When you work on the discussion section of the CHILDES write-up, I ask you to compare your results to some that were reported by Valian (1991). However, one thing that I should really point out: The MLU counts you found in the first section of the lab are probably incomparable to Valian’s.

This is inconvenient, but the reason is this: The mlu command in the CHILDES transcript browser computes the Mean Length of Utterance in terms of average morphemes per utterance by default. Valian’s results, however, were stated in terms of average words per utterance. So, to really do the comparison properly—that is, to compare Nina’s behavior to what Valian’s subjects did—you will probably need to basically do the first step of the lab assignment again, except this time telling the mlu command to count words instead of morphemes.

The way you do this is to provide -t%mor as a parameter to the mlu command, which means “omit the %mor tier. That is, you want a parameter string like:

+t*CHI -t%mor nina0*

(“include only the CHI tier (child utterances), but exclude the %mor tier (morpheme-by-morpheme breakdown), for all of the files whose name begins with nina0.”)

Incidentally, an additional unnecessarily confusing aspect of this, the results are still reported as being a ratio of morphemes over utterances, even with the -t%mor option included, but what it means is the ratio of words to utterances in that case.

To have discovered this for yourself, you could have just pressed the Run button with no parameters after selecting mlu as the command; this displays the “help” summary for the mlu command, which contains the warning: “MLU NOW WORKS ON “%mor:” TIER BY DEFAULT. TO RUN MLU ON MAIN SPEAKER TIER PLEASE USE “-t%mor:” OPTION.” You might have had no reason to look, but it’s good to know how to get the help text in general.