale, Philip S.]

I, too, think it is essential to distinguish the within-language use of MLU or similar measures from the cross-linguistic use. But the latter use has a whole set of challenges, of which the primary one is knowing what is the proper criterion or set of criteria for judging how satisfactory the measure is. That is, if we are using the measure to "line up" children acquiring different languages, how do we do know we're doing it appropriately? We can't use chronological age as a benchmark (i.e., the relation of the measure to age), because that assumes a uniformity of rate of acquisition of grammar across languages. We can't use rate of vocabulary growth as the benchmark, even though there is a high correlation of grammar with vocabulary, because rates of vocabulary growth show some substantial variation across language (compare English, Chinese, Danish). In general, we can't use a comparison of similar, specific grammatical forms across languages (i.e., do suffixes expressing tense/aspect appear at the same level of the measure?), because the most common use of such a measure across languages is precisely to compare *different* forms expressing similar meanings. 

I'd be interesting in hearing suggestions about proposed evaluation criteria for a new measure.

Philip Dale