first words

[Brian MacWhinney]

Dear Info-CHILDES,
    Thanks to Alison for posting the clarification regarding her son  
Tilden's production of "bears" at 4 months and "elk" and "igloo"  
shortly thereafter.  These are indeed real months, not months after  
the beginning of speech as I (perhaps facetiously) had suggested.
    What happens to early word learning theory if we take  
observations of this type seriously (as I think we should)?  I see  
some possibilities:
1.  The observations are correct, but the actual productions were so  
sporadic and rare that they can be dismissed as chance combinations.
2.  Tilden actually heard the words in the adult speech and they  
managed to creep into his babbling repertoire as "amalgams" or  
"frozen forms" copied in their entirety.
3.  There was some subliminal shaping going on through which Tilden  
said something like "bears" or "elk" and Alison and her husband then  
latched onto this and shaped up production of the sound.

Until this possible phenomenon is more fully documented, it would be  
premature to even attempt to decide between such possibilities.  But,  
in principle, one can easily imagine words being learned as sound  
forms long before they are learned as meaningful sound-meaning  
associations.  It seems to me that this is exactly what the recent  
burst of interest in statistical learning would predict.  If children  
are doing such great segmentation, shouldn't they be storing the  
results of the segmentation as raw sound forms?  And if a particular  
child, such as Tilden perhaps, is rather good at auditory- 
articulatory matching or mapping, then that child could indeed  
produce such "words" long before the onset of the first real word.

I think I observed something like this in my older boy Ross, but it  
is the type of thing that, if you see it, you tell yourself you must  
be dreaming.  I wonder if anyone else besides Alison has spotted this?

Brian MacWhinney