coding t-units

[Ngoni Chipere]

I would like to measure sentence length in a corpus of children's writing.
The punctuation is unreliable and so I have to annotate the sentence
boundaries myself. I've decided to use the T-unit measure and I would
welcome some feedback on the following queries. I'm sure someone has
grappled with these problems before and perhaps even written about them.

1) How to deal with reported speech, e.g. I said, "I'm going". Is it two
t-units (main clause + main clause) or one t-unit (main clause + subordinate
clause)?

2) How to deal with single-word dialogues, e.g.

"Yes!"
"No!"
"Yes!"
"Okay"

The problem here is not so much the direct speech is not introduced but that
it does not constitute a clause.

Many thanks,

 Ngoni


*********************************************************
Dr Ngoni Chipere
Research Fellow
School of Education, University of Reading
Bulmershe Court, Earley, Reading, RG6 1HY, UK
tel 0118 9875123 ext 4943