Premium
Assessing the Validity of ACTFL Can‐Do Statements for Spoken Proficiency: A Rasch Analysis
Author(s) -
Tigchelaar Magda,
Bowles Ryan P.,
Winke Paula,
Gass Susan
Publication year - 2017
Publication title -
foreign language annals
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.258
H-Index - 49
eISSN - 1944-9720
pISSN - 0015-718X
DOI - 10.1111/flan.12286
Subject(s) - rasch model , language proficiency , psychology , construct validity , statement (logic) , construct (python library) , linguistics , polytomous rasch model , psychometrics , item response theory , mathematics education , clinical psychology , developmental psychology , computer science , philosophy , programming language
Abstract The NCSSFL‐ACTFL Can‐Do Statements describe what language learners can do at the various ACTFL proficiency sublevels. Unlike the European equivalent of the Can‐Do Statements (the Common European Framework of Reference for Languages), few researchers have assessed the construct validity of the NCSSFL‐ACTFL statements. Concerns have included whether the difficulty levels of the skills described in the statements match the statements’ assigned proficiency levels and whether each statement accurately indicates the underlying construct: language proficiency on the ACTFL subscales. This study addressed those two concerns. Undergraduate Spanish learners at an American university (N = 382) self‐assessed their speaking proficiency by responding to a selection of 50 NCSSFL‐ACTFL Can‐Do Statements. A Rasch analysis revealed 15 misfitting items that did not fit the model. The suspected reasons for the misfit were that the statements were vague, described experiences that the learners may not have had, or assessed multiple skills in a single statement. However, 35 statements fit the model. Discussed are how the NCSSFL‐ACTFL Can‐Do Statements can be used to self‐assess proficiency and how the statements should be assessed for content validity and psychometric value.