A Reverse Turing Test: Is Good Language Teaching Robot-Proof?
The always brilliant Gene Glass provides a list of jobs that are “0% Automatable” (cannot be done by a robot) and “100% Automatable.” He uses the list to examine what schools teach and the future of our economy, but it is a question you may want to ask of your own job.
Robots and computer software can very efficiently deliver traditional “drill and kill” in language arts, as Frank Smith pointed out. A Korean researcher has just discovered that robots work wonderfully on mindless (and utterly ineffective) tasks like phonemic awareness training, and someone in Japan is looking into “pedagogical machines.” But does it takes a human mind to provide compelling language input? I think the answer is “yes.”
Perhaps we need a sort of “Reverse Turing Test” for language teaching: If an observer can’t distinguish what you’re doing in the classroom from what a robot can do, you’re probably doing the wrong thing.
ETS Introduces Software to Suck the Life Out of All Student Reading Material
In a paper just released this week, the Educational Testing Service reports on preliminary results of its online service, Language Muse. Study authors explain that it can take a teacher’s “input text,” analyze it for vocabulary and syntax, then spit out “activities” based on the student’s level:
Teachers upload a classroom text into LM [Language Muse]. The engine automatically generates over 20 activities based on linguistic features identified in the text. Teachers then select activities to create an “activity palette” . . . — a set of text-specific activities — to support one or more learning objectives, such as “practice with derivational word forms”.
No real data were reported in their “in progress” report, though the authors did say the results were “difficult to interpret.” Sounds like bad news to me. They added that “gains were observed in some components, but there were also score losses – potentially due to motivation, since the assessments were no-stakes” (p. 214).
“If only we could make all assessments high-stakes!” an ETS spokesman didn’t say, but was likely thinking.
Er íslenskt að vana (Is Icelandic Disappearing)?
If these AP and New York Times stories are correct, children and teenagers in Iceland are shifting so quickly to English, they aren’t even fluent enough to read the one Nobel Prize-winning author the island has produced (and yes, I’m talking about Halldor Laxness). A former president laments the language may soon be in the “Latin bin,” a comparison that will make her no friends in Finland. Perhaps they should apply for a grant from the new Endangered Languages Fund?
Gap Widens Between Low-Achieving and High-Achieving First Graders; Researchers Stumped
Parent Magazine reports on a new study just published in Educational Researcher (paywall) that claims kids entering first grade are doing better overall at reading than they did 12 years ago. The gap between low-achieving and high-achieving readers has also narrowed, but only on isolated skills such as letter recognition. When it comes to actually reading, the gap between the two groups is greater now than before.
Could it be that all that early literacy skills training of the past decade wasn’t the way to go?
One of the study’s author, Jerome D’Agostino, said, “There’s a missing link between teaching low-achieving students basic literacy skills and having them actually put those skills to use in reading,” he said. “We don’t know what that is yet.” His co-author, Elaine Rodger, however, has a radical solution sure to be rejected by saner voices in the field: “We’re probably spending too much time emphasizing basic skills for the low-achieving students, when we should be giving them more opportunities to actually read text.” (Hat tip: Warren Ediger)
Can You Learn Spanish from Listening to Dodger Baseball?
Los Angeles Times columnist Steve Lopez muses that you can improve your Spanish by listening to the Spanish-language broadcasts by the veteran LA Dodger announcer, Jaime Jarrín. I haven’t tried that, but it certainly could be a good source of “narrow listening” with (for a Dodger fan) compelling input, especially of Clayton Kershaw is pitching.
More interestingly, the article describes the loss of the home or heritage language by 3rd generation immigrants, including by Jarrin’s own grandson. Lopez thinks this loss is a “surprise,” but it is in fact the rule rather than the exception. (Hat tip: Warren Ediger)
Learning English Can Boost Your Salary
A new survey of more than 5,000 businesses around the world by Cambridge English finds that nearly all companies think English is important for their success. Most reward employees with good English skills. No surprises here.
Three Anthropology Graduate Students Try to Learn Turkish with Duolingo and Hate It
Or as they put it, “Narrative Perspectives on Self-Directed Foreign Language Learning in a Computer- and Mobile-Assisted Language Learning Context.” They didn’t like the “stilted robotic voice,” “the same routinized patterns,” and the frequent “absurd sentences.” After struggling to finish the first level, they dropped out (just like most people). Languages (open access)
German Learning Helps Students With Dyslexia?
An ASU professor claims teaching German “also lets her help students overcome dyslexia.” After mentioning the frequently cited and completely bogus statistic that 20% of American students have “language-based dyslexia,” the university press release quotes Dr. Sarah Lee, who says she sees herself as “a dyslexia specialist when it comes to teaching foreign language, especially German”:
“The difference is really in the language itself.” Lee explained. “In English, there’s much more of a variety of how different sounds can be represented by different letters or how one letter can have different sounds. … In German, it’s much more of one letter represents one sound.” . . . “It’s not quite as confusing. There are not so many different ways of pronouncing and spelling words,” Lee said. “So it’s actually easier to speak and learn German than it is to speak and learn English.”
Not to rain on the Frau Lee’s new-found specialty, but, despite what you read in the papers, the case has been made that fluent reading is not primarily a matter of phonology and transparent orthographies (see here, here, here, and here, just for starters). On the specific question of L1 reading “skills” and L2 reading, a 2016 study by Alderson and colleagues in Modern Language Journal (paywall) of Finnish ESL students found that the factors typically associated with “reading disabilities” and dyslexia in the L1 are not the most important in differentiating good and poor L2 readers (see also a 2013 meta-analysis by Melby-Lervåg and Lervåg). All the phonological theorizing is rather unnecessary.
Military Drops Giant Grant Bomb in Latest Attempt to Accelerate FL Acquisition
According to a press release:
Researchers at the Texas Biomedical Device Center (TxBDC) at The University of Texas at Dallas have been awarded a contract worth up to $5.8 million from the Defense Advanced Research Projects Agency (DARPA) to investigate a novel approach to accelerate the learning of foreign languages.
The newly-awash-in-grant-money TxBDC will spend it on some really nice hotel rooms at the next conference “precise activation of peripheral nerves.” More specifically, the researchers will “focus on developing an approach that uses vagus nerve stimulation (VNS) during training to specifically reinforce neural networks that are involved in learning a particular task,” which is apparently more difficult than it sounds. Specifically:
It involves sending a mild electric pulse through the vagus nerve in the neck. When stimulated, the vagus nerve affects the brain, where it boosts the release of chemicals called neuromodulators. These chemicals facilitate synaptic plasticity, a process in which the connections between brain cells change and strengthen during learning. . . “We believe that we will be able to substantially increase the rate of language learning. With VNS, we may be able to improve on the brain’s natural ability to learn,” Kilgard said.
At least part of the aim of VNS is to improve working memory, which has heretofore not been very amenable to training (Developmental Psychology, paywall). But it seems from other evidence that even if you can improve working memory, that itself does not improve other types of “cognitive performance,” such as reading comprehension; the prolific Dr. Melby-Lervåg and her colleagues have the details here.
Cambridge University Press Has a Few Psycholinguistics Articles for Free Until May 31st
I could find only eight articles, but may be worth a quick look.
In Case You Missed It:
- Memo to The American Conservative: Bilingual Education Will Not Actually Make Californians Bilingual
- Rakuten, Japan’s 21st Century Online Powerhouse, to Launch 19th Century Language School
Quick Takes:
- L1 Reading Habits Correlate With L2 Proficiency for Intermediate Students – There is a lot of noise in the data, however.
It appears that the “beginning” and “intermediate” groups actually scored about the same on the L2 assessment, which if true makes the conclusions rather doubtful. Data reporting is incomplete. System (paywall) - Perfectionism Correlates With Language Test Scores – But the language measure (a modified Michigan Test) was at least 1/3 grammar, and another study using university course grades found the opposite.
- Your Brain on L1 versus L2 Reading – “[I]mportant brain regions for L1 are carried over to L2 reading, maybe more so in highly proficient bilinguals.” Brain and Language (paywall)
- Carrot and Schtick: Reading Helps Vegetable Consumption – From my new favorite journal, Appetite (paywall): Researchers found that kids who were engaged in “interactive reading” of picture books ate more vegetables than those who were read to with hand puppets (no, I don’t know how that works, either). Study also give us our Sentence of the Week: “The main finding was that interactive reading produced the greatest carrot consumption.”
- Bilingual Advantage Losing Its Publishing Advantage – Three researchers found “bibliometric” evidence of an uptick in studies questioning the cognitive advantages of bilingualism from 2014-2015. If we did science by vote counting and “consensus,” this would be really important to know. PLoS (open access).
Did I miss something? Send submissions for next week to jeff [at] backseatlinguist.com
Follow Us!