Beserman morphological analyzer

This is a rule-based morphological analyzer for Beserman (formerly a dialect of Udmurt udm; Uralic > Permic). It contains a formalized description of Beserman morphology as established by the Beserman documentation project, based mostly on the spoken data from the village of Shamardan (Yukamenskoye district, Udmurtia). It uses uniparser-morph for parsing. It performs full morphological analysis of Beserman words (lemmatization, POS tagging, grammatical tagging, glossing).

This package uses a project-internal Latin-based spelling system. Cyrillic and UPA analyzers will hopefully follow later. Right now, see translit-udmurt for transliteration options.

Warning: This is a project-internal tool. If you think you might need it, you are probably wrong. If what you need is standard Udmurt analyzer, you can find one here. If you are not sure, feel free to send an email to the developer (Timofey Arkhangelskiy, timarkh@gmail.com).

How to use

Python package

The analyzer is available as a Python package. If you want to analyze Beserman texts in Python, install the module:

pip3 install uniparser-beserman-lat

Import the module and create an instance of BesermanLatAnalyzer class. After that, you can either parse tokens or lists of tokens with analyze_words(), or parse a frequency list with analyze_wordlist(). Here is a simple example:

from uniparser_udmurt import BesermanLatAnalyzer
a = BesermanLatAnalyzer()

analyses = a.analyze_words('Gožtemjosəz')
# The parser is initialized before first use, so expect
# some delay here (usually several seconds)

# You will get a list of Wordform objects
# The analysis attributes are stored in its properties
# as string values, e.g.:
for ana in analyses:
        print(ana.wf, ana.lemma, ana.gramm, ana.gloss)

# You can also pass lists (even nested lists) and specify
# output format ('xml' or 'json')
# If you pass a list, you will get a list of analyses
# with the same structure
analyses = a.analyze_words([['A'], ['Mon', 'tone', 'jaratišʼko', '.']],
	                       format='xml')
analyses = a.analyze_words(['Gožtemjosəz', [['A'], ['Mon', 'tone', 'jaratišʼko', '.']]],
	                       format='json')

Refer to the uniparser-morph documentation for the full list of options.

Disambiguation

Apart from the analyzer, this repository contains a set of Constraint Grammar rules that can be used for partial disambiguation of analyzed Beserman texts. They reduce the average number of different analyses per analyzed token from about 1.7 to about 1.4. If you want to use them, set disambiguation=True when calling analyze_words:

analyses = a.analyze_words(['Mon', 'tone', 'jaratišʼko'], disambiguate=True)

In order for this to work, you have to install the cg3 executable separately. On Ubuntu/Debian, you can use apt-get:

sudo apt-get install cg3

On Windows, download the binary and add the path to the PATH environment variable. See the documentation for other options.

Note that each time you call analyze_words() with disambiguate=True, the CG grammar is loaded and compiled from scratch, which makes the analysis even slower. If you are analyzing a large text, it would make sense to pass the entire text contents in a single function call rather than do it sentence-by-sentence, for optimal performance.

Word lists

Alternatively, you can use a preprocessed word list. The wordlists directory contains a list of words from a 200-thousand-word Beserman multimedia corpus (wordlist.csv), list of analyzed tokens (wordlist_analyzed.txt; each line contains all possible analyses for one word in an XML format), and list of tokens the parser could not analyze (wordlist_unanalyzed.txt). The recall of the analyzer on the corpus texts is about 98%.

Description format

The description is carried out in the uniparser-morph format and involves a description of the inflection (paradigms.txt), a grammatical dictionary (lexemes.txt), and a short list of analyses that should be avoided (bad_analyses.txt). The dictionary contains descriptions of individual lexemes, each of which is accompanied by information about its stem, its part-of-speech tag and some other grammatical/borrowing information, its inflectional type (paradigm), and Russian translation. See more about the format in the uniparser-morph documentation.