Final discussion

Robert Daland points out that amplitude is another cue to sentence boundaries - Amanda answers that indeed there are many others (such as initial strengthening), but the point remains that younger babies were less able to attend to single cues than older babies.
Jerry Jeager points out Levelt's model does not address actual production. Lisa G agrees that Levelt's model is useful, and further clarifies that she was trying to imply that there's a need to get converge in terms of categories with articulatory phonology.
Matt notes that S. Blumstein also found in L2 more transfer for non-words
* what is the nature of the mechanism governing attention?
* Lisa D. is asked whether she makes a distinction between 'lexical' and 'semantic'. She answers that, for the purposes of the task, it doesn't matter: in fact, it is possible that anything that 'made them care' would have helped.
Amanda (?) asks whether it is an explicit or implicit task - and whether an active versus a passive task would have made a difference.
Mary Beckman points out that perhaps the answer is having a *symbolic* category.
Jennifer Cole notes that we have to factor in that it is language - not any arbitrary cue can be used, since some things are unlearnable [I think she may have referred to tasks where you had to learn that a given rule applied only to female speakers, but not male ones - on this note, doesn't that actually happen in speech, that an alternation occurs only in one sociolect?]
Lisa D. points the effects of telling participants that they are about to hear a foreign language (e.g. Russian) actually improves their performance.
[I think there was a return to the question of frication versus vocalic information]
Matt wonders also about the fact that in speech social information is coded primarily in the vowels leading people to attend more to them, but notes that it is also in encoded consonants.
Amanda suggests that perhaps inventory size is a reason for it.
Lisa D. confirms this with her experience with Bengali speakers who... ??
Mary Beckman wonders how do we know why we attend to something - Jennifer Cole agrees: in fact, it seems so hard conceptually that we are able to converge on the right cues given the multidimensionality of speech. Jessica points out that we couldn't just throw out information that is not adjacent to the sound in question in cases of long distance dependencies such as vowel harmony. How does one determine the limits, the distance at which one will stop paying attention.
Ken Dejong (?) points out that we have been using the term 'attention' without really defining it. Attention could mean recruitment of resources and usage of information. There are two issues attached to this: we tend to underestimate people's ability to use something and, for development, we need to address how we learn to focus attention.
Matt asks whether it is just interference of categories that allows to focus attention. For instance, Dell did a study where people had to learn that engma appeared in syllable initial position, and it was really hard.
Mary Beckman relates this to babies' inability to learn that a nasalized vowel is followed by t, given that this sequence is illegal in English.
Amanda answers that perhaps it isn't given that sometimes the nasalized vowel is the only cue to the presence of a nasal consonant and this one is dropped.
Jessica remembers that when she heard the stimuli used in that study they sounded like vowel+nt sequences to her.
?? Ken Dejong asks Lisa G whether it is a question of salience - Lisa D adds whether it is mirroring
LG answers that when a novel word was at the end of the sentence, the novel word was not more variable, but the rest of the sentence was.