Google Translate was slow and difficult to use requiring repeated attempts to convey a message
Source
Hwang (2022). Testing the use of translation apps to overcome everyday healthcare communication in Australian aged-care hospital wards-An exploratory study. Nursing Open.
Description #
Under the "Communication" theme, although Google Translate (unlike the phrasebook apps) could translate patient responses, its free-text voice translation was at times slow and difficult, requiring repetition to convey a message; one observer needed seven attempts before a single question translated. The discussion attributed this partly to microphone issues and the often noisy ward environment.
"While Google Translate was able to translate responses, the utility of it was at times slow and difficult, requiring repetition to convey a message: "Needed to ask, 'Have you had a shower yesterday' 7 tries before translation." (Observer, site 1, Google Translate)." (Hwang, 2022, p. 582)
"For Google Translate, while its voice-to-voice translation feature is innovative, issues related to the microphone and the often noisy ward environment can render the conversational exchange slow and difficult." (Hwang, 2022, p. 583)
Methods Context #
What? #ⓘ
The observable: usability and speed of Google Translate's voice translation, characterized qualitatively as difficulty and number of attempts needed to convey a message.
"Accessibility and functionality relate to the capabilities of the device and translation apps." (Hwang, 2022, p. 582)
How? #ⓘ
Inductive qualitative content analysis of observer field notes and open-ended survey responses; the app was the free-text voice engine A-0016ArtifactA-0016Initial AI draftGoogle TranslateA free, general-purpose machine translation app increasingly used as an ad-hoc communication tool in healthcare settings to bridge language barriers with LEP patients, including via voice-to-voice translation. "One such…, used in the ward alongside two phrasebook apps.
"Google Translate was selected on the basis of feedback from ward staff that this and similar translation apps were already being used in the ward on an ad-hoc and unofficial basis." (Hwang, 2022, p. 580)
Who? #ⓘ
Nursing/allied health staff and older CALD patients with limited English proficiency during the six Google Translate observations (of 21 analyzed) and survey responses across four aged-care hospital wards.
"Of the 21 remaining observations, there were 10 observations for CALD Assist, 6 observations for Google Translate and 5 observations for TalkToMe." (Hwang, 2022, p. 581)
Other Notes #
Speed/usability failure is distinct from the accuracy failure (accents/dialects) captured in the parallel EVD; both bear on the reliability of free real-time machine translation in clinical settings.
Caveats #
- Observational data may be subject to selective reporting bias The study set no specific rules governing when observations of app use were collected, leaving observation to staff discretion in a busy, unpredictable ward. The authors caution that the observational data may therefore be subject to selective reporting bias, which constrains how representative the recorded interactions and the descriptive usage/engagement findings are of app use overall.