Online Translators in Language Classes: Pedagogical and Practical Considerations

By Florencia Henshaw, Ph.D., Director of Advanced Spanish, University of Illinois, Urbana-Champaign

Nothing sparks more debate among language educators than the mere mention of Google Translate, other than perhaps the role of explicit grammar instruction. We might not be on the same page about whether online translators (OTs) are a useful tool or unauthorized assistance, but we all agree that they are here to stay and continuously improving. The efficiency appeal of a tool like Google Translate is hard for students to resist and for instructors to deny: using OTs is easier and faster than unassisted language production (and for some languages, just as accurate or more!). 

In classes with a face-to-face component, or “mask-to-mask” in the post-pandemic world, preventing the use of unauthorized resources is somewhat easier: no take-home writing assignments. However, in fully online courses, that is not an option. As the designer and supervisor of several online Spanish courses, ranging from novice to advanced, I have given this issue some thought. In this article, I share my insights and propose some pedagogical and practical considerations for dealing with OTs in the language classroom. 

Perceptions and policies

Students view OTs as one more tool, akin to dictionaries and spell check. Indeed, studies have found that the most popular use of OTs reported by undergraduate students is to look up words they want to understand or convey (Clifford, Merschel, & Reisinger, 2013; Faber & Turrero Garcia, 2020). Ironically, that’s when the accuracy of machine translation goes down, since the algorithm thrives on context. Students also rely on OTs to confirm or revise what they wrote, which seems odd from the instructor’s point of view, since online translators are not exactly known for their accuracy.

Understanding how instructors’ and students’ perceptions differ is crucial to draft course policies that make sense to all stakeholders. If students can look up a word in Linguee, why can’t they look up the same word in Google Translate? If spell check or similar “grammar checker” sites are allowed, why can’t they use an OT for editing purposes? It all boils down to the extent of the assistance that each tool affords. Unlike Linguee, Wordreference, or spell check, OTs could potentially tell students, for better or worse, exactly what they should say, without involving any meaningful decision-making, other than whether or not to trust its output. And in that sense, it would be comparable to asking someone else to complete the assignment for them. Sites like Linguee, on the other hand, offer several choices and leave it up to the students to make the final decision, thus reflecting their abilities more closely. As evident as this may seem to us, it may not be to our students.

Practices and pitfalls

A clear policy with respect to authorized and unauthorized resources is absolutely necessary, but we know it is not enough to dissuade students from resorting to OTs. Language educators often ask each other for ideas and advice to grapple with OT use, and responses are as varied as there are teachers. I will briefly explain and critically reflect on five of the most common approaches, and I will later propose a sixth alternative that tackles the inevitable omnipresence of OTs from both pedagogical and practical angles.

Approach #1: “Can’t trust this”

This is probably the very first tactic ever used against OTs: telling students that machine translation is highly inaccurate and results in ridiculous literal translations. This argument used to be quite convincing, but online translators have improved vastly in the last few years, at least for some languages. OTs are still struggling with songs and some discipline-specific jargon without enough context (e.g. see Google Translate’s attempt at the term “heritage speakers” below), and their output might never be flawless, but it can certainly handle the type of simple sentences on familiar topics that novice and intermediate learners tend to write about. I would venture that Google Translate is at least “Advanced-Low”(ACTFL, 2012) in Spanish, though its proficiency certainly varies by language. 

The most fascinating pitfall of this approach is that students seem to be well aware of the fact that OTs are inaccurate, and yet they continue to find them useful. According to Clifford et al. (2013), 91% of students surveyed said they detected errors, and yet 94% considered OTs to be always or sometimes helpful. Why? Because using Google Translate, as opposed to only an online dictionary, seems to help them write more accurate essays (O’Neill, 2016, 2019). It is clear that we need a different approach.

Approach #2: “One word only”

Some instructors allow the use of OTs to look up a single word or idiomatic phrase, presumably equating OTs to dictionaries. The issue with this approach is that the more isolated the input, the less accurate the OT is. For instance, “paper jam” is inaccurately translated by itself (resulting in something along the lines of “marmalade of paper”), but it is perfectly accurate when one adds the words “in printer.” Therefore, the one-word approach presents more cons than pros, given that it almost certainly leads students to inaccurate results. 

In a similar vein, the “ten items or less” approach entails setting a maximum number of words a student can look up, and asking them to disclose which words they obtained from outside sources. Unfortunately, the issues that we know result from word-for-word translation prevail, and some instructors anecdotally report a “slippery slope” effect, where the number of words students end up looking up is much greater than what they are disclosing. Additionally, could disclosing the source of the word make students feel less responsible for any inaccuracies? 

Approach #3: “Zero tolerance”

It is common for educators to dissuade OT use by warning students that doing so results in a zero on the assignment, on the basis of academic dishonesty. Although this sounds effective in principle, there are some issues that arise in practice because institutions tend to have a process in place for handling academic dishonesty cases, which helps to ensure that students are treated fairly. And when a case involves alleged use of an OT, things can get tricky. First, it is almost impossible to determine unequivocally the source of the unauthorized help. If we were accusing students of having copied someone else’s words without proper citation, it would not be enough to say “this sounds like something someone else wrote; I don’t know who, but I am giving you a zero because you must have copied it from somewhere.” 

Another issue with this approach is that research has shown that instructors are not always accurate at determining which writing sample was produced with the help of a translator. In other words, it could be possible to falsely accuse a student of cheating. In O’Neill’s (2013) study, the error rate of false positives (i.e., suspecting the assistance of a translator when that was not the case) was as high as 26.5%. Of course, it is quite possible that if participants in O’Neill’s (2013) study had been rating their own students’ samples, the accuracy rate might have been much higher. Nonetheless, the takeaway is that academic dishonesty allegations should not be taken lightly, and all educators should familiarize themselves with their own institutional policies and procedures before jumping on the “zero tolerance” bandwagon. 

Approach #4: “Where did you learn that?”

This approach entails strict limits: students are required to use only words or structures they have learned within the confines of the course. Odds are that the OT will produce language that has not been taught, which is seen not only as evidence of cheating, but also presumably harmful to the learning process because it might confuse students. On a practical level, the main challenge with this approach is that it is not clear how the instructor can keep track of all of the words and structures students have learned over time (i.e., several courses with possibly other instructors). 

More importantly, this approach raises a number of serious concerns from a pedagogical standpoint. First, don’t we want students to express what they want to communicate, and to expand their language abilities, as opposed to limiting them to recombine learned material or forcing them to say things in a particular way? Approaching a communicative act in such a restrictive manner sends a clear signal to students that the focus of the assignment is not what they say (i.e., meaningful content), but rather how they say it (i.e., grammar).

Second, we are doing our students a disservice by sheltering them from language that exists outside of the classroom walls. Our goal as language educators should not be limited to helping students do well on exams that test their knowledge of specific material. We should be motivating them to become lifelong learners. Don’t we want students to seek the target language outside of class, so they can continue developing their language skills? I personally encourage students to use some of these resources for Spanish on their own, participate in conversation groups, and embrace any chance they get to learn from other speakers of the target language. 

Last but not least, this approach seems to be based on the problematic assumption that the only acceptable dialects are those that are used by the instructor and/or the textbook. Shouldn’t we be fomenting an appreciation for linguistic diversity? And by the way, OTs are also unhelpful in embracing dialectal differences, since their output cannot be tailored by dialect, whereas spell check in Microsoft Word, for instance, acknowledges the existence of language varieties. 

Approach #5: “If you can’t beat them, join them”

Rather than banning OTs, proponents of this approach maintain that their output can be used to heighten students’ awareness of various aspects of language, mainly grammatical forms, through translation and editing exercises (Correa, 2014; Enkin & Mejías-Bikandi, 2016; Faber & Turrero Garcia, 2020). The first potential pitfall of this approach is that it tends to treat language as the object of study. Faber and Turrero Garcia (2020), for instance, see OTs as a tool “to test students’ understanding of the fine-grained aspectual distinctions studied in this part of the course” (Translator activity section), which implies a certain degree of emphasis on the importance of metalinguistic knowledge. The question that arises is: if our pedagogical goals don’t include learning about the language, what do OTs have to offer?

Another drawback of dissecting inaccuracies, especially when they involve the complex issues that OTs struggle with (e.g., long-distance agreement, mood selection in subordinate clauses, etc.), is that it might contribute to perpetuating the notion that the teacher expects perfection. In reality, according to ACTFL Proficiency Guidelines (2012), the writing of learners at the Intermediate-High level (i.e., after several years of study) may have “numerous and perhaps significant errors,” that may at times result in possible “gaps in comprehension.” If we are subjecting the output of OTs to ruthless scrutiny, when it is actually more accurate than what novice and intermediate learners can produce unassisted, we might be inadvertently sending the wrong message about our expectations. 

Other challenges with this approach, which Faber and Turrero Garcia (2020) acknowledge, are that (1) OTs may have already or might eventually overcome those grammatical struggles, resulting in a never-ending quest for faulty output, and (2) some learners are wrongfully concluding that you can’t trust the grammatical accuracy of OTs, but they are still helpful for single words, when in reality, isolated words are the Achilles heel of OTs. 

Also stemming from the premise that banning OTs is futile, the “enjoy responsibly” approach advocates for training students on how to use OTs so they can get the most out of them. Indeed, O’Neill (2019) found that writing tasks produced by students who received training on using Google Translate were rated higher than those submitted by all other groups. However, it is unclear to what extent we can consider those submissions to be a reflection of the students’ abilities in the target language, as opposed to their abilities in using Google Translate. Moreover, does this practice contribute to language development? According to the same study, the answer appears to be no: students who received training on using Google Translate performed worse than the other groups when they had to write a composition on their own, presumably because they had become overly reliant on Google’s help (O’Neill, 2019). More research is needed to further explore the link between OT use and language development. For now, the role of OTs as pedagogical tools remains unclear. 

A two-pronged proposal

The challenges and pitfalls with the myriad of approaches educators have come up with underscores the complexity of this issue. Even though upholding academic integrity is of utmost importance, trying to police and prosecute OT users in light of the challenges of the “zero tolerance” approach seems to be doing more harm to our rapport with our students than good in terms of curbing their reliance on OTs. Furthermore, while it is true that bilingual professionals use online tools out in the real world, and we should equip students with the knowledge necessary to make good choices on their own, how do we ensure that we are not creating the perfect loophole for academic dishonesty?

What I propose instead is to direct our efforts toward reducing students’ motivation to use OTs, by implementing both pedagogical and practical changes to our courses. My goal is essentially to create what Ducar and Schocket (2018) call a “Google-irrelevant” classroom, but without so much emphasis on translation activities and dissection of OTs’ strengths and weaknesses, given some of the concerns I delineated above. 

Pedagogical considerations

If our goal is to help students develop communicative ability in the target language at a certain proficiency level, every single component of the course, from the materials to the grading rubrics, should reflect that. Our expectations should be in line with the proficiency level of the students, and our instructional practices should be in line with what we know about second language acquisition. Here are some questions to ask ourselves:

1. Are students at the targeted proficiency level able to do what we’re asking them to do?  If you ask an Intermediate-Low learner to write a 200-word paragraph about something memorable that happened during a trip (with good control of past tense), you are bound to see an increase in OT use. And the students are bound to mistakenly conclude that they are just not good at learning languages.
2. Are students being evaluated on what they can do with the language at the targeted level? If our rubrics insist on a level of grammatical accuracy that is unrealistically high for their level, we are giving students reasons to want to use OTs because, despite some errors, their output is still more accurate than what many of our students can produce on their own, at least for some languages like Spanish.
3. Have we adequately prepared students to do the assignment in a relatively unassisted way? Doing computer-graded mechanical practice (e.g., matching vocab words with pictures, or filling in blanks with the correct verbs) does not prepare students to write a short paragraph about their family or favorite vacation spot.
4. Are we moving too quickly from input to output? Is the proportion of interpretive communication, relative to presentational and interpersonal communication, adequate for the level of the students? 
5. Are we being realistic with respect to what we are aiming for? If students entering the course are Intermediate-Mid, we cannot expect them to be Advanced-High after a few weeks.

Practical considerations

It is important to recognize that even if we have done everything right pedagogically, some students will resort to OTs to complete the simplest of assignments, merely because of convenience. If they can save 2.5 seconds by asking Google how to say “My name is…,” as opposed to trying to express that on their own, they will. Therefore, we also need to make course modifications on a more practical level.

First, as you plan your online course, classify your assignments into one of these three categories: 

• Green: students are likely to succumb to the temptation of OTs to complete them. Some examples would be written forum posts or open-ended responses; reading comprehension based on texts that can be translated automatically with the Google Translate browser extension.
• Yellow: students could seek help from OTs to complete them, but the assignment format makes it less convenient. For example, short video presentations that are rehearsed but unscripted (i.e., they can plan what to say, but they cannot read), or comprehension-based activities where the text is presented in the form of a screencast video (e.g., example of a “video reading”), which also helps students hear the words they are reading. 
• Red: OT use would be extremely impractical or impossible. Examples in this category include remotely proctored written assessments, synchronous sessions with the instructor or other speakers of the target language through platforms like LinguaMeeting or TalkAbroad (I’ve reviewed several of them here: “Video chatting with Native Speakers: Why, What, and How”), and unrehearsed asynchronous video responses on Extempore, where you can set up the recording to start automatically, as soon as students see or hear the prompt.

Then, the weight of course components is distributed in a way that the highest percentages (15%+) go to “red” assignments, followed by “yellow,” and the lowest being “green” assignments. And I mean truly low-stakes: for example, bi-weekly forum posts might only count for 5% of the course grade. Needless to say, all of the assignments, regardless of their weight, should be carefully crafted in light of the pedagogical considerations outlined above.

Absolutely paramount is to help learners understand that even though low-stakes assignments might not be worth much in terms of their final grade, they are still worth their time because they are meant to help them prepare for higher-stakes (“red”) assignments, and that connection should be explicitly made in the syllabus. If they choose to rely extensively on an OT for the low-stakes assignments, they will not gain much in terms of points, and they will actually lose a lot in terms of learning. They will miss out not only on developing communicative ability, but also on the cognitive benefits and personal growth that learning a language entails, from improving their ability to multi-task and process information to building humility, empathy and confidence, just to name a few. I suppose this approach could be called “it’s their loss.”

Caveats and calls to action

This approach is not without its limitations. For instance, the practical solutions outlined above do not address the unique situation of fully online composition courses, where high-stakes writing assignments (i.e., 1,500-word essays) cannot be eliminated or proctored. All language educators would probably welcome a tool that can effectively monitor at-home writing assignments, without the privacy concerns surrounding online proctoring services. 

Another caveat about the approach I propose is that it involves undoing the hardwired perception that only near-perfect work deserves full credit, and that takes time and trust. Do students believe us when we say we are not expecting perfection? And do we really mean it, or could there be some mixed messages when it comes to performance expectations?

We would also benefit from more research on potential benefits and drawbacks of using OTs and online dictionaries for language development, along the lines of O’Neill (2019), to better guide our policies and practices. Likewise, we need action research to confirm whether using OTs with advanced students, as suggested by Correa (2014) and Enkin and Mejías-Bikandi (2016), leads to an improvement in their self-editing skills: does raising students’ metalinguistic awareness through OT editing activities help them become better unassisted self-editors?

On a broader level, it would be helpful to have a more uniform view, across institutions, on whether using OTs constitutes a form of plagiarism or cheating. The lack of consensus among educators, reflected by the seemingly contradictory policies students encounter from course to course, have left OTs in academic integrity limbo. 

More research, conversations, and reconsiderations are warranted. Even if my proposal is not a good fit for your courses, I hope to have provided some food for thought. No, Google, not literally.

References

ACTFL. (2012). ACTFL Proficiency Guidelines 2012 [Electronic version]. Retrieved May 27, 2020, from https://www.actfl.org/publications/guidelines-and-manuals/actfl-proficiency-guidelines-2012

Clifford, J., Merschel, L., & Reisinger, D. (2013). Meeting the challenges of machine translation. The Language Educator, 8, 44–47.

Correa, M. (2014). Leaving the “peer” out of peer-editing: Online translators as a pedagogical tool in the Spanish as a second language classroom. Latin American Journal of Content & Language Integrated Learning, 7(1), 1-20. 

Ducar C. & Schocket D.H. (2018) Machine translation and the L2 classroom: Pedagogical solutions for making peace with Google Translate. Foreign Language Annals. 51: 779–795. 

Enkin, E. & Mejias-Bikandi, E. (2016). Using online translators in the Second Language Classroom: Ideas for Advanced-Level Spanish. LACLIL, 9(1). 138-158.

Faber, A. & Turrero Garcia, M. (2020). Online Translators as a Pedagogical Tool. FLTMAG. Retrieved from https://fltmag.com/online-translators-as-a-pedagogical-tool/ 

O’Neill, E. (2013). Online translator usage in foreign language writing. Dimension: 74-88.

O’Neill, E. (2016). Measuring the impact of online translation on fl writing scores. IALLT Journal of Language Learning Technologies, 46(2), 1-39.

O’Neill, E. (2019). Training students to use online translators and dictionaries: The impact on second language writing scores. International Journal of Research Studies in Language Learning. 8 (2): 47-65.