Skip to Main Content

Labelers were given examples of each category, and annotated with access to both transcript and speech source. The guidelines used by the annotators are presented in Appendix B. Appendix C includes some examples of each class of ACWs, as labeled by our annotators. Inter-labeler reliability was measured by Fleiss's κ (Fleiss 1971) as Substantial at 0.745.2 We define the majority label of a token as the label chosen for that token by at least two of the three labelers; we assign the “?” label to a token either when its majority label is “?”, or when it was assigned a different label by each labeler. Of the 5,456 affirmative cue words in the corpus, 5,185 (95%) have a majority label other than “?.” Table 2 shows the distribution of discourse/pragmatic functions over ACWs in the whole corpus.

Table 2

Distribution of function over ACW. Rest = {gotcha, huh, yep, yes, yup}.


alright
mm-hm
okay
right
uh-huh
yeah
Rest
Total
Agr 76 58 1,092 111 18 754 116 2,225 
BC 395 120 14 148 69 757 
CBeg 83 543 630 
CEnd 12 
PBeg 65 69 
PEnd 11 12 218 20 15 278 
Mod 18 1,069 1,092 
BTsk 32 40 
Chk 49 63 
Stl 15 19 
36 12 150 10 55 271 
Total 236 478 2,265 1,258 169 903 147 5,456 

alright
mm-hm
okay
right
uh-huh
yeah
Rest
Total
Agr 76 58 1,092 111 18 754 116 2,225 
BC 395 120 14 148 69 757 
CBeg 83 543 630 
CEnd 12 
PBeg 65 69 
PEnd 11 12 218 20 15 278 
Mod 18 1,069 1,092 
BTsk 32 40 
Chk 49 63 
Stl 15 19 
36 12 150 10 55 271 
Total 236 478 2,265 1,258 169 903 147 5,456 

Close Modal

or Create an Account

Close Modal
Close Modal