KidEval

[Brian MacWhinney]

Dear Info-CHILDES,

    In the context of the AphasiaBank project, we have developed a system for automatic profiling from language transcripts called EVAL.  The predecessors of this system were the CLAN programs of MORTABLE which dumps out counts of all grammatical morphemes from the %mor line to rows in an Excel spreadsheet and MEASURES which does the same for additional things like MLU, TTR, pauses, retraces and so on.  MORTABLE and MEASURES are designed primarily for researchers who are conducting cross-sectional studies to speed up and synchronize their work.  In these programs, the basic idea is that rows in the Excel spreadsheet are participants and columns are measures. Computation of the various measures depends on accurate use of CHAT transcription and then running of the MOR program to create a %mor line.  All recent work in AphasiaBank and the majority of the corpora now in CHILDES have these features. 
    The next step has involved configuring this system into something more relevant to clinicians.  In that case, the idea was to focus not on output for groups, but for individual participants.  Here, what is particularly interesting is the ability to compare the participant with some reference group, perhaps normal controls or perhaps other people with Broca's aphasia.  For AphasiaBank, this is easy, because the whole project was designed to collect data in this format.  This system is called EVAL and it is now operational.  However, it is primarily conceived of as a method for studying people with aphasia.
    This use of reference databases can also be extended to child language, much as is done now in the SALT framework.  We are now working with Nan Bernstein Ratner to modify the MEASURES program for use with child language transcripts.  The new program would be called KidEval.  Here, we envision a combined usage, both with individual children and with groups.  In addition to the columns currently in MEASURES, we hope to add DSS, VOCD, IPSyn, and counts of the 14 morphemes of Brown (1973).  A big challenge here will be the addition of reference data sets.  In child language, there are so many possible reference sets, varying by age, language, bilingual status, topic, method of elicitation, and so on.  My sense is that we want to make available all possible reference datasets and allow the user to select the ones that best match the current child or group of children.  So, this will be a big project.
    This same method could also possibly be extended to the PHON program that Yvan Rose and Greg Hedlund have built.  Again, either individual children or groups of children would be compared against some standard set of control or norm data for phonological development.  
  We would love to receive suggestions regarding this project, including ideas about reference databases and additional automatic measures,  either posted to the list or sent directly to me, Nan, or Yvan, depending on the focus.

-- Brian MacWhinney

-- 
You received this message because you are subscribed to the Google Groups "Info-CHILDES" group.
To post to this group, send email to info-childes at googlegroups.com.
To unsubscribe from this group, send email to info-childes+unsubscribe at googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.