[HN Gopher] The race to teach sign language to computers
       ___________________________________________________________________
        
       The race to teach sign language to computers
        
       Author : sohkamyung
       Score  : 45 points
       Date   : 2021-03-07 12:41 UTC (2 days ago)
        
 (HTM) web link (www.economist.com)
 (TXT) w3m dump (www.economist.com)
        
       | vitorbaptistaa wrote:
       | Portuguese-only, but in Brazil we have VLibras [1], which is a
       | free open-source software to automatically translate between
       | written text and a 3D model using sign language.
       | 
       | [1] https://www.vlibras.gov.br/
        
         | elliekelly wrote:
         | That's really cool. Was this funded by the government? I ask
         | because it's on a .gov site.
        
       | ars wrote:
       | Can someone explain why this is necessary (important) when typing
       | into a computer seems like a reasonable alternative?
       | 
       | I'm tying to make an analog to Google home like devices, but I
       | can't imagine someone putting cameras all over a room to capture
       | sign language from any angle.
       | 
       | Can someone help me understand what's the use case for computers
       | understanding sign language that is not served by typing?
        
         | toomuchredbull wrote:
         | What if you had deaf friends and could wear glasses that would
         | translate sign language for you? What if you were deaf and had
         | a robotic assistant? What if you wanted to do translations for
         | a hearing audience?
        
         | curtisf wrote:
         | Sign language is not a way to speak English using your hands --
         | it is a distinct language with distinct grammar and distinct
         | vocabulary.
         | 
         | If you are born Deaf, you will not have the opportunity to
         | natively acquire a spoken language. Our brains are not designed
         | to acquire spoken language through writing -- and our spoken
         | languages are designed to be acquired through listening, not
         | reading.
         | 
         | It's easy to think that a Deaf signer living in America is a
         | native speaker of English, but they very well may not be,
         | having never heard the sounds and therefore been unable to
         | internalize all of the many, many necessary grammatical and
         | morphological rules of the language.
         | 
         | Expecting all Deaf people to simply use written English when
         | dealing with a computer is like expecting all Americans to use
         | Spanish to interact with their computers -- technically
         | possible (Spanish is a perfectly good language, and relatively
         | accessible starting even from grade school here), but ignores
         | the human reality that people want (need) to speak their own
         | language.
         | 
         | There are not yet standard ways to write sign languages. Partly
         | this is because of difficulties in translating the medium to
         | text -- unlike speech, sign language production is not
         | _entirely_ linear, in addition to the fact that your hands can
         | be articulated comprehensibly in much more sophisticated ways
         | than your mouth can.
        
           | ars wrote:
           | Does that mean people with use Sign language as their first
           | language do not read books? How do they learn anything?
           | 
           | It really seems to me you are implying they can't read
           | regular English - doesn't that make regular life quite a
           | challenge? I would have assumed learning written English
           | would be high priority.
           | 
           | Your answer just makes me even more confused. If they can't
           | read English, what is the computer translating the sign
           | language into? What are they doing with the results?
           | 
           | And this just brings up even more questions - how do they
           | speak remotely to people? I was thinking that a video phone
           | was an amazing advance, except you'd have to put the phone
           | several feet away from you, making it pretty hard to see the
           | screen.
           | 
           | Every single scenario I'm thinking of just makes the need for
           | written language even more critical - otherwise such a person
           | is almost entirely isolated, except when near other signing
           | people.
        
             | cowboysauce wrote:
             | Most people who know ASL can also communicate in English,
             | but language skills are a spectrum and some people may know
             | enough to get by in daily life, but may not know enough
             | English to understand more complicated topics. It's really
             | no different than a native Spanish speaker living in the
             | United States. Some may only know a basic level of English
             | while others may be equally comfortable in both languages.
             | 
             | This isn't even unique to the Deaf community. Something
             | like 15 to 20 percent of the general population are
             | considered functionally illiterate.
        
           | bluefirebrand wrote:
           | > Expecting all Deaf people to simply use written English
           | when dealing with a computer is like expecting all Americans
           | to use Spanish to interact with their computers
           | 
           | Kind of a false equivalence here. For many Americans there is
           | no real reason to learn Spanish. However, I have to imagine
           | that for literally all deaf people learning to read and write
           | the dominant local language is required in order to
           | participate in the larger society. Signs will be written in
           | that language, restaurant menus, newspapers, business emails,
           | etc. You will be surrounded by writing of your local language
           | every day.
           | 
           | There doesn't really exist a world where a person can get by
           | without being able communicate with others in their vicinity.
           | Deaf people are more often surrounded by non-deaf people who
           | they will have to communicate with somehow. Sharing a common
           | spoken language is kind of out of the question, sharing a
           | common written language is the next best thing.
           | 
           | So, sure. Create a written sign language, make unicode
           | support for it and create new locale settings for each
           | dialect of sign language. This will not eliminate or even
           | really reduce deaf people's need to learn to write their
           | locally dominant language.
        
         | dlandis wrote:
         | One thing to keep in mind is that ASL, for example, is the
         | first language of many people and it is a totally different
         | from the English language.
        
       | kingant wrote:
       | Check out this paper: https://arxiv.org/abs/1908.10546 (not
       | mine)...but links to some recent work using deep learning.
        
       | karlkatzke wrote:
       | Complete aside: How cool of a name is "Dr. Zsolt Robotka"? Pretty
       | sure Dr. Robotka will also be the name of the next Bond villain.
        
       | neonate wrote:
       | https://archive.is/ex59g
        
         | _Microft wrote:
         | If you wonder why you might not be able to access this, keep in
         | mind that archive.is and Cloudflare DNS do not play nicely with
         | each other [0]. It might also bite you if you are using Firefox
         | because of the DNS over HTTPS service from Cloudflare that
         | Firefox uses [1].
         | 
         | Loading the article directly into Firefox's reader mode by
         | prepending _about:reader?url=_ to the URL works fine though.
         | 
         | [0] https://twitter.com/archiveis/status/1018691421182791680
         | 
         | [1] https://support.mozilla.org/en-US/kb/firefox-dns-over-https
        
       | runnr_az wrote:
       | Signal me when they've taught em semaphore.
        
         | jfengel wrote:
         | That should be way, way simpler. It sounds like a neat computer
         | vision class project.
        
           | runnr_az wrote:
           | That's what I'm sayin'... duolingo for flag communication.
        
       | TheAdamAndChe wrote:
       | I can't read it, but what makes interpreting sign language so
       | difficult? You can have a model that inputs two pictures of the
       | person 5/60 seconds apart. If the change in movement is below a
       | certain threshold(indicating a word is being displayed) then crop
       | it around the hands and run it though a resnet trained to
       | categorize images based off of words.
       | 
       | Is there something I'm missing? If it's such a big deal then I
       | may work on it myself.
        
         | schwartzworld wrote:
         | American Sign Language grammar is subtle and complex.
        
         | sohkamyung wrote:
         | This paragraph from the article highlights some of the
         | problems.
         | 
         | > Research led by a team at Microsoft, a big computing firm,
         | and published in 2019, estimated that a typical publicly
         | available corpus of a spoken language consists of around a
         | billion words from as many as 1,000 different speakers. An
         | equivalent data-set in a sign language might have fewer than
         | 100,000 signs from just ten people. Besides large numbers, a
         | good corpus also needs variety. This means conversations
         | between native signers of diverse backgrounds, dialects and
         | levels of fluency. Because deaf people more often have physical
         | disabilities than do those with unaffected hearing,
         | representing those with restricted fluency of movement is
         | important.
        
         | PeterisP wrote:
         | Fragmentation also is a big issue. Not only in the sense that
         | there are separate very different sign languages (American Sign
         | Language has little in common with British Sign Language, they
         | are not mutually comprehensible), but even for each separate
         | sign language you also get different 'dialects' in different
         | areas/major cities, which are mutually comprehensible but have
         | lexical variation depending on the speaker's location/community
         | that needs to be properly represented in your datasets in order
         | to be acknowledged and understood. In some sense it seems the
         | equivalent of what the fragmentation was for spoken languages
         | before the printing press and age of nationalism facilitated
         | language standardization.
        
         | hmry wrote:
         | Sign languages don't only use hand shape, but also hand motion,
         | body pose, facial expression, mouthing words, etc. Cropping
         | still images around the hands won't do you much good at all.
        
           | medstrom wrote:
           | Not to mention polysynthetic signs - I don't want to be the
           | one to program that.
           | 
           | An example sentence of mostly polysynthetic signs:
           | 
           | "i was out driving, just enjoying the scenery, the nature and
           | all the trees passing by me," (3 signs) "then suddenly" (just
           | raise the eyebrows and look scared at some point in front of
           | you, with your hands still on the imaginary steering wheel),
           | "my car crashed into a wall (or other immobile object) (and
           | the hood got all deformed)" - 1 sign.
           | 
           | So that takes about 5 seconds to sign, but up to a minute to
           | represent all the embedded information in English. But in
           | practice, you'd skip a lot of the embedded information when
           | translating to English, depending on what was actually
           | important. So how do you make the machine know what was
           | important? It depends on who you're translating for!
        
         | codyogden wrote:
         | All sign languages use much more than handshape and position.
         | And even outside of facial expression grammar (ie non-manual
         | communication) there is a whole host of other complexities that
         | make parsing sign language difficult.
         | 
         | Take classifiers: there are eight morphological types that
         | allow users to express complex ideas without the use of any
         | type of vocabulary-based utterances. You can certainly sign the
         | equivalent of, "Did you see the car parked close to the tree,"
         | (CAR PARK NEAR TREE SEE?) using vocabulary. Sure. It gets the
         | point across, but it's not efficient, and doesn't really
         | utilize sign language in a way that feels normal for users.
         | 
         | But that same question can be posed in a more condensed format
         | if you use a classifier, (#CAR 5:CL-"tree"/3:CL-"car parked
         | really close" SEE?). In the second utterance, one use the
         | lexicalized (fingerspelled "car"), then holds up a five hand
         | with bent elbow in their non-dominant hand, use the three hand
         | representing the car placing it reeeeeeally close (with
         | grammatical, NMC emphasis) to the tree, and finally asking the
         | receiver if they saw it.
         | 
         | In that example alone, I'm using two types of classifiers:
         | element and locative. Imagine a complex story using all eight
         | classifier types with just a little vocabulary mixed in? I
         | doubt AI/ML could interpret the context, characters, actions,
         | events, descriptions, etc. accurately. Anyway, I hope this
         | example shows there is a lot more complexity in computerized
         | interpretation of signed languages well beyond
         | gesture/vocabulary recognition.
         | 
         | Note: My glossing may not be perfect. It's been a while, and
         | I'm a bit rusty.
        
           | whitten wrote:
           | Is this an ASL thing or some other system?
           | 
           | What would be a good source to get the full list of the eight
           | morphological types, and how they are used?
        
             | codyogden wrote:
             | I learned this in a Linguistics of ASL course as a part of
             | my degree in ASL. My understanding is that classifiers
             | exist in every natural signed language and isn't exclusive
             | to ASL.
             | 
             | https://www.allenisd.org/cms/lib/TX01001197/Centricity/Doma
             | i...
             | 
             | The linked PowerPoint does a good job of explaining a brief
             | overview of the different types of classifiers. (I actually
             | referenced it to correct my own answer because I knew
             | something felt off. lol)
        
           | codyogden wrote:
           | Correcting myself: The classifiers used in my example are
           | "semantic and locative," not "element."
        
         | medstrom wrote:
         | Essentially, vocal languages are one-dimensional, thus easy to
         | solve. Filter out disturbances, background noise, and train the
         | machine to adapt to dialects, and you're halfway home.
         | 
         | Sign language is at least three-dimensional, it can embed more
         | information. The ASL signs for "forget" and "because" differ
         | only in how you mouth them - easy enough, but sometimes you
         | even skip mouthing them because there is sufficient context.
         | What's worse is the location of the hands can vary depending on
         | how lazy the signer is feeling, so "forget" is supposed to be
         | signed from the forehead but it can also be signed from some
         | empty point in front of your shoulder, off a "phantom person"'s
         | imaginary forehead. You rely on the listener to understand that
         | you intend for there to be a phantom person there.
         | 
         | Parsing sign language means you have to turn a 2D recording
         | into a 3D model - even humans can have difficulty with this,
         | parsing signs over Zoom - and you need to solve parsing facial
         | expressions, reading lips and more. _And_ you need the machine
         | to be smart enough to detect the assignment of phantom persons
         | to various locations around the signer. That 's my impression
         | as a deaf person anyway.
        
           | kleiba wrote:
           | _Essentially, vocal languages are one-dimensional, thus easy
           | to solve._
           | 
           | cough.
        
       ___________________________________________________________________
       (page generated 2021-03-09 23:03 UTC)