Article Preview
TopIntroduction
Technology has had and continues to have an impact on language learning and teaching. The latest interactive apps that students use in the language classroom, for example, Kahoot!, appear to represent a huge leap over earlier technological advances from another era. Yet these advances, in which students enjoy greater autonomy with their own devices, coupled with instant feedback, still require learners to acquire and use foreign languages in traditional ways: memorizing new words and understanding how to manipulate grammatical patterns in the target language. Thus, these latest technological advances are largely enhanced ways for students of foreign languages to do what they have always done in an effort to reach their goal of becoming proficient in a new language. In this sense, these pedagogical advances can be deemed incremental, rather than categorical. Further, they do not take advantage of new developments such as artificial intelligence or big data in any way.
The leap in categories alluded to here refers not so much to a pedagogical technique, but rather the driving force among many who decide to learn a foreign language, which happens to be the need to communicate with those who are speakers of the target language. This need, however, is increasingly being filled by instant and accurate machine translation (MT) at the tap of an app, obviating the motivation, at least the instrumental kind, for learning to write and read, and to some extent, speak and listen in foreign languages (Crossley, 2018, p. 547).
Among all machine translators, Google’s service, Google Translate (GT) may be the most well-known and used because of its link to its popular search engine and its recent upgrade in 2016, which uses a neural network approach to analyze sequences of words, called Google’s Neural Machine Translation (GNMT) (Wu, Schuster, Chen, Le, & Norouzi, 2016). This recent advance has resulted in translations that are significantly superior to those produced prior to the upgrade. Wu et al. (p. 20) claim that the “GNMT system approaches the accuracy achieved by average bilingual human translators on some [of their] test sets.”
As the enhancement of MT has progressed over the years, there has been a considerable amount of research conducted on students’ use of MT in foreign language learning contexts, much of which has been descriptive and attitudinal in nature. For example, Jolley and Maimone (2015), White and Heidrich (2013) and Clifford, Merschal and Munne (2013) reported that among their Spanish, German and Romance language learners respectively, most of their students used machine translators at least some of the time. As for students’ attitudes towards online translators, studies (Bahri & Mahadi, 2016; Clifford 2013; Cornell, Dean & Tomas 2016; Farzi 2016; O’Neill 2019) have had mixed findings across native speakers of several languages: students do use MT, largely GT; however, there remain concerns about its accuracy. Almost all of this research, which has largely investigated students’ and teachers’ impressions of the usefulness and accuracy of GT has produced mixed results at best, although notably, most of these studies were conducted before GT’s enhanced GNMT had replaced older systems.