Neil deGrasse Tyson on linguists and Arrival

« previous post | next post »

This is a guest post submitted by Nathan Sanders and colleagues. It's the text of an open letter to Neil deGrasse Tyson, who made a comment about linguists on Twitter not long ago.

Dr. Neil deGrasse Tyson,

As fellow scientists, we linguists appreciate the work you do as a spokesperson for science. However, your recent tweet about the film Arrival perpetuates a common misunderstanding about what linguistics is and what linguists do:

In the @ArrivalMovie I'd chose a Cryptographer & Astrobiologist to talk to the aliens, not a Linguist & Theoretical Physicist

Neil deGrasse Tyson (@neiltyson), 1:40 PM – 26 Feb 2017

Though the term linguist is often used by the public to refer generally to anyone whose occupation is related to language (especially translators and interpreters), the type depicted in Arrival is a special kind of linguist who engages in the scientific study of human language: its structures, its uses, its underlying similarities, and its surprising diversity. A cryptographer simply cannot replicate the specialized training that a linguist like Louise Banks has, which takes years to learn and decades to master.

Most importantly, a cryptanalyst would likely be much less suited to the task of communicating with aliens than a linguist would (a cryptographer even less so, since they work on encryption, not decryption). Cryptanalysis relies on decrypting coded messages from a known language. If the source language and the encryption method are both unknown, ordinary cryptanalytic methods will fail. This is why the Native American code talkers of the 20th century were so invaluable to the US in both world wars: their languages were not understood by enemy cryptanalysts, so their encrypted versions could not be cracked, unlike with well-known languages like English.

A linguist's interactive methodology is more likely to result in successful communication with aliens. Whereas cryptanalysts generally work with a static corpus of encrypted messages and cannot obtain new ones of a particular type on demand, linguists are trained in a variety of techniques to elicit targeted utterances from speakers, as broadly demonstrated by the elicitation sessions in Arrival. These elicitation sessions are designed to bring to light subtle information about the atomic units of a language, how they are combined into longer units, what those units mean, and how they are used. These methods are used for analyzing the structure of well-known languages as well as for documenting and analyzing endangered languages that the linguist may not speak with any fluency and may be typologically quite different from widely spoken languages of the world.

Perhaps instead of true cryptanalysis, you were thinking more along the lines of machine translation, with the idea that languages can be treated as codes of each other or of some universal interlingua. This idea was popular among 17th-century philosophers and has been explored by computational linguists since the mid-20th century, but with limited success, because human languages simply are not structured as codes of one another. They can differ not only in how they express information (sometimes by word order, sometimes by extra words or pieces of words), but also in what information they express (such as grammatical gender, levels of politeness, and verb tense), and these differences are often difficult to adequately translate from one language to another. Any methodology relying on the assumption that all languages are merely variant realizations of the same underlying concepts is doomed to fail. An alien language, with its associated alien thoughts and alien culture, would be even less amenable to such methods.

Linguists have specialized fieldwork techniques and an understanding of what kinds of information a language is likely to express and how that information may be realized, which are necessary tools for understanding any new language, human or alien. Linguists are thus exactly the kind of researchers you would want on hand in an alien encounter.

We thank you for your time and attention, and we hope that your commitment to educating the public about the sciences will include and celebrate the important contributions of linguistics.


Kevin Schluter
Postdoctoral Researcher, New York University Abu Dhabi

Nathan Sanders
Visiting Assistant Professor, Haverford College

Stephen Politzer-Ahles
Assistant Professor, The Hong Kong Polytechnic University

Carrie Gillon
Research Associate, University of Manitoba

The above is a guest post submitted by Nathan Sanders and colleagues.


  1. Not Neil deGrasse Tyson said,

    March 3, 2017 @ 12:09 pm

    Tyson has demonstrated on multiple occasions that he suffers from a severe case of Engineer's Disease.

  2. Yuval said,

    March 3, 2017 @ 12:11 pm

    It's as if NdGT didn't even watch the movie before commenting on it, no?

  3. Dan Mac said,

    March 3, 2017 @ 1:13 pm

    Tyson is great at rhetoric. As Sagan's successor, this makes sense. He's also very funny and should consider professional comedy, outside of science.

    But his philosophy is really terrible and ignorant.

  4. John said,

    March 3, 2017 @ 1:18 pm


    Perhaps he did not read the short story. I would have found the movie a bit confusing had I not read the short story.

    I can see where he is coming from because the movie gave me the idea that the heptapod language was just a type of code for English concepts (since we don't get to see the perspective of any other languages).

    IMO, the short story has a completely different message from the movie, which is surprising since Ted Chiang was involved in the movie.

  5. Jenny Chu said,

    March 3, 2017 @ 1:18 pm

    Can you get some more signatories?

  6. Christian DiCanio said,

    March 3, 2017 @ 1:37 pm

    I'll sign if they want more signatures.

  7. Brook D Lillehaugen said,

    March 3, 2017 @ 2:07 pm

    I would sign, too.

  8. Nathan Sanders said,

    March 3, 2017 @ 2:07 pm

    We intended the letter to be hosted here on Language Log, so adding signatories after posting wasn't on our original agenda. (Not that we are opposed to the idea; it just didn't come up in our discussions.)

    If we decide to convert it to a more dynamic form elsewhere that would easily allow for additional signatures, we'll spread the word.

    In the meantime, please consider yourself an honorary signatory if you share this post! The more widely it's disseminated, the more likely he is to read it.

  9. N.D. said,

    March 3, 2017 @ 2:14 pm

    While I agree that we need to promote science awareness now more than ever and I appreciate NdGT's efforts in doing so, this Clickhole article is spot-on:–3565

    I'm glad this letter was written because the most I could do in response to his tweet was a hard eye-roll.

  10. ngage92 said,

    March 3, 2017 @ 2:15 pm

    I can't think of anyone who's gone from "Wow this guy is so cool" to "Wow this guy is a tedious, pedantic bore" faster than DeGrasse Tyson.

  11. Guy said,

    March 3, 2017 @ 2:30 pm

    Although the short story showed its linguistic work, I was actually pretty underwhelmed by the movie's handling of linguistics, so I can see how someone with little familiarity with linguistics might fail to understand why a linguist would be well-suited to the task. As for the physicist, the short story indicated there was a large team of researchers at each site, and the physicists were there primarily to try to engage in an exchange of technology with the aliens once communication become possible, not really to learn their language. This might also not have been clear in the movie.

  12. Arturo Magidin said,

    March 3, 2017 @ 3:29 pm

    While I agree that linguists are definitely better choices than cryptoanalysts, I would take a bit of issue with some of your comments regarding cryptoanalysis. Note that the code talkers, for example, were engaged in tactical information exchange: these are "trench codes", messages were you need very fast decryptions/encryption, have a very small shelf-life (so you don't care if the enemy can decode them in three hours, because by then the information is no longer important), and where authentification has a very high premium (to ensure that the information being provided comes from a trusted source). These are messages used to pass information and orders in the middle of a battle or an operation, while the operation is on-going.The code talkers were particularly good for these because decryption, although not technicaly difficult (at the level of the PURPLE or ENIGMA codes) was sufficiently time consuming for non-natives; the ease of distinguishing non-native speakers gave strong authentification protocols; and the use of bilingual operators made encryption and decryption very fast. These are tactical codes. By contrast, codes such as ENIGMA and other codes that we are more familiar with and that were broken were generally signal codes: the messages had a much longer shelf-life (orders to deploy a carrier group are good to know for several days after the message is sent), authentification was not really an issue, and the velocity of encryption and decryption was also not an issue for authorized users (ENIGMA was relatively slow, but they didn't care because the information did not need to be acted on immediately). Finally, note that familiarity with the underlying language is not generally strictly required for cryptoanalysis: for example, the cryptoanalysts in charge of cracking the Japanese PURPLE code were not speakers of Japanese (either as first or second language). What is true is that understanding the plaintext will require familiarity with the underlying language once the cryptoanalyst has completed their work, and that is where Dr Tyson's remarks really fail.

  13. Marianna Pool said,

    March 3, 2017 @ 4:04 pm

    Give the guy a break. How many ists dos you know who don't have a clue about what we do? But most intelligent people get it once it's explained.

  14. Marianna Pool said,

    March 3, 2017 @ 4:07 pm

    Sorry, my autocorrect is wacky. Pretend I said that in English.

  15. Evan Harper said,

    March 3, 2017 @ 4:18 pm

    Sam Kriss is not one of my favorite people but his Neil deGrasse Tyson: pedantry in space is on point

  16. Jason Baldridge said,

    March 3, 2017 @ 4:58 pm

    It's fair to respond to say that Tyson is being unfair to linguistics here, but there are ways to improve this response:

    – It focuses on MT as being based on interlinguas, and that's not how modern MT works. The best systems use gobs of parallel texts and they can learn very interesting phrase reordering patterns and much more.
    – I'd say that MT is incredibly successful as an endeavor at this point. No, it can't create beautiful translations of great literature, and it screws up discourse effects tragically. But it conveys a great deal of the meaning of the source language — provided it has sufficient training material. I'd take that argument out.
    – The question of training material is the key failing point for modern MT in the alien setting: it is generally hungry for many example pairs of sentences in two languages of interest. Obviously, we wouldn't have parallel text for English/Alien to train these models on, so this seems like a dead end. However, there is work in MT that doesn't use parallel text and in fact is based on ideas from decryption! E.g.
    * Ravi and Knight (2011):
    – MT approaches are nonetheless data hungry and need lots of examples. So, I'd instead emphasize the importance of elicitation that a linguist can provide. The potential for the linguist is to get some grip on the language of interest with orders of magnitude fewer examples.
    – You could also look to combining human and machine. E.g. there are methods for building MT systems that have no parallel text, but which make use of bilingual lexicons (which our linguist could get started). E.g.:
    * Klementiev et al (2012):
    – Another reason you want the linguist along is they can help figure out which parts of the utterances are meaningful. Is the language based on sound, visual elements, both, something else?

    What would possibly work the best is to have a linguist guiding the elicitation with some great tools—based on ideas from cryptoanalysis perhaps!—to run in the background to help generate hypothesis and narrow down the search space. I'd focus on the need for data, and the ability to figure out how to get any data at all, as why the linguist is needed here.

  17. ex0du5 said,

    March 3, 2017 @ 5:08 pm

    As one who has a fascination for both linguistics and cryptanalysis, I do think there is use for both. Particularly through the connection with complexity theory. It does seem like exist a (single or small collection of) minimally complex machine operations in the real world on which to build a complexity theory of messages. There seems to be a natural language of operations on which to measure Kolmogorov complexity – at least providing a non-uniform spectrum of measures.

    So we can define mappings of a given message to other messages of lower complexity and compare that to the complexity of the machine that makes that mapping. This comparison can help reveal "likely" decode paths and provides a way to identify proper translations.

    Additional tools that cryptographers use – lexical placement, syntactic distributions, etc. – all eventually revert back to these complexity-theoretic measures when philosophically challenged. They are all fundamentally ways of extracting likely low-complexity mapping functions (interpretations / translations / semantic implementations / etc.).

    Because the meaning of what's going to be said can be measured in ways outside the direct establishment of semantic affirmation. There is the possibility of cryptanalysis to give some assistance here.

    But it's also pretty weak compared to getting that affirmation directly. So having a linguist, one versed in the philosophy of semantics and theories of meaning and truth, was certainly best in the particular scenarios of the movie.

    The physicist could have helped the cryptographer build physical complexity languages, and the astrobiologist could have helped unveil the history they carry with them in their bodies, so honestly, having all come for the ride would probably have been the better choice. And really, it could be the end of the world, but they can only afford to bring two? This sounds like a strange bargain.

  18. Matt McGarrity said,

    March 3, 2017 @ 5:41 pm

    I'm torn. I agree with the sentiment if the letter, but I like Tyson's general approach to public intellectualism. I'm a rhetorician, so I feel the sting of misuse. Even in this comment thread there has been the "oh, that's just rhetoric" comment. I try not to take popular uses of the term as professional slights.

  19. jick said,

    March 3, 2017 @ 6:12 pm

    I feel that this letter does the same kind disservice to cryptography/cryptanalysis/NLP that Tyson's tweet did to linguistics. In particular,

    > Any methodology relying on the assumption that all languages are merely variant realizations of the same underlying concepts is doomed to fail.

    This sounds like a straw-man argument. An NLP researcher does not assume that different languages are "variant realizations of the same concepts" any more than a physicist assumes that cows are spherical balls floating in frictionless vacuum. It's just a useful approximation which can get a surprising amount of job done given the right circumstances.

  20. Guy said,

    March 3, 2017 @ 8:07 pm


    I don't understand the sentence you cite to claim that cryptanalysts believe that all languages are coded forms of each other, I understand it to be saying that believing that a cryptanalyst is better suited to learning and translating a completely unknown language than a linguist expresses such a belief. Are there cryptoanalytic methods that are able to translate or comprehend a message in an unknown language without access to any translated corpora or translation dictionaries or syntactic knowledge for that language?

  21. Nathan Sanders said,

    March 3, 2017 @ 8:44 pm

    Arturo writes "What is true is that understanding the plaintext will require familiarity with the underlying language once the cryptoanalyst has completed their work, and that is where Dr Tyson's remarks really fail."

    Yes, absolutely. This is what we intended by "If the source language and the encryption method are both unknown…" Perhaps this point would have been more clear had we said "…the meaning of the source language…" And thanks for pointing out the differences in how the different codes were used! Very fascinating stuff.

    Jason, your points are well-taken, though as for not needing parallel texts, in the paper you cite, Ravi and Knight state "we imagine that this corpus of foreign strings 'is really written in English, but has been coded in some strange symbols'". They get good results for English-Spanish, which are rather similar languages from a morphosyntactic standpoint (largely fusional, SVO, no significant case system, verbal inflectional largely restricted to tense and subject agreement).

    But I would want to see how their methods work with languages with much different morphosyntax, because I've done MT work for English-Tuvan, and even with training from parallel texts (including massive sets of paradigmatic sentences to flesh out the case system), the results are still pretty poor. Tuvan's agglutinative morphology, SOV word order, extensive case system, and rich tense-aspect-mood verbal inflection proved to be quite difficult for MT to handle.

    In addition to this problem of translating between languages with wildly different morphosyntactic types, I still believe not being able to safely assume human-like semantics would be an insurmountable barrier, without using interactive elicitation methods. Note crucially that Ravi and Knight's methods rely on semantically related Spanish and English data (temporal expressions, movie subtitles). We couldn't do that with aliens, because we have no a priori notion of what they even talk about, and we certainly can't guarantee that our alien corpus would have suitable semantic similarity to any human corpus. And if even we could somehow guarantee it, we'd have no idea which human corpus to use.

    That said, I do think computational methods would be valuable, if not outright necessary; they would just make a poor substitute for the methods used by field linguists.

  22. jick said,

    March 3, 2017 @ 9:33 pm


    I interpreted the sentence as "Any such method is doomed to fail, in this world, to produce a meaningful method of translation." Which sounds like a straw-man argument to me.

    However, it might be that you are correct and the sentence was referring to the choice of professions inside the movie, in which case, well I have no opinion (I haven't even seen the movie!) — it's going to be about as interesting as arguing why Batman didn't bring a machine gun to kill Joker. (In other words, not very interesting for me.)

    …Although I have to wonder what makes linguists so confident that what they learned about *human* languages can be applied to aliens. After all, when the subject of animal communications are brought up here (and note that these animals are infinitely close to us than aliens, genetically speaking), the general mood seemed to be polite indifference.

  23. Ray said,

    March 3, 2017 @ 9:41 pm

    I haven’t seen arrival, but it kinda reminds me of how, while watching a documentary on cave exploration, for example, you’ll hear the narrator say (in a very deep and important voice): “this is the first time a human being has ever set foot in these waters” — and then you’ll realize that a whole camera crew and production team had already traversed those waters in order to film the lone, intrepid explorer.

  24. Nathan Sanders said,

    March 3, 2017 @ 10:07 pm

    > Although I have to wonder what makes linguists so confident that what they learned about *human* languages can be applied to aliens.

    We aren't.

  25. Joyce Melton said,

    March 3, 2017 @ 11:18 pm

    As an amateur linguist, holder of a degree in math, and a one-time professional cryptanalyst, I can say, with some parallax on the problem, that what is needed for the job in the movie is a team of different sorts of linguists and at least one mathematician with a specialty in information theory.

    The cryptanalyst could draw cartoons in the top secret logbooks which was also part of my job. Well, it was something I did even if it wasn't part of my job.

    [(myl) Good advice! And in addition to Joyce, someone like Kevin Knight would be a good choice for the team — see these slides from his 2013 ACL tutorial, or his work on the Copiale Cipher.]

  26. Nathan Sanders said,

    March 3, 2017 @ 11:35 pm

    Joyce, I like your plan!

  27. Matt said,

    March 4, 2017 @ 12:05 am

    It would be nice to get him to pay attention to this matter. He's an influential source of information, and if we can get him to send the right message, that'd be great for outreach. Is there some way we can pester him enough in a politic manner so as to achieve such an effect? An open letter is certainly a good step.

    It'd be great if we could somehow get a generative linguist on StarTalk somehow.

  28. Lugubert said,

    March 4, 2017 @ 12:07 am

    For understanding alien languages, do include an architect on the team.

    I am of course referring to the role of Michael Ventris in deciphering Linear B. He started with no idea of which language he was addressing, and had no population on which to use field linguistics.

    Assuming unlimited funds available, what kind of team would you people suggest for deciphering the Indus Valley script?

  29. GH said,

    March 4, 2017 @ 4:39 am

    @Jason Baldridge

    Also, the film eventually shows that the researchers have developed an application that allows them to generate written Heptapod sentences from fragments indexed in English (making the appropriate adaptations/conjugations as they are put together – though it's unclear just how advanced the underlying language processing is), so there is apparently some element of MT/computer linguistics in the program.

  30. Edwin Schmitt said,

    March 4, 2017 @ 8:40 am

    Ha, apparently he just can't help himself. See the comments he made during a Joe Rogan interview.
    Around minute 35 he really wants to talk about it but holds back and instead explains his critique of the movie Gravity. Hopefully if enough of us @ him on Twitter he'll realize the importance of this reply.

  31. Ken said,

    March 4, 2017 @ 9:43 am

    I think I'd want a range of scientists on the team, a la H. Beam Piper's Omnilingual. However the aliens in that story were long-dead, so point-and-say wasn't possible.

  32. J.W. Brewer said,

    March 4, 2017 @ 10:42 am

    My native-speaker intuition finds the "As fellow scientists" opening rather ridiculous-sounding, not just because of Tyson's dubious status, although that's part of it, but because linguists self-identifying as "scientists" sounds ridiculous to me. However, this is certainly not because I find the contention that linguistics is a "science" ridiculous or even false. Rather, it's because "scientist," at least in my variety of English, has narrower semantic scope than "anyone who works professionally in a field that falls within a broad-but-reasonable sense of 'science.'" (That the two related words may have different scope should not be surprising to anyone familiar with the scientific field of linguistics.)
    Similarly, with the exception of a few things off to one side of the discipline (e.g. what they do at Haskins Labs), it would sound weird at least to my native-speaker ear to use "laboratory" to mean "building where linguistics people do the scientific stuff that they do."

    Perhaps my idiolect is idiosyncratic, but if it's not, I would respectfully suggest that self-identification as "scientists" is likely to be counterproductive for the generally worthwhile project of increasing respect for linguistics as a rigorous (and, indeed, "scientific") scholarly discipline. Think of it as akin to the phenomenon of persons with an Ed.D. who self-consciously insist on being referred to as "Dr. SURNAME." Rather than leading to increased respect it often leads to decreased respect, because it tends to focus attention on the differences and distance between the Ed.D. and the paradigm referent of someone addressed as "Dr. SURNAME" rather than on the similarities.

  33. Nathan Sanders said,

    March 4, 2017 @ 11:43 am

    I'll readily admit that our opening line was purposefully chosen to immediately challenge such restricted notions of what science and scientist mean, because it's all too common for linguistics and linguists not to be counted. If the line triggered you, then good, it was supposed to!

    For what it's worth, I know plenty of linguists who have a lab(oratory), in both function and name. My colleague David Harrison at Swarthmore College runs the "Laboratory for Endangered Languages Research and Documentation", my office there is physically located within the "Phonetics Lab", and the research I'm doing with Dan Grodner on eye-tracking is being conducted in the "Psycholinguistics Laboratory".

  34. Zeppelin said,

    March 4, 2017 @ 12:41 pm

    J.W. Brewer: This seems to be the English-specific division between "sciences" and "the humanities" again.
    The German Wikipedia article on "studia humanitatis" defines the English term "humanities" (which has no equivalent in German) as "a categorisation of sciences widespread in the English-speaking world which subsumes all sciences relating to humans (as individuals or collectives)", for example.

  35. Adso said,

    March 4, 2017 @ 12:56 pm

    I'd like to add that there is also historical precedent for the idea that cryptanalytic techniques might not be applied productively to an unknown writing system, because that was exactly the project of cryptographer William F. Friedman and his wife. Friedman was a skilled code-breaker and—among many other accomplishments—notably broke the Japanese PURPLE code during World War 2. One of his other pastimes was attempting to decipher unknown languages, most famously the Voynich manuscript, but at one point he and his wife were given a grant to apply cryptanalytic techniques to the then-largely-undeciphered Mayan writing system. Despite Friedman's stellar background in cryptography, he and his wife failed to decipher even a single glyph.

    The decipherment of Mayan wasn't exact the same as the decipherment of Heptapod as it appeared in the movie, of course, but I'd still contend that this provides historical evidence of the assertion that the cryptographer's skill-set is comparatively ill-suited to deciphering new languages. (And to the degree that the movie's translation effort differed from the decipherment effort of something like Mayan, it does so in ways that would make a linguist even more indispensable: for example, the Mayan writing system had no remaining users who could produce new texts, but the Heptapods could—making a linguist with training in elicitation even more valuable!)

  36. Guy said,

    March 4, 2017 @ 1:55 pm


    This potentially varies by person, but for me "the humainities" refers to things like what would be covered by the typical art and literatuture classes at a university, and does not include the social sciences. Social sciences often get lumped together with the humanities because of similar subject matter. Oddly, physical sciences don't get lumped in with engineering on the basis of subject matter, but rather engineering tends to get lumped in under the "science" heading.

  37. Zeppelin said,

    March 4, 2017 @ 3:50 pm

    Guy: For what it's worth, the academic study of literature is called "literary science" in German.

  38. The Other Mark P said,

    March 4, 2017 @ 5:25 pm

    Give the guy a break. How many ists dos you know who don't have a clue about what we do?

    If he did not know what you do, then he should have kept his mouth shut. That's what sensible people do when they are out of their field of expertise.

  39. J.W. Brewer said,

    March 4, 2017 @ 5:50 pm

    Well, if we were speaking German we could just call someone who did linguistics stuff for a living a Sprachwissenschaftler (or -lerin, as appropriate), which is I dare say a much more awesome-sounding name than "scientist," as well as being more specific.

    I appreciate Prof. Sanders' information on the present usage of "laboratory." For the psycholinguistics eye-tracking stuff it sounds wholly appropriate to me. As for the endangered languages lab, the name primes me to expect a more extensive and cooler inventory of lab equipment and scientific apparatus than merely microphones, tape recorders (or I guess digital equivalent) and computers with software suitable for producing fancy analyses of recorded speech, but if that's not the case maybe my own intuitions are out of synch with current usage, at least in Pennsylvania-Quaker-Academic dialect, and if so I am happy to become better informed.

    The problem (and I encourage Prof. Sanders and his colleagues to gather data from a representative cross-section of AmEng speakers, because it may well be that my reaction is not representative) is that the use of "scientist" did not "trigger" me in a way that was actually productive. I.e., it did not move me, after having been initially jarring, toward greater reflection on the question of "what is a scientist, anyway, and how does that relate to what linguists do?" And FWIW I am temperamentally as well as philosophically enough of a descriptivist that unexpected uses often elicit a "hmm, that's interesting" response from me rather than a "no that's wrong wrong wrong" response. But not here. But more to the point, my response here was not actually "no that's wrong wrong wrong," but rather "the use of that word makes the signatories come off as insecure about their social status within the academy and fishing for emotional validation." That's imho an affirmatively counterproductive reaction to be eliciting, given what I take the object of the letter to be. It's even worse than, e.g. "the use of that word makes the signatories come off as pompous and self-important," which could easily be written off as merely par for the course in an academic setting. But, again, whether my reactions are representative of a meaningful percentage of the letter's target audience is uncertain, and is an empirical question which further research could perhaps elucidate.

  40. Nathan Sanders said,

    March 4, 2017 @ 6:03 pm

    I strongly suspect that the vast majority of native English speakers who truly understand what linguistics is would not balk at linguists calling themselves scientists. But this is only a hunch drawn from my informal observations, not a rigorous experiment.

  41. Sascha Griffiths said,

    March 4, 2017 @ 9:02 pm

    This is probably a peripheral point but Neil deGrasse Tyson's tweet is getting the linguist's role in the film wrong at another level. He assumes that there are no cryptographers in the movie. This appears to be incorrect.

    Jessica Coon, the linguistics consultant on the film, stated her instructions by the filmmakers in an interview as:
    "I was taken to the set, and into one of the military tents and they said, 'Your job is to translate an alien language. You have a team of 15 military cryptographers—tell us what happens next.'"

    To be fair, the teams which Louise Banks and Ian Donnelly lead were only mentioned once or twice in the film. There is, for example, one instance in which Dr. Banks says something to the effect that they would make more progress if the two teams worked together. However, even if it was never specified what kind of analysts the "teams" consisted of it seemed fairly obvious to me while watching the film that the military had already brought in some code breakers but they did not know how to extract the information needed which is why they bring in Louise Banks and Ian Donnelly.

    While Neil deGrasse Tyson tweeted that *a* cryptographer would have been useful I believe that the filmmakers figured that a whole tent full of them would not be enough.

  42. Lai Ka Yau said,

    March 5, 2017 @ 4:27 am

    @J. W. Brewer I understand and respect your point about the use of the words 'scientists' and 'laboratory', and indeed from the perspective of prototype theory, I don't think linguists are close to the prototypical scientist, or that endangered languages labs are close to the prototypical lab. However, I was wondering what you think of computer labs used by statisticians and computer scientists? Since such labs are generally not equipped with more advanced equipment than linguistics labs – I major in linguistics and statistics, and I think the ultrasound equipment is 'fancier' than what I find in the computer labs of the stats department – would you consider it strange to call them labs as well?

  43. GH said,

    March 5, 2017 @ 6:17 am

    @ J.W. Brewer:

    In the universities I have worked in, a "lab" does not primarily refer to a physical space, but to the smallest administrative unit of the research organization: a group of graduate students and postdocs working under one professor, sharing space, resources, budget, and usually some common research focus. Their "lab space" can be any room in which they do their research (e.g. if all graduate students share a large office, that might be considered the lab space), and may not feature any particular scientific equipment. The lab I was in had a lab space that was essentially a meeting room, optimized for conducting interviews and user studies, and with table and wall space for data analysis on paper.

    (Sometimes a lab grows beyond the "single professor" size. For example, at Cambridge the computer science department is called the "Computer Lab" for historical reasons.)

  44. Rodger C said,

    March 5, 2017 @ 1:00 pm

    @Zeppelin: When I took Comp. Lit. 501 in 1968 under Ulrich Weisstein, he made a great deal of the difference between vergleichende Literaturwissenschaft and vergleichende Literaturgeschichte. Do the young folks today learn that? I doubt it. *sniff*

  45. Stephan Hurtubise said,

    March 5, 2017 @ 4:50 pm

    I reached out to Dr. Jerry Coyne, the evolutionary biologist who took up the mantle of defending linguistics (well, Chomsky, at least) against Tom Wolfe in the Washington Post last year. He's been kind enough to share this letter with his readers over at Why Evolution Is True:

  46. Jason said,

    March 5, 2017 @ 7:26 pm

    While the linguists here understably wish to defend their territory, why has no one considered that an "amateur" hyperpolyglot like Richard Simmcott might actually be a better option over an academic linguist? This concept would even be Hollywood friendly, as they love stories where the plucky and highly gifted self-taught amateur runs rings around the professionals; witness what they did with "The Imitation Game", where they turned Joan Clarke (aka Alan Turing's love interest) from a dowdy career nerd who was headhunted because of her mathematics PhD from Cambridge into a Keria Knightly beautiful amateur who aces the admissions test after the evil sexists ask if her boyfriend is taking the exam.

    You can see it now: the intuitive hyperpolyglot (with a Dr. House type disease) who leaps ahead in intuitive leaps paried with an academic linguist who plods along, trying to restrain their more extreme flights of fancy.

  47. rcalmy said,

    March 5, 2017 @ 11:36 pm

    @J.W. Brewer

    Would you mind expanding on what in your opinion qualifies someone to be described as a scientist? I'm curious, given that you seem to take the position that it is not an appropriate descriptor of either linguists or of Dr. Tyson.

  48. db48x said,

    March 6, 2017 @ 6:44 am

    Nathan said "We couldn't do that with aliens, because we have no a priori notion of what they even talk about, and we certainly can't guarantee that our alien corpus would have suitable semantic similarity to any human corpus. And if even we could somehow guarantee it, we'd have no idea which human corpus to use."

    This isn't true though. If they build any high-tech artifacts (like spaceships, or radio telescopes), then they must know the laws of physics. The laws of physics are universal across the whole of the universe, and are described using mathematics. Thus we have a guaranteed 1:1 mapping between the mathematical subsets of any two languages, no matter how unrelated they are otherwise. From counting through arithmetic, algebra, and statistics you can proceed to descriptions of physical objects, and with that as a basis I believe it would be straight forward to start talking about language itself. (At least as straight forward as talking about language ever is; it is after all rather abstract. Luckily mathematics has some tools for dealing with abstractions, such as variables, self-reference, quoting, etc.)

    In fact, this is the basis of the Evpatoria message that was broadcast about 20 years ago: The message ( makes a great puzzle that anyone with a high-school education should be able to solve. Any 17th-century mathematician would be able to do most of it, though it contains certain facts that would be quite difficult to verify.

    I've no doubt that linguists would be the top choice though, particularly for face-to-face interactions.

  49. Idran said,

    March 6, 2017 @ 9:42 am

    @db48x: Have you read the actual short story that "Arrival" was based on? Because a big point of it was that our specific instantiation of the laws of physics aren't the only accurately descriptive and mathematically consistent one. I mean, I even remember seeing a fully-developed alternative physical model not even based on a number system at all but rather fully based on set theory, developed purely as a thought experiment to see if it was possible to do such a thing. There is no requirement of even arithmetic to develop explanatory and predictive models of the universe.

  50. Keith said,

    March 6, 2017 @ 10:32 am

    In the @ArrivalMovie I'd chose a Cryptographer & Astrobiologist to talk to the aliens, not a Linguist & Theoretical Physicist

    Neil deGrasse Tyson (@neiltyson), 1:40 PM – 26 Feb 2017

    I might have taken @neiltyson more seriously if he had written "I would have chosen …"

  51. Lai Ka Yau said,

    March 6, 2017 @ 11:20 am

    @db48x: I think that would invalidate Sapir-Whorf, which is central to the movie. In the famous 'Science and linguistics' paper, BL Whorf sketched a version of mechanics in Hopi terms that dispensed of notions dependent on the idea of time, and he believed that such a system of mechanics can also be consistent with reality.

    I also remember reading Ian Stewart write (I can't remember in which book) that a problem with sending mathematics to aliens is parochialism, because we have no way to know whether, say, pi is also a central part of mathematics in extraterrestrial cultures. I think it makes sense. Human statistics, for example, is based on probability theory, but what is to say that extraterrestrial statistics is also based on that? If the alien version of stats isn't based on probability, then the key theorems in modern human statistics (e.g. Central Limit Theorem, Laws of Large Numbers, Cramer-Rao inequality, Neyman-Pearson Lemma) won't make much sense to them either, since they all need the idea of probability.

  52. db48x said,

    March 6, 2017 @ 11:27 am

    Idran says: "[…] fully-developed alternative physical model not even based on a number system at all but rather fully based on set theory […]"

    No, I haven't read it, but if you want to posit aliens that developed set theory before counting, that's no problem. Humans started with counting and only later developed set theory, but we had set theory before we had radio telescopes. I'd say it's likely that some alien theorist came up with arithmetic as a theoretical exercise at some point, as an alternative to set theory. Later, having developed radio telescopes (or viable interstellar travel), they discover the Evpatoria message (or visit Earth), and discover that it's semantically isomorphic to that weird alternate form of mathematics called "arithmetics". It may take them longer to understand the message, but it won't be impossible.

  53. db48x said,

    March 6, 2017 @ 11:40 am

    Lai Ka Yau said: "[…] we have no way to know whether, say, pi is also a central part of mathematics in extraterrestrial cultures."

    Careful :)

    If they have circles, then they have pi (or possibly tau, or some other less likely multiple). To miss pi entirely they would have to miss quite a lot of other things as well, like complex numbers, sinusoidal waves, electromagnetism, etc. It's rather difficult to build radio telescopes without knowing about electromagnetism, I'll wager.

    The point about statistics is fair(er), though I point out again that any model that they have must necessarily be isomorphic to the laws of physics, and therefore to our own model. Note that we do have plenty of mathematics which is _not_ isomorphic to any known law of physics, but we don't use them to build radio telescopes (or spaceships, so far).

  54. J.W. Brewer said,

    March 6, 2017 @ 11:49 am

    To rcalmy's question, the answer for Dr. Tyson is simple: while his background and professional training is as a scientist, his present celebrity (and thus the reason anyone might care about what he says on the internet) is derived from his separate career as a show-business personality. I guess "science educator" is the least pejorative way to describe what he currently does, and one need not have a background as an actual scientist to do that well, any more than one needs to have a background as an actual cinematographer (or actor or what have you) to be a competent and effective movie critic.

    As to the lexical status of linguists, I agree with the point alluded to by Lai Ka Yu that it's often helpful to think of semantic scope in terms of prototypes and then other instances of varying distance from the prototype. My claim is that even though linguistics is presumably the same distance away from prototype-science as linguists are from prototype-scientists, the two words do not, at least in some varieties of English, have the same semantic scope and thus the distance-from-prototype at which they stop being the apt word to use is not the same. Simple parallel: there are lots of fields that are plausibly considered part of the "arts" whose practitioners are not typically called "artists." That's not a fact about the inherent dignity or professional qualifications or metaphysical status of practitioners in those fields, it's just a contingent fact about English usage, and the sort of "illogical" lack of complete symmetry between the semantic scope of related words that is ubiquitous in natural language but tends to create agita for prescriptivists.

    If one tries to abstract the notion of paradigm-scientist from a visual image involving white lab coats and bunsen burners, maybe a practical way to think of it is "how much of a typical scholarly article in the field involves either equations or the presentation and analysis of data in a statistically-rigorous way"? And of course there always has been a subset of linguistics work that was more like that, and perhaps that's been increasing over the three decades since it was (regrettably?) possible for me to obtain a B.A. in linguistics without using any math skills whatsoever. (And one can go too far the other direction, e.g. academic economics, on some accounts, has gotten so excessively math-skewed that people doing useful work in subfields that aren't as math-oriented allegedly feel constrained to stick gratuitous equations into their journal articles just so they give off the right superficial appearance.)

    I think "lab" turns out to be a more complex and polysemous word than I had unreflectively considered it to be when using it as, in hindsight, a poor parallel in an earlier comment. Indeed, I now recall that as a college freshman I was supposed to spend a certain number of hours a week at the "language lab" listening to recordings that would hopefully improve my aural comprehension of German. (By contrast, there was no "language lab" aspect of the class when I signed up for New Testament Greek or Old Irish . . .) That seems like probably an extended usage from "lab" in the school-jargon sense of "that thing you're supposed to do as part of an intro science class to practice practical skills at a different time and place than the regular class sessions where the professor is up front talking." But obviously the German-class usage of "lab" does not suggest that its student users are scientists-in-training or even that its faculty designers (optimistically assumed to be scholars who have thought in a rigorous, data-driven, and indeed "scientific" way about effective L2 pedagogy for 18-year-olds) are scientists.

  55. db48x said,

    March 6, 2017 @ 1:03 pm

    Incidentally, if anyone else is interested in translating the Evpatoria message, I recommend my own website at

  56. Jess Tauber said,

    March 6, 2017 @ 3:56 pm

    Linguists don't always have the facts they need, based on the expectations they have. For example I've discovered that in Zulu, some of the verbal extensions appear to have animacy and control features (similar in some ways to equivalent forms in Salishan languages) in addition to aspect notions. For example there are two different types of 'reversive' suffix, one of which is a duplicated version of the other. The simplex much more often deals with removal of home-grown coverings/occupants which can be thought of abstractly as 'feminine' in a fictive manner, while the duplications largely detail removal of artificially imposed and/or penetrating objects of a more linear character (fictively 'masculine). The 'feminine' is generalized to higher-dimensional objects (2 and 3 dimensions) while the 'masculine' generalizes to lower (0 and 1). As far as I can tell nobody had noticed this before. If human derivational categories can be such mysteries to us, imagine how alien ones might present problems.

  57. JPL said,

    March 9, 2017 @ 4:19 am

    Sorry to have missed this post on the Arrival film, since I loved it and was at the same time time disappointed by the skipping over of the linguist's procedures. (I haven't read the story.) I commented earlier that what seemed to be needed in this situation was a field worker, and specifically one who was expert in Ken Pike's "monolingual method". The request by the military people in the film for a translator was off the mark, since in that situation translation can't be used. The point of the monolingual method is to learn the language of the informant with a view to communicating with speakers of that language in that language. In the process of learning, one does not use translation, since there is no common language to use as a basis. Also, unlike normal language learning in the natural (non-classroom) context, one has the need to compress to a very brief period the time to communication. With the monolingual method, the give and take of social interaction is absolutely essential, especially establishing the signals for question and acceptable answer, puzzlement and satisfaction, so the point in the film where she takes off her helmet and approaches the barrier, I said, "Now we're getting started, that's more like it" (not holding up a board with the English word 'human' on it). (BTW, I suggested in an earlier comment that instead of looking in Chomsky's Aspects for helpful hints, she should have pulled out Turing's "Computing machinery and intelligence" or some such thing.)

    I haven't been able to read carefully all the comments, but I see no need to be hard on Prof. Tyson for his lack of familiarity with our obscure field; he definitely is playing a valuable role in countering the current anti-science attitudes and promoting the scientific approach to understanding the world. I would say, welcome to linguistics; get to know us better! This film Arrival raises some very interesting questions about the cosmic phenomenon of language and its roots in action and the logical structure of action that would bear further discussion. If I were in the story, I would be optimistic about the possibility of learning the "heptapod" language.

  58. Paul Topping said,

    March 9, 2017 @ 12:26 pm

    We'll only know which is right after first contact. I haven't read the short story but I didn't like the movie much and I'm a big fan of sci-fi, "first contact with aliens" in particular. Why only two experts anyway?

  59. MWarhol said,

    March 9, 2017 @ 4:57 pm

    The Academy has determined that Doctor Tyson possesses insufficient gravity to be classified as a scientist. The Academy has not yet decided whether he should now be classified as a scientoid or a scientesimal.

RSS feed for comments on this post