subordinate clauses were identified and counted.
To assess the second language performance of the two participants in terms of accuracy, complexity and fluency, the following measures were used( see summary in Table 1):
1. Measures of accuracy
The percentage of error-free AS units was calculated. Self-corrections were identified and counted. Error types were identified, distinguishing between lexical, phonological and syntactical errors. Errors per 100 words were also calculated.
2. Measures of complexity
A number of different measures were used. To capture syntactic complexity, the number of subordinate clauses was divided by the number of AS-units, as a higher proportion of subordination is generally perceived as a sign of increased complexity( Norris & Ortega, 2009). Another measure of syntactic complexity used was the average length of AS-units. This was calculated by dividing the total number of words by the number of AS-units. Finally, a more finely-grained analysis of complexity was conducted by analysing the range of verbs, tenses and moods used by both learners as well as the type and range of clauses.
3. Measures of fluency
Second language fluency is generally perceived as the ability to produce second language speech at a normal rate without undue pauses and interruptions( Ellis & Barkhuizen, 2005). To this end, all pauses were identified and measured. A distinction was made between short pauses( no longer than 2 seconds) and long pauses( over a second).
To measure speech production fluency, the number of false starts, hesitations and pauses over 2 seconds were counted and divided by the total number of words produced.
In addition, the whole conversation section( 7 minutes) was marked by two experienced assessors using the official VCAA assessment criteria and guidelines, which consist of three criteria of 10 points each( VCAA, 2010: 16-19). The first criterion focuses on communication and is divided into two subcategories:‘ the capacity to maintain and advance the exchange appropriately and effectively’ and‘ clarity of expression’ rendered through‘ pronunciation, intonation, stress and tempo’. The second one is concerned with content:‘ relevance, breadth and depth of information, opinions and ideas’, whilst the third focuses on‘ language’, which encompasses‘ accuracy of vocabulary and grammar’ and‘ range and appropriateness of vocabulary and grammar’.
RESULTS
Overall performance
Overall, Patrick produced longer and more complex AS-units and spoke more than Olivia( Table 2).
Accuracy
Both learners had similar relatively low accuracy scores in terms of error-free AS-units( Table 3), but Patrick had a slightly higher accuracy score in terms of errors per words( Table 4). Table 4 also shows that both learners had a similar number of lexical errors. The mastery of articles( gender and form) and preposition seemed difficult for both learners. Olivia used half as many prepositions as Patrick. Olivia seemed to use a very limited number of prepositions, and avoided agreement errors by using the impersonal c’ est + noun / adjective which does not require the adjective or noun to agree. This expression was indeed used 14 times in Olivia’ s speech production out of 35 conjugated verbs in total( see Table 6 & 7). Nonetheless, she used the wrong verb form on three occasions whilst Patrick managed to use the correct verb form each time. Despite Patrick’ s clearly superior mastery of the language overall, both learners struggled with articles. Patrick for instance failed to contract de into du with the definite article le on two occasions( line 5, and 32). Like Olivia, he also used the wrong type of article twice( an indefinite instead of a definite, line 8 and 25 for Patrick and line 9, 16 and 25 for Olivia). They also both struggled with article-noun agreement( line 18 and 21 for Olivia and line 10 for Patrick).
Interestingly, although both performances presented opportunities for grammatical and lexical self-correction, the learners made very few such corrections( see Table 3). All three attempted corrections were grammatical.
Complexity
In terms of syntactic complexity, Table 5 indicates that Patrick’ s language was linguistically more complex. Patrick’ s transcript had many more subordinate clauses, thus yielding a higher subordinate / AS-unit score( 36.3 % compared to only 12.9 % for Olivia), and the average length of his AS-units was 12.5 words compared to 8.2 words for Olivia.
Results of the analysis of verb forms, presented in Table 6 and 7, also signal differences in complexity. In Olivia’ s text, most verbs( 32 out of 35) were in the present tense. Almost half of these verbs were in the third person impersonal form. Such verb constructions are easier to conjugate because they can be used as chunks( e. g. c’ est). The other half were in the first person singular, and these are also easier to formulate because they are used more frequently by students and are thus more familiar than plural
INTERESTINGLY, ALTHOUGH BOTH PERFORMANCES PRESENTED OPPORTUNITIES FOR GRAMMATICAL AND LEXICAL SELF- CORRECTION, THE LEARNERS MADE VERY FEW SUCH CORRECTIONS.
Volume 47 Number 2 7