oxaliq.net/archive/beginning-latl.scm

(article
 (h1 "beginning work on LATL")
 (p
  "a little introduction to a project i started thinking about in early 2020,
   as i was learning how to do web development and wanted to try my hand at
   combining some hobbies")
 (p (a ((href "#the-point")) "skip the story and get to the point"))
 (section ((id "the-story"))
  (hgroup
   (h2 "what even is this?")
   (p (em "a story of linguistics and hubris"))
   (hr))
  (p
   "so, when i was a wee little child, i discovered language. like, most
    wee little children discover language, but some wee little children
    get given the lord of the rings when they're ten or something and get
    into the narrative and the characters, sure. but those appendices....
    there's .. a language here? called quenya? and it's written in this
    script? called tengwar? and there's a related language, sindarin? and
    this tolkien guy just made all this up? ")
  (p
   "so, i did the precocious kid thing and said to myself \"if this dead
    british weirdo (who's like really into the concept of royalty) can
    make a whole new language, then surely i can too\"")
  (p
   "i did not, in fact, make a language. i made a relex of the language
    i already spoke (my own peculiar idiolect of u.s. english,) but we all
    gotta start somewhere" )
  (p
   "for a long time the internet to me was mostly "
   (a ((href "https://listserv.brown.edu/archives/conlang.html"))
      "the brown university conlang listserv.")
   " here i learned about different phonology, what the heck
    morphosyntactic alignment is, diachronicity and how different
    languages can be related. i got exposed to awkwords and sca2 (tools
    for generating words and modeling changes to those words.) but i
    never interacted with ppl much. i was a kid and the ppl in there
    really knew their stuff, and i never much felt comfy with the idea
    of being just "
   (em "in public") " on the internet like that")
  (p
   "anyway i got better at making languages and learning about how
    language works, and it's just been a (sometimes more consistent, at
    times less) hobby of mine for most of my life")
  (p
   "flash forward to a few years ago, i'm working on a language that has
    turned into the passion project that i've spend the majority of my
    conlanging on and i'm getting frustrated with my tools. other
    conlangers have made some really cool tools that i've used forever,
    but i wanted to change my workflow. i wanted to do more complex,
    phonological-feature-aware, sound change rules; i wanted diachroinc
    tooling for morhposyntactic bits of my languages; i wanted a tighter
    feedback loop between defining some changes, or lexical items, or
    whathave you, and seeing the results; i wanted to see how a
    derivational pattern at one stage in a parent form of a language
    branches out or collapses into different derivational patterns in a
    child language")
  (p
   "all was not well in my conlanging")
  (p
   "i would need to " (strong "do some programming"))
  (p
   "the thing is, i had just learned javascript. i had built a few
    little toy web apps, and i was not ready to execute on a vision
    for a multi-purpose conlanging tool that was beginning to take shape in my head")
  (p "i tried anyway")
  (p "and i made "
     (a ((href "/settled/1"))
        "a bad first draft of a sound change tool"))
  (p
   "there's an e-bnf grammar in that project somewhere! the hubris i
    had then, to think i could make a little javascript-backed
    language tool with all of the bells and whistles i needed! with
    the knowledge that i had then! (or more accurately, didn't have)")
  (br)
  (p "but now..."))
 (section ((id "the-point"))
  (hgroup
   (h2 "reviving this project")
   (p (em "whatcha gonna do sorrel?"))
   (hr))
  (p "i'm still not ready")
  (p
   "but at least, today i'm forgoing bells and whistles for
    execution models. core abstractions. experience using "
   (a ((href "https://racket-lang.org"))
      "a (programming) language-oriented programming language")
   " maybe. there's a lot to think about")
  (p
   "so, i'm going to try reviving this project. or, more accurately,
    reimagining this project. from the little sound change tool that
    was merely an unusably buggy iteration on tools that other
    conlangers had executed better, to a more robust environment for
    conlanging. one which gives language nerds the power to solve
    all of their language nerding problems in one runtime, with an
    extensible and reflective interface. and which is written in such
    a way, with the appropriate abstractions, that lanuage nerdy
    hackers can hack their own tools on top of it")
  (p
   "and i'm going to be doing "
   (a ((href "https://www.recurse.com/"))
      "a hecking hacking retreat")
   " about it. i'll spend some full-time programming days making a
    goofy little thing for goofy little language nerds surrounded by
    goofy programming nerds doing whatever rad things they happen
    to be getting up to when i happen to be there")
  (p
   "like many of the communities i am a part of, i am a quiet part of
    the conlanging community. so, we'll see how this goes. maybe it
    is only ever something of interest to me and the conlangers i
    interact with in meatspace. maybe it only teaches me some things
    about designing and building a tool of the type that i'm designing
    and building. hoooopefully it is helpful in a way to other
    conlangers such that it evolves beyond what i could do on
    my own (cause i ain't no whiz kid 10x programmer)")
  (p
   "but that's all me getting ahead of myself (as is often my wont.)
    first! to pack for the trip!"))
 (section
  (hgroup
   (h2 "packing for the trip")
   (p (em "a biggg heap of things to expand"))
   (hr))
  (p
   "i'm just gonna fly right through these. there's going to be a lot
    to read up on, to explore, to think and talk through, to build
    toy versions of before it gets time for larger prototyping and
    building a workable tool. so, here's the very start of the
    thinking-in-public. each of these little thoughts is going to get
    at least it's own writeup (if not several) and i'll update with
    additional sections and links to the writeups as i go. a note: it
    would be impossible for me (and probably most ppl) to create the
    full vision in the time i'll be working on it fulltime")
  (section
   (h3 "notes on the goals")
   (p
    "it's good to have goals! to define what it even is yr tring to
     do. so here's my attempt. my goal is to create a foss+ working
     runtime that *can* serve as a singular tool for conlangers with
     some basic programming capacity. it should be linguistically
     *theory neutral* to the extant that such a thing is possible.
     it should allow for the definition of all relevant structures
     of a language *in code* and support the construction of grammars,
     dictionaries, text documents and (stretch goal) non-programmer
     friendly tools *from code*. it should support phonological,
     lexical, syntactic, morphological, semantic, etc transformations
     across different 'epochs' be they diachronic or synchronic.
     these transformations should work on transformations, ie
     morphological transformations should themselves be transformed
     by phonological transformations. users of the tool should be
     able to edit their work and see the consequences of their changes
     quickly. hopefully, this can all happen on the web, i guess,
     cause ppl like using their browser as the everything app"))
  (section
   (h3 "notes on existing tools")
   (p
    "i'm going to talk a bunch about existing tools made by
     conlangers or made for professional linguists and used by
     conlangers or just yr regular old spreedsheets and such. i'm
     going to be doing some research about how conlangers do their
     conlanging, lest this truly be a just-for-me type project"))
  (section
   (h3 "notes on primitives")
   (p
    "i think some interesting things start to bubble out of some of
     those goals up there. like the idea of supporting a bunch of
     different kinds of transformations that should themselves be
     transformable. the primitives we're working with have to work
     below the level of \"lexeme\" or \"phoneme\" to have something
     abstract that can apply to both. i'm not going to get into
     specifics even a little bit at this point-you'll have to stay
     tuned for more"))
  (section
   (h3 "notes on execution model")
   (p
    "this is going to be the crunchiest stuff to think about. i'm
     going to make some arguments for some things that (it is my
     belief) fall out of these goals. things like dynamic typing.
     things like reflectivity (this is sooo important it gets its
     own section.) things like lazy evaluation (because languages
     are big and sometimes you only want to look at a subset of a
     language.) there's also some hypothetical notions that will
     require some experiments to determine applicability to the
     project. i'm just going to list them as questions and not
     expound upon or defend their existence as questions here:
     propositions as a means of combination? propagation networks
     as primitives?"))
  (section
   (h3 "notes on reflectivity and interaction")
   (p
    "this is so important! i want conlangers to use this tool to
     iterate and that requires ~something like a REPL~ in a
     dynamic environment. my dream is to have a bidirectional
     flow of information; whereby a user of the system can define
     some rules about their language, evaluate those rules on some
     data in an interpreter, inspect those data, modify those data
     and run the interpreter such that it produces possible rules
     that would yield the new result. wether or not that experience
     results from the project, users should be able to rebind
     anything their project needs *at runtime*"))
  (section
   (h3 "notes on portability")
   (p
    "as the project develops i will have to give some thought to
     how to make it available to conlangers with different kinds
     of computation experience. the ideal would be for a full
     featured web implementation that interoperates with hosted
     implementations, but we'll see"))
  (section
   (h3 "notes on the substrate")
   (p
    "not a lot to say here. i will begin this project in racket
     because i have some familiarity with it and it is
     specifically built for developing languages. i will be using
     other tools to explore other aspects of the project and may
     ultimately move away from the racket ecosystem, but a girl
     has gotta start somewhere")))
 (section
  (hgroup
   (h2 "about the name")
   (p (em "it's pronouced ˈlæ.ɾɫ̩"))
   (hr))
  (p
   "i don't know dude, naming things is hard. it's a working name,
    it works good as an extension. "
   (code "my-conlang.latl") "anyone?")
  (p
   "i also just like syllabic laterals? they sound neat to me. the
    official initialism is " (u "l") "inguistic " (u "a") "nalytic "
    (u "t") "ransformation " (u "l") "anguage, but it also double
    entendres to " (u "l") "ower " (u "a" ) "nterior " (u "t")
    "emporal " (u "l" ) "obe, so that's just kind of fun.")
  (p
   "and the l means there can be lambda iconography-this is crucial
    for a new computer project. [ˈλæ.ɾɫ̩] as a typographic image is
    so perfectly specific to the kinds of things i dream about.")                             
  (p "maybe it'll have a different name some day, idk")))