Update to MOR tagset?

[Erin Quirk]

Hello,
I've been working with three corpora from the CHILDES database
(Bloom70, Brown, and Valian) and I've come across some discrepancies
in the MOR tagset and am hoping to get some clarification. I'll try to
summarize them here:

In Eve (Brown) files 1-19, auxiliary do and all modals are tagged as
mod while auxiliaries have and be are tagged as aux. Here are some
examples:

does --> pro|it mod|do&3S ?
may --> mod|may pro|I v|have pro|it ?
have, be --> co|well pro:wh|what post|else aux|have pro|you aux|
be&PERF part|do-PROG ?

However, in Eve file 20 and all of Bloom70 and all of Valian, all
auxiliaries and modals are tagged as aux. Here are some examples:

does --> det|that pro:indef|one aux|do&3S~neg|not n|roll aux|do&3S pro|
it ?
may --> aux|may pro|I v|top pro|it ?
have, be --> co|say pro|you~aux|have aux|be&PERF part|drink-PROG
pro:poss:det|your n|eggnog aux|have~neg|not pro|you .

Also, in the CHAT manual currently available from the CHILDES site, on
page 109, the convention just mentioned is shown (i.e. modal
auxiliaries are tagged as aux, as in aux|can), but on page 107 a
different convention is shown. Table 45 says that "Auxiliary verbs,
including modals" are tagged as v:aux. This is repeated on page 109.

Which is the most up-to-date convention? If I had to choose, I would
say that modal auxiliaries should be tagged as mod while auxiliary do,
have and be, which can be inflected, ought to be tagged as aux with
the appropriate inflectional suffixes (e.g. aux|do&PAST).

Another discrepancy that is less problematic is the use of conj:sub
for subordinating conjunctions also in Eve files 1-19 while conj:subor
is used elsewhere. The manual recommends the use of conj:sub.

I appreciate your help clearing this up.
Thanks,
Erin Quirk