I've developed some small-scale tools for corpus analysis and management. All are available under an open source licence via Sourceforge:
Concordance Manager (https://github.com/dfg-silpac/conman):
Specialist tool for managing and annotating concordances. In particular, it was designed to facilitate substantial post-processing of data extracted from closed-source corpora (e.g. pos tagging, lemmatization, or parsing) and conversion of a variety of outputs into a standard format
2026 status: Version 1.2.0 published January 2025. Minor developments still ongoing.
Old French Lemmatizer (https://github.com/ILR-Stuttgart/old-french-lemmatization-tools):
Wrapper for Helmut Schmid’s excellent RNN Tagger (must be downloaded separately here), improves lemmatization by adding lexical lookup from data drawn from gold lemmatized corpora, by allowing lemma disambiguation to be based on Gold Part-of-Speech tags, and by adding a ‘reliability score’ for each lemma to guard against hallucinations and signal common sources of error. Adds multiple input/output formats (CONLL-U, csv, some XML) and converts part-of-speech tags to UD to allow it to combine data from multiple sources.
2026 status: Complete. Minor developments still ongoing.
Syllabic Verse Analysis (https://sourceforge.net/projects/syllabic-verse-analysis/):
Script designed to assist in the generation of metrical annotation for Romance syllabic verse, essential for the creation of the Old Gallo-Romance Corpus. The first stage splits orthographic forms into syllables while the second stage scans the result assigning each syllable to a metrical position in the line of verse. Exports to PAULA-XML suitable for use with ANNIS. See Rainsford (2022).
2026 status: Fully functional, major upgrade in progress (August 2026)
Tokenized Text Aligner (https://sourceforge.net/projects/tokenized-text-aligner/):
Automatically aligns two similar versions of the same text token by token. Useful when combining annotation from corpora with different tokenization policies and/or comparing different editions of the same manuscript and/or different manuscripts of the same text. The quality of the result obviously depends on the similarity of the source texts but I've found it to be surprisingly robust.
2026 Status: Version 2.1.0 complete (January 2026). Future development if necessary.
KNIC Concordances (https://sourceforge.net/projects/knicconcordances/):
Backend for TIGERSearch/TIGER-XML to generate concordance-style tabular results from treebank queries. See also Rainsford and Heiden (2014).
Status: Complete 2014. No future development planned.
IMPORTANT CAVEAT: These tools are provided as-is without warranty or guarantees any kind. In particular, they are developed on Linux and I have no plans to test them on other operating systems.