LT-PROC(1)

📦
version 3.8.3 · 2 versions
, in ubuntu stonking

💖 LT-PROC(1) — the lt-proc era 💖

okay bestie, pull up a chair, because we're talking about lt-proc, the lexical processor for Apertium, from the General Commands Manual. this is the girl who does the actual ~thinking~ about words. she's a whole icon. let's go 🦋

✨ NAME ✨

lt-proc — lexical processor for Apertium. that's it, that's the girl. 💅

✨ SYNOPSIS ✨

lt-proc
        [-a | -b | -o | -c | -d | -e | -g | -h | -p | -s | -t | -v | -h | -z | -w]
        [-W] [-N -N] [-L -N] [-i icx_file] fst_file
        [input_file [output_file]]

(and yes, -h really does show up twice up there. she's just that into helping, i guess. we don't question it. 🙈)

✨ DESCRIPTION ✨

so lt-proc is the application that serves up the four lexical processing functionalities. four. a whole squad:

  • morphological analyser (option -a)
  • lexical transfer (option -n)
  • morphological generator (option -g)
  • post-generator (option -p)

and how does she do all that? not by guessing, bestie. she reads binary files that contain a compact and efficient representation of dictionaries — a class of finite-state transducers called augmented letter transducers. and those files? they're generated by lt-comp(1). so like, lt-comp makes the dictionaries, lt-proc uses them. it's giving collaborative. it's giving duo. 🫶

now pls pay attention because this part matters: some characters are special chars used for format and encapsulation — specifically [, ], $, ^, /, +. if you actually want to use them literally, you have to escape them. for instance, [...] get ignored, and the format of a linefeed is ^...$. lowkey important. don't forget it. no cap. ⚠️

✨ OPTIONS ✨

she has a LOT of them. every one of them. let's go down the whole list, no skipping, periodt.

  • -a, --analysis — tokenizes the text into surface forms (aka lexical units, the way they actually show up in real texts) and then, for every single surface form, hands you back one or more lexical forms made up of lemma, lexical category, and morphological inflection info. and ngl, the tokenization is not straightforward, for two reasons: contractions exist, and multi-word lexical units exist. for contractions, the system takes in one surface form and spits out the corresponding sequence of lexical forms. multi-word surface forms get analysed left-to-right, longest-match style. those multi-word surface forms can be invariable (think a multi-word preposition or conjunction) or inflected — like, in es, “echaban de menos”, “they missed”, is a form of the imperfect indicative tense of the verb “echar de menos”, “to miss”. 😭 she's so smart. there's also limited support for some kinds of discontinuous multi-word units. and single-word surface form analysis gives you output like these examples:

“cantar” → “^cantar/cantar<vblex><inf>$” or “daba” → “^daba/dar<vblex><pii><p1><sg>/dar<vblex><pii><p3><sg>$”.

  • -b, --bilingual — does lexical transference, attaching queues of morphological symbols that aren't specified in the dictionaries. just like analysis mode, she supports multiple lexical forms in the target language for a given lexical form in the source language. typically works with the output of apertium-pretransfer(1). 💫
  • -o, --surf-bilingual — same vibe as -b, BUT it takes input from apertium-tagger(1) -p, with surface forms, and if the lexical form isn't found in the bilingual dictionary, it just outputs the surface form of the word. she's flexible like that. 🌸
  • -O, --surf-bilingual-keep — same as -o, but it also outputs the surface form before the readings. extra. we love extra. 👑
  • -c, --case-sensitive — uses the literal case of the incoming characters. caps lock respected fr fr.
  • -d, --debugged-gen — morphological generation, but with all the stuff. yes. all of it. every last crumb. 🍓
  • -e, --decompose-compounds — tries to treat unknown words as compounds, and decompose them. she's doing detective work. 🔍
  • -w, --dictionary-case — uses the case information contained in the lexicon, instead of the surface case. heads up: this one is only applied in analysis mode.
  • -g, --generation — delivers a target-language surface form for each target-language lexical form, by suitably inflecting it. generation era. ✨
  • -n, --non-marked-gen — morphological generation (like -g) but without the unknown word marks (that's the asterisk *).
  • -b, --tagged-gen — morphological generation (like -g) but retaining part-of-speech tags.
  • -p, --post-generation — performs orthographical operations like contractions and apostrophations. the post-generator is usually dormant (she just copies the input to the output 😴) until a special alarm symbol contained in some target-language surface forms wakes her up to perform a particular string transformation if necessary — and then she goes right back to sleep. nap queen. main character energy. 💜
  • -s, --sao — input processing is in orthoepikon (previously sao) annotation system format: https://orthoepikon.sf.net.
  • -t, --transliteration — apply a transliteration dictionary.
  • -i icx_file, --ignored-chars icx_file — ignores characters specified in the file icx_file. out of sight, out of mind. 🙈
  • -z, --null-flush — flush output on the null character.
  • -C, --careful-case — use dictionary case if it's present, else surface. she checks first. careful girlie. 🎀
  • -N, --analyses — output no more than N analyses (and if the transducer is weighted, it's the N best analyses). only the best for you.
  • -L, --weight-classes — output no more than N best weight classes (where analyses with equal weight constitute a class). groups only. 🫶
  • -W, --show-weights — print final analysis weights (if any).
  • -v, --version — display the version number.
  • -h, --help — display this help.

✨ FILES ✨

  • input_file — the input compiled dictionary. 🗂️

✨ SEE ALSO ✨

apertium(1), apertium-tagger(1), lt-comp(1), lt-expand(1) — the whole friend group. go say hi. 🦋

Copyright © 2005, 2006 Universitat d'Alacant / Universidad de Alicante. this is free software. you may redistribute copies of it under the terms of the GNU General Public License: https://www.gnu.org/licenses/gpl.html. 💖

✨ BUGS ✨

Many... lurking in the dark and waiting for you! 💀😭

not even joking bestie, "Many... lurking in the dark and waiting for you!" is the actual bug report. they really said no further comment. be careful out there. 🔥

and that's the whole page!! 💖 huge shoutout to Apertium and to that March 23, 2006 date — the girl has been serving lexical processing since 2006 and she's still iconic. mwah. 🎀✨👑

📦 where this page comes from