morphologytwo

Morphologytwo is a theoretical framework and computational model for analyzing and generating word forms in natural language. It treats morphology as a two-layer system consisting of a lexical layer that encodes base meaning and grammatical features, and a surface layer that encodes phonological form and inflectional realization. The two layers are linked by a mapping that assigns surface realizations to combinations of lexical content and morphosyntactic features, allowing the model to separate semantic intent from grammatical realization.

Core concepts include dual-layer representation, a concise set of correspondence rules, and the ability to integrate finite-state, statistical, or neural components. This separation enables the framework to capture cross-linguistic inflection and derivation while remaining compatible with both rule-based and data-driven approaches.

In computation, a Morphologytwo system stores lexical entries with features; the generation module constructs surface forms by applying feature-to-form rules, while the analyzer inverts the process to predict features from a surface form. The architecture supports hybrid implementations and can accommodate languages with rich morphology, including affixal and non-concatenative processes.

Examples include English forms such as walk plus past tense yielding walked, and walk plus present participle yielding walking; in agglutinative languages, the surface form results from a sequence of affix realizations mapped from the surface layer. Compared with traditional two-level morphology, Morphologytwo emphasizes modularity and explicit separation of meaning and form, and aims to support typological research and multilingual NLP.

Applications span morphological analysis and generation in NLP pipelines, language education tools, and cross-language typology studies.