Identifying Second Language Speech Tasks and Ability Levels for Successful Nurse Oral Interaction with Patients in a Linguistic Minority Setting: An Instrument Development Project

Abstract

One of the most demanding situations for members of linguistic minorities is a conversation between a health professional and a patient, a situation that frequently arises for linguistic minority groups in North America, Europe, and elsewhere. The present study reports on the construction of an oral interaction scale for nurses serving linguistic minorities in their second language (L2). A mixed methods approach was used to identify and validate a set of speech activities relating to nurse interactions with patients and to derive the L2 ability required to carry out those tasks. The research included an extensive literature review, the development of an initial list of speech tasks, and validation of this list with a nurse focus group. The retained speech tasks were then developed into a questionnaire and administered to 133 Quebec nurses who assessed each speech task for difficulty in an L2 context. Results were submitted to Rasch analysis and calibrated with reference to the Canadian Language Benchmarks, and the constructs underlying the speech tasks were identified through exploratory and confirmatory factor analyses. Results showed that speech tasks dealing with emotional aspects of caregiving and conveying health-specific information were reported as being the most demanding in terms of L2 ability, and the most strongly associated with L2 ability required for nurse–patient interactions. Implications are discussed with respect to the development and use of assessment instruments to facilitate L2 workplace training for health care professionals.

ACKNOWLEDGMENTS

This research was made possible by support to the Health-Care Access for Linguistic Minorities (H-CALM) research team, which is part of the Training and Human Resources Development Project (THRDP) based at McGill University, funded by Health Canada. The research was also supported in part by a grant from the Social Sciences and Humanities Research Council of Canada to NS and CT. Some of the data from this study were previously presented at the annual meeting of the Canadian Association of Applied Linguistics (2009, Ottawa, Canada) and at the Language Testing Research Colloquium (2010, Cambridge, UK). We are grateful to Maia Yarymowich, our research coordinator, for her support throughout the project, to the many institutions and individuals who helped us get access to nurses, and to our nurse participants.

Notes

¹As outlined earlier, letting the numbers drive the data by “throwing out” speech tasks was not pursued in this study. This was due to the nature of the low-stakes instrument being developed and to the importance placed on the domain experts' input in the development and validation of the list of speech tasks. However, in the interests of examining whether the data could support a better fitting model if statistically redundant speech tasks were to be removed, models that optimized fit by excluding speech tasks with large unique variance components were examined. A three-factor solution with nine speech tasks yielded the best fitting model, and all indices fell within the Schermelleh-Engel et al. (2003) “good fit” range (p = .22; RMSEA = .04; CFI = .99; NNFI = .99). This suggests that an assessment instrument for nurse licensure purposes could potentially be created using the same data in light of fit statistics; however, this was not the purpose of the present study.