6 Exercises
8.3 CHIP
8.3.4 Summary Measures
In addition to analyzing utterances and creating separate dependent tiers, CHIP also produces a set of summary measures. These measures include absolute and proportional values for each of the coding categories for each speaker type that are outlined below. The definition of each of these measures is as follows. In these codes, the asterisk stands for any one of the four basic operations of ADD, DEL, EXA, and SUB.
Total # of Utterances The number of utterances for all speakers regardless of the number of intervening utterances and speaker identification. Total Responses The total number of responses for each speaker type regardless
of amount of overlap.
Overlap The number of responses in which there is an overlap of at least one word stem in the source and response utterances.
No Overlap The number of responses in which there is NO overlap between the source and response utterances.
Avg_Dist The sum of the DIST values divided by the total number of overlapping utterances.
%_Overlap The percentage of overlapping responses over the total number of responses.
Rep_Index Average proportion of repetition between the source and re- sponse utterance across all the overlapping responses in the data.
*_OPS The total (absolute) number of add, delete, exact, or substitution operations for all overlapping utterance pairs in the data.
%_*_OPS The numerator in these percentages is the operator being tracked and the denominator is the sum of all four operator types.
*_WORD The total (absolute) number of add, delete, exact, or substitution words for all overlapping utterance pairs in the data.
%_*_WORDS The numerator in these percentages is the word operator being tracked and the denominator is the sum of all four word operator types.
MORPH_* The total number of morphological changes on exactlymatching stems.
%_MORPH_* The total number of morphological changes divided by the number of exactly matching stems.
AV_WORD_* The average number of words per operation across all the over- lapping utterance pairs in the data.
FRONTED The number of lexical items from the word list that have been fronted.
EXACT The number of exactly matching responses.
EXPAN The number of responses containing only exact matches and additions.
REDUC The number of responses containing only exact-matches and deletions.
SUBST The number of responses containing only exact matches and substitutions.
8.3.5 Unique Options
+b Specify that speaker ID S is an “adult.” The speaker does not actually have to be an adult. The “b” simply indicates a way of keeping track of one of the speakers.
+c Specify that speaker ID S is a “child.” The speaker does not actually have to be a child. The “c” simply indicates a way of keeping track of one of the speakers.
+d Using +d with no further number outputs only coding tiers, which are useful for iterative analyses. Using +d1 outputs only summary statistics, which can then be sent to a statistical program.
+g Enable the substitution option. This option is meaningful in the presence of a word list in a file specified by the +h/-h switch, because substitutions are coded with respect to this list.
+h Use a word list file. The target file is specified after the letter “h.” Words to be included (with +h) or excluded (with -h) are searched for in the target file. The use of an include file enables CHIP to compare ADD and DEL categories for any utterance pair analyses to determine if there are substitutions within word classes. For example, the use of a file containing a list of pronouns would enable CHIP to determine that the instances of ADD of “I” and DEL of “you” across a source and response utterance are substitutions within a word class. Standard CLAN wildcards may be used anywhere in the word list. When the transcript uses CHAT-style morphological coding (e.g., I-’ve), only words from the word list file will match to stems in the transcript. In other words, specific morphology may not be traced within a word list analysis. Note that all of the operation and word-based summary statistics are tabulated with respect to the word list only. The word list option may be used for any research purpose including grammatical word classes, number terms, color terms, or mental verbs. Note also that the -h option is useful for excluding certain terms such as “okay” or “yeah” from the analysis. Doing this often improves the ability of the program to pick up matching utterances.
+n This switch has three values: +nb, +nc, and +ns. See the examples given earlier for a discussion of the use of these switches in combination.
+qN Set the utterance window to N utterances. The default window is seven
utterances. CHIP identifies the source-response utterances pairs to code. When a response is encountered, the program works backwards (through a window determined by the +q option) until it identifies the most recent potential source utterance. Only one source utterance is coded for each response utterance. Once the source-response pair has been identified, a simple matching procedure is performed.
+x Set the minimum repetition index for coding.
CHIP also uses several options that are shared with other commands. For a complete list of options for a command, type the name of the command followed by a carriage return in the Commands window. Information regarding the additional options shared across commands can be found in the chapter on Options.