Establishing A Conceptual Framework Of Designing Picture-based Writing Tasks

Amidst the growing development of multiliteracies and multimodality in 21~stt century around the world,the use of multimodal resources in language learning,teaching and assessment has been advocated in curriculum requirements,thus occupying a place in designing instructional materials/activities and language tests.Language tests have thus seen an increasing use of visuals like graphs or pictures as stimulus materials in recent years.Due to concerns of validity and reliability in language testing,using multimodal resources as test task input merits careful examination.Yet,the design of such tasks as a whole has received scant research attention.Inspired by existent graph-based writing studies as a result of the rise of academic English writing,the researcher proposed the term“picture-based writing”(PBW)to investigate the far less researched area of using pictures in writing tests.The present research aimed to develop a conceptual framework of designing PBW tasks,addressing three primary research questions:(1)What design parameters should be incorporated in a framework ofdesigning PBW tasks?(2)How do test takers perceive PBW tasks designed in light of the framework?(3)To what extent do the design parameters affect test takers’writingperformance,if any?After reviewing models of multimodal information processing and theories of writing task design,the researcher proposed a conceptual framework consisting of four major task variables from test designers’perspective interacting with two task variables from test takers’perspective,consistent with test task characteristics—input and response—proposed by Bachman and Palmer.The parameters pinpointed in the framework include topic,nature of image,pictorial feature,and interaction between picture prompt and rubric,which were in turn taken into account in the design of two types of PBW tasks.To address the second and the third primary research questions,an empirical study was designed to examine both the attitudes of test takers towards PBW tasks as opposed to non-PBW tasks and the effects of those design parameters on test takers’writing performance in terms of the writing product and the writing process.A three-level-mixed-methods research design has been adopted,moving from survey analysis to writing product analysis to writing process analysis.At the first level,a large-scale survey was conducted among 711 questionnaire respondents from a university who were required to sequence two different PBW task types and their non-PBW counterpart in terms of preferences for the three tasks.The results showed that PBW tasks received much greater preference than non-PBW task,yet the first PBW task type requiring test takers to write on the basis of their own interpretations of the picture prompt was far less favored than the second PBW task type that specifies for test takers an idea to develop while giving the picture prompt.The next level of research focused on the effects of design parameters embodied in the two PBW task types on test takers’writing product using statistical analysis of writing scores and textual analysis of writing scripts.Univariate ANOVAs were conducted on writing scores assigned to the writings of 474 matched participants performing six PBW tasks on three topics(Topic A,B,and C).Results showed that no significant statistical differences were found across the three topics,indicating no topical effect because of the meticulous selection of topics related to test takers.In addition,the two PBW task types exerted no statistically significant effects on test takers’writing scores across three topics except for Topic A.The interpretation of the picture prompts related to the three topics may have constituted a source of difference due to the interaction between the salient pictorial feature of verbal cue and the rubric.The last analysis at the product level,employing two textual analysis tools of LIWC and Antconc,was directed to the lexical aspect of writing scripts collected from 80matched participants performing two other PBW tasks specifying the same idea yet using two different types of image—an abstract picture and a concrete one.LIWC results showed that the abstract picture contributed to greater use of emotion-related words while the concrete picture brought about more social words.Meanwhile,Antconc analysis revealed that different picture prompts contributed to different frequency of certain classes of words,pronouns in particular,in the two corpora.The effects of design parameters on test takers’writing process were analyzed at the third level of case studies using eye-tracking(ET)technology complemented by think-aloud-protocols(TAPs).Two series of eye-tracking experiments—verbal cue series and task type—series were conducted to examine the planning-specific cognitive processes of test takers performing PBW tasks.It showed that the picture prompts affected the planning process of test takers of varying proficiency levels in both the attentional allocation quantified by ET data and the planning progress from initial task representation to advance planning demonstrated by TAPs data.It is concluded that the design of PBW tasks deserves more efforts to expand the research scope in order to develop a complete and dynamic framework in the future.