ACQ Vol 13 No1 2011

Communicative efficiency was measured by the number of correct information units per minute (CIU/min) relating to how quickly and correctly each topic was produced. The procedure for calculating CIU/min followed Nicholas and Brookshire’s (1993) rules for analysis. Breakdown in language production was investigated at word and utterance levels. The percent of words in mazes (revisions, repetitions, and filler words) (Miller & Chapman, 2002), number of word errors (Word errs) and number of utterance level errors (Utt errs) were measured. The number of dysfluencies as indicated by the percentage of words in mazes may be an indication of the participant attempting to correct difficulties in communication either before speaking, or once she had started speaking (Merlo & Mansur, 2004). Word errors occurred when an incorrect word was produced. Utterance errors included utterances that provided incorrect information, or did not add to the overall flow of the discourse (Ciccone, 2003). These were coded during transcription and calculated as the total number of each type of error contained in each sample. The amount of content recalled by the participant was measured by the percent of predetermined main ideas (%MI) and optional ideas (%OI) (Li et al., 1995). Results Table 3 provides an example of the procedure, for a familiar and an unfamiliar topic, played to the participant as well as the participant’s corresponding discourse samples. For each measure, the results were grouped according to familiarity or unfamiliarity to allow for statistical comparison (see table 4 for the results). Comparisons were undertaken using a paired-samples t- test or a Mann-Whitney U test when assumptions of normality and homogeneity of variance were violated. The alpha level was set at 0.05. Significant differences in the discourse measures between familiar and unfamiliar topics were noted. The unfamiliar topics resulted in a reduction in the speed and accuracy of discourse production. The slower rate of production was characterised by a decreased number of words per minute, an increase in the total amount of pause time, and a reduced number of correct information units per minute. The samples also contained a larger number of utterances that provided incorrect information or information that was conveyed ineffectively. These utterances had an increased number of filler words, repetitions and revisions. The unfamiliar samples also contained fewer optional ideas. Discussion There were statistically significant differences between the discourse samples produced in response to topics rated as familiar and unfamiliar. The more familiar topics resulted in higher quality discourse samples. The number of main ideas recalled was similar for both familiar and unfamiliar samples. This result is consistent with the findings of Li et al. (1995) and when considered in light of the significant difference in the number of optional ideas recalled suggests that the unfamiliar topics had an impact on the participant’s ability to recall all procedural details (Williams et al., 1994). It may also be evidence of the individual’s lack of previous exposure to the experiences outlined in the unfamiliar procedural topics. No significant differences were found between familiar and unfamiliar topics on the measures of mean length of utterance (MLU), type token ratio, and the number of word errors. Williams et al. (1994) found the syntactic complexity of the utterances increased when participants produced

Table 2. Complete list of familiar and unfamiliar topics Familiar topics Unfamiliar topics Going grocery shopping Going mountain climbing Going out to dinner Saddling a horse Clearing the table after dinner Making a clay bowl Getting children ready for bed Making a bean bag Getting a haircut Painting a watercolour landscape

Changing bed sheets Making a cup of tea

Participating in a marathon walk

Writing a haiku poem Auditioning for a play Conducting a symphony Preparing to scuba dive

Having a shower

Going to the doctors Making a sandwich

average of 6 utterances, with 11 words per utterance, and included an average of 7 main ideas and 3 optional ideas. The number of main and optional ideas were predetermined, consistent with Williams et al.’s study (1994). Average word frequency was calculated for all topics using the word frequency lists from the MRC psycholinguistic database (Wilson, 1998). A t-test showed no statistically significant difference in word frequency between familiar and unfamiliar topics t (18, 17.21) = .137, p = 0.715. Procedure The participant attended two 60-minute data collection sessions conducted by the first author. The first session involved collection of case history information, the completion of BDAE, and the ranking of procedural topics. During the second session, the 10 pre-recorded discourse samples were presented via a laptop computer to the participant in a random order. Presentation of stimuli and instructions was consistent with the retell tasks in Williams et al.’s (1994) and Li et al.’s (1995) studies. After listening to the discourse sample once, the participant retold the procedure in her own words. For each discourse topic, she was prompted to provide as much detail as she could recall. During discourse production nonspecific prompting was used such as “can you tell me anything else?” to encourage as much output as possible for each topic. Samples were recorded and timed using a JNC USB-350 digital voice recorder with a lapel microphone. Discourse samples were transcribed using Systematic Analysis of Language Transcripts software (SALT; Miller & Chapman, 2002) and analysed by the first author. Discourse analyses The discourse samples were analysed using the measures outlined below. Mean length of utterance (MLU) measured in words and type token ratio (TTR) were calculated. Mean length of utterance is a measure of syntactic complexity (Miller & Chapman, 2002). TTR is a ratio of the number of different words produced compared to the total number of words produced and reflects diversity in the lexical items produced in response to the discourse topics. Speech rate was measured by the number of words produced per minute (WPM) which reflects the speed with which the participant was able to formulate and produce the language required for each sample. The amount of pause time compared to the total discourse time was calculated as a percentage figure (%pauses). This percentage reflected the amount of additional time required to formulate the language output.

9

ACQ Volume 13, Number 1 2011

www.speechpathologyaustralia.org.au