ParleLint

ParleLint is a static analysis tool that scans source code and detects comments that are not fully bilingual (English and French). It enforces a policy where every comment must have a parallel translation. The tool integrates with popular linters like ESLint and Pylint, and with CI pipelines. It uses language detection and machine translation to suggest missing translations. Built for Montreal-based companies, it addresses the city's cultural and legal expectations of bilingualism in the workplace, ensuring codebases are inclusive for both English and French-speaking developers.

In Montreal, many software companies operate in a bilingual environment. However, code comments are overwhelmingly written only in English, which can alienate French-speaking developers and may violate local language policies. Existing linters do not enforce language requirements, leaving teams to manually check comments. Developers need a tool that automatically detects incomplete translations and promotes bilingual inclusivity.

• Detect comments that lack a translation in the other official language
• Provide automated translation suggestions via machine translation APIs
• Integrate seamlessly with ESLint, Pylint, and other popular linters
• Offer configuration for language pairs (e.g., English-French)
• Generate compliance reports for audits
• Support both inline and separate translation storage modes

• Not a full translation service—translations are suggestions, not mandatory
• Does not enforce language on variable names, strings, or code logic
• Not a general i18n tool
• Will not guarantee translation accuracy for highly technical terms

Node.js core, Language detection library (franc), Machine translation API (DeepL or Google Translate), ESLint plugin API, Pylint plugin API, GitHub Actions for CI integration

ParleLint is implemented as plugins for existing linters. It works by scanning AST comment nodes, using a language detection library (e.g., franc) to identify the primary language. If a comment is monolingual and no corresponding translation is found within a configurable distance (e.g., next line or block), it flags a warning. For auto-fix, it calls a translation API and inserts the translated comment below or adjacent. The plugin is configurable for language pairs and output style.

• False positives: comments like acronyms or code snippets that shouldn't be translated
• Machine translation quality may be poor for domain-specific terms
• Performance overhead from language detection and API calls on large codebases
• Developer resistance—some may prefer English-only comments
• Maintenance burden of keeping translation API keys and handling API changes

• Should we support language pairs beyond English-French? (e.g., for Brussels)
• How to handle very long comments? (truncate, split, or skip?)
• Should translations be stored inline (same line or next) or in separate .po files?
• What threshold of confidence for language detection should trigger a warning?

The idea was presented at a Montreal tech meetup but gained little traction due to the niche market—most startups preferred English-only codebases. The manual effort to refine translations and handle edge cases also felt too high for a side project, so it remained a plan.

ParleLint could be extended to other bilingual cities like Brussels (French-Dutch) but was originally conceived for Montreal's specific linguistic landscape. The tool's name is a portmanteau of 'parler' (to speak) and 'lint'.