GiellaLT provides an infrastructure for rule-based language technology aimed at minority and indigenous languages, and streamlines building anything from keyboards to speech technology. Read more about why, and see also our privacy document.
Lexicon files are a part of the langs/crk/src/morphology infrastructure. This documentation is not intended to be an exhaustive document for the structure of the lexicon, but so far just the lexicon elements that have an effect on the display of the dictionary through NDS.
Entry structure
level
Source attribute
In order to display a dictionary source in the entry, it should be included as an attribute on the node. This should be the full text that you wish to be displayed.
<e src="nêhiyawêwin : itwêwina / Cree : Words">
TODO: example image
level
TODO: example crk entry
level
contains one or more (translation group) which can contain: ### - a word `TODO: example entry with ` ### - a phrase `TODO: example entry with ` ### - An explanation: a sentence which explains the meaning of a word, but can't be used in the translation. `TODO: example entry with ` ### - Restriction - gives a restriction for the translation, f.ex. norwegian _vest_ has the restriction _of clothes_, to separate it from the navigational direction. `TODO: example entry with ` ### level TODO: ### Audio specifications fall under the node in an node. Within this node are nodes, which contain the following attributes: - @href - a link to the audio file, hosted on some server - @speaker - initials, or anonymized name for the speaker - @dialect - a code for the dialect. There may be more attributes, use whatever you like and document them here. NDS will chiefly rely on @href, but may use other attributes such as @dialect, in order to render more information to the user on what they will hear. Filenames should be unique to the sound. If the sound needs to reappear anywhere, the same filename may be used. ``` lemma ``` During the lexicon build process, audio will be extracted and compressed for web-friendly formats, so that files dictionary users need to download are wtihin the ~50kb range. Locally hosted paths to the file will be automatically added during the build process as long as filenames are unique, so that files specified may be whatever size. ### node Currently the node contains the word inflectional class. For some words it will be used to determine how to generate a miniparadigm. ``` kiskinwahamâkêwikamikohkêwVAI-v ``` It is also used to display the inflectional class alongside the POS. ``` VTA-1 -> wâpamêw (Verb, VTA) ``` See the userfriendly tag relabels file for `crk` to adjust how any of these appear in the interface. ### level TODO: any attributes ### node The stem will be included with the analysis column along the right side of all entries that match up with an analysis. The stem will be pulled from the node within the node. ``` kiskinwahamâkosiwVAI-vkiskinwahamâkosi ``` The stem will also be used in analyses along the right side. ``` nikî-nitawi-kiskinwahamâkosin ⟵ PV/nitawi + √ kiskinwahamâkosi- + Intransitive Animate Verb + Independent + Past + 1s ``` TODO: ### attribute documentation TODO: ### Example sentences TODO: ### Other files TODO: ## Generated miniparadigms Miniparadigms are generated in lexicon entries in order to help users. They vary from POS to POS and sometimes within POS. There are also some constraints on generation for orthographical reasons. ### FST restrictions # `+Err/Orth` - Forms marked with this will not be displayed in generation. ## This may affect forms without hyphens, where hyphens are expected. # `+Use/Ped` - ? # `+Use/NGminip` - ? ### Lexicon attributes in XML that affect generation For now generation relies purely on animacy tags in the FST. ### Verbs **TA** | person | INDEPENDENT | tag | CONJUNCT | tag | | ----------------- | ----------- | ---------------------- | ---------- | ---------------------- | | 1s → 3s | niwâpamâw | V+TA+Ind+Prs+1Sg+3SgO | wâpamak | V+TA+Cnj+Prs+1Sg+3SgO | | 2s → 3s | kiwâpamâw | V+TA+Ind+Prs+2Sg+3SgO | wâpamat | V+TA+Cnj+Prs+2Sg+3SgO | | 1p exclusive → 3s | niwâpamânân | V+TA+Ind+Prs+1Pl+3SgO | wâpamâyâhk | V+TA+Cnj+Prs+1Pl+3SgO | | 1p inclusive → 3s | kiwâpamânaw | V+TA+Ind+Prs+12Pl+3SgO | wâpamâyahk | V+TA+Cnj+Prs+12Pl+3SgO | | 2p → 3s | kiwâpamâwâw | V+TA+Ind+Prs+2Sg+3SgO | wâpamâyêk | V+TA+Cnj+Prs+2Sg+3SgO | (Example table for now, but copy format for other tables) **II** **AI and TI** **Context** Independent pronouns are included for context. ### Nouns ### Proper nouns ### Adjectives ### Numerals ### Pronouns ### Personal pronouns ## Pregenerated paradigms Plains Cree currently does not reply on pregenerated entries in NDS. ## Analysis specifics Plains Cree has some tags that appear before the lemma. There is a special rule in NDS to handle this, but more tags may need to be added occasionally. TODO: document tags # FST Miscellany - Preverbs and reduplication tags come before the lemma, and so there is a tag reformatter in place which shifts these tags after the lemma, and may replace the slot formerly occupied by the lemma with `Tpl/Lemma` for analysis work in the templates.