Morphoscope

Morphoscope is a small SWI-Prolog morphological analyzer for English and Spanish verbs.

It maps a word form to a lemma and grammatical features. It also generates every supported form for a known lemma.

Value

The project demonstrates bidirectional morphology in a compact, inspectable codebase.

The English lexicon contains 224 lemmas with regular and irregular entries. The Spanish lexicon contains regular -ar, -er, and -ir verbs plus common irregular paradigms.

Ambiguous forms return multiple answers. For example, walked can be a past form or a past participle.

Architecture

• morphoscope.pl exposes the public API.
• src/engine.pl validates input and dispatches language operations.
• src/languages/english.pl stores English data and spelling rules.
• src/languages/spanish.pl stores Spanish data and conjugation rules.
• bin/morphoscope.pl provides a command-line demo.
• test/test_morphoscope.pl contains deterministic plunit tests.

Each answer uses a SWI-Prolog dict named features. Finite forms include person, number, tense, mood, and form.

English supports these slots:

• infinitive;
• six present indicative person-number slots;
• six past indicative person-number slots;
• past participle;
• gerund.

Spanish supports these slots:

• infinitive;
• six present indicative person-number slots;
• six preterite indicative person-number slots;
• six imperfect indicative person-number slots;
• past participle;
• gerund.

Full paradigm means every slot implemented by the selected language rule set.

Setup

Install SWI-Prolog 9.x or later. Run commands from the repository root.

Check the installation:

swipl --version

Run the test suite:

swipl -q -f none -s test/test_morphoscope.pl -g run_tests -t halt

Run the command-line demo:

swipl -q -s bin/morphoscope.pl -- analyze en walked
swipl -q -s bin/morphoscope.pl -- generate es hablar
swipl -q -s bin/morphoscope.pl -- paradigm en go

Use the library from a Prolog query:

?- use_module(morphoscope).
?- analyze(en, went, Lemma, Features).
Lemma = go,
Features = features{form:past, mood:indicative, number:singular,
                    person:first, tense:past}.

?- paradigm(es, hablar, Forms).

Sample output

The first command emits one answer for each compatible reading. Output is shortened here.

lemma=walk features=features{form:past,mood:indicative,number:singular,person:first,tense:past}
...
lemma=walk features=features{form:past_participle,mood:participle,tense:past}

A generation query emits surface forms with their features.

form=hablar features=features{form:infinitive,mood:infinitive,tense:present}
form=hablo features=features{form:present,mood:indicative,number:singular,person:first,tense:present}
form=hablando features=features{form:gerund,mood:participle,tense:present}
form=hablado features=features{form:past_participle,mood:participle,tense:past}

Test status

Tests cover regular and irregular analysis in both languages. Tests cover English and Spanish paradigm generation. CI installs SWI-Prolog and runs the suite on pushes and pull requests. Local execution was unavailable for this release because swipl was absent from the environment.

Limitations

The analyzer covers verbs only. It does not infer unknown lemmas or use a corpus. English spelling rules use curated exceptions and do not model dialect variants. Spanish rules omit subjunctive, future, conditional, pronoun clitics, and regional variants. Lexicons are hand-curated fixtures, not frequency lists. An analysis does not provide confidence scores.

License

Morphoscope is released under the MIT License.