270 likes | 501 Views
How t o Write a Reliable Listening Comprehension Test ?. Zhenlin Qiao Defense Language Institute April 16, 2009. Characteristics of oral communication Aspects of listening comprehension Defining the construct Approaches to assess listening Selecting text Creating tasks
E N D
How to Write a Reliable Listening Comprehension Test? Zhenlin Qiao Defense Language Institute April 16, 2009
Characteristics of oral communication Aspects of listening comprehension Defining the construct Approaches to assess listening Selecting text Creating tasks Properties of a good test Toward a Better Listening Comprehension Test
Encoded in the form of sound, acoustic signal Linear and takes place in real time with no chance to review Linguistically different from written language Phonological modification Stress and intonation to indicate clausal boundaries Redundancy, hesitation, accent Various speech rate Discourse structure Shared knowledge Non-verbal signals Characteristics of Oral Communication
Listening comprehension is an active construction of meaning, and that this is done by applying knowledge to the incoming sound. (Buck, 2001) Aspects of Listening Comprehension
Declarative knowledge(Anderson, 1976) Procedural knowledge(Anderson, 1976) Spreading activation(Cole and Jakimik, 1980) Processing idea unit(Sachs, 1967) Processing connected discourse/cohesion (Halliday and Hasan, 1976) Applying the knowledge of language
Inference: - propositional inference (Hildyardand Olson, 1978) - enabling inference (Hildyardand Olson, 1978) - pragmatic inference (Hildyardand Olson, 1978) - bridging inference(Haviland and Clark 1974) - elaborative inference Script theory (Schank and Abelson, 1977) Schemata (Rumelhart and Ortony, 1977;Rumelhart, 1980) Using world knowledge
Sociolinguistic appropriacy (Hymes, 1972) Pragmatic interpretation (Leech, 1983) Speech acts (Austin, 1962; Searle, 1969) Grice’s Maxim (Grice, 1975): - maxim of quantity - maxim of quality - maxim of relation - maxim of manner Principles of analogy and minimal change (Brown and Yule, 1983) The cognitive environment (Sperber and Wilson, 1986 The context of communication
The flow chart approach (Nagle and Sanders, 1986) 1) echoic memory 2) working memory 3) long-term memory Mental modes: there are 3 main ways in which texts can be represented mentally (Johnson-Laird, 1983;1985): 1) as proposition 2) as mental models 3) or as image Building mental representations of meaning
The thing we are trying to measure is called construct (Buck, 2002). Construct validity is to make a test somehow measures that construct (Buck, 2002). Messic(1989, 1994): - construct-underrepresention - construct-irrelevant variance Defining the Construct
General language proficiency Assessing achievement Diagnostic testing Purpose of assessing listening comprehension
Collaborative or non-collaborative listening Differing interpretations of texts Intervening variables Available resources Practical constraints
Competence-based construct Task-based construct A construct based on the interaction between competence and task A default listening construct Developing the construct
Competence: The knowledge, skills and abilities we think test takers should have. Knowledge here refers to procedural knowledge- the ability to apply the knowledge in efficient and automatic language processing. Language competence Strategic competence Competence-based construct
Grammatical knowledge Discourse knowledge Pragmatic knowledge Sociolinguistic knowledge Language competence
Cognitive strategies - comprehension processes - storing and memory processes - using and retrieval processes Metacongnitive strategies - assessing the situation - monitoring - self-evaluating - self-testing Strategic competence
Discrete-point approach (Lado, 1961) Integrative approach (Oller, 1979) Communicative approach (Caroll, 1972) Approaches to Assess Listening
Providing suitable texts - assimilated/authentic - topics - levels of difficulty Ensuring good quality sound - speech rate - background noise Selecting Text
Task Characteristics: Setting (place)/time of day Test rubric - instruction - structure - time allotment - scoring Creating Tasks
Format (long/short texts) Language of input Topical knowledge a. use tasks that depend on knowledge that everyone has; b. use tasks that depend on knowledge that no one has, c. use tasks that depend on knowledge that has been provided in the test Input characteristics
Retention tasks Conversation tasks Self-evident comprehension tasks (true/false) Picture tasks Body movement tasks Tasks for testing understanding of literal meanings
Understanding gist General passage comprehension Information transfer tasks (e.g. following direction on a map; filling in information in a grid) Tasks for going beyond local literal meanings
Short answer/open-ended Multiple-choice True/false Inference question: The main problem is that it is very difficult to say with certainty that any inference is completely wrong. Question types
Usefulnessincorporates the test’s measurement properties, as well as the social consequences of test use and the practicality of the test for its particular purpose. It is a function of six properties: reliability, construct validity, authenticity, interactiveness, impact and practicality. (Bachman and Palmer, 1996) Properties of a Good Test
It is concerned with how accurately the test measures. We should always get the same result testing the same person in the same situation. The more important the decision based on the test, the more reliable the test needs to be. Reliability
It is the extent to which the test measures the right construct. Construct validity
Authenticity Interactiveness Impact Practicality