📄 Predicate Semantics in Old English Chronicle and Medical Texts: a Comparative Analysis

Summary

The paper asks whether the genre of a non-fiction text shows up in the meaning of its predicates. It compares two Old English sources: annual entries from the Anglo-Saxon Chronicle and recipes from the medical collection Lacnunga. Each entry or recipe is treated as one supraphrasal unit. The author sorts predicates into semantic types, using the classification edited by O. Seliverstova and G. Zolotova's split between actional and non-actional verbs. The analysis combines componential, contextual and transformational methods with semantic-syntactic role analysis.

Physical action dominates in both texts, but the profiles differ:

• Chronicle: motion, giving (donative) and activity predicates, with animate participants and past indicative verb forms.
• Recipes: more state, change-of-state and causation predicates, with inanimate participants and second-person imperatives.

The author argues that this distribution forms each text's "rhematic dominant," the shared type of new information across its units. That dominant accounts for the narrative character of chronicles and the instructional character of recipes. The result supports predicate-type profiles as a measurable genre signal.

Key points

• The material is the Anglo-Saxon Chronicle and Lacnunga. The Chronicle survives in eight manuscripts labelled A–H and covers events up to the early 12th century. Lacnunga contains 191 recipes, each naming a disease and then prescribing a remedy.
• Chronicle predicates are mostly full verbs in the past indicative, or a copula plus participle I or an adjective. Their actants are mostly animate nouns.
• Lacnunga predicates are mostly second-person imperatives, sometimes a causative verb (lætan, habban) with an infinitive. Their actants are mostly inanimate nouns.
• Recipes that include a charm contain more predicates of speech acts, intersubjective relations and mode of functioning. The verse parts of the Chronicle contain more deverbal nouns, which form semi-predicative constructions.
• The transformational method recovers hidden predicates inside compounds. Examples are sunlight (the light comes from the sun) and the Old English beahgyfa, "ring-giver."

Interesting ideas

• I1. Describing a genre by the distribution of predicate semantic types, rather than by vocabulary, gives a structural fingerprint of a text type.
• I2. The same physical-action verb behaves differently depending on whether its actants are animate or inanimate, and this contrast separates the two genres.
• I3. Mood and tense correlate with genre: imperatives mark prescriptions, while past indicatives mark a sequence of events.
• I4. Transformations can unfold compounds and deverbal nouns into explicit predicates, such as ring-giver becoming "someone gives rings."
• I5. Embedded sub-genres shift the predicate profile, as charms inside recipes add speech-act predicates.

For Mova

How these ideas could serve Mova, our small open language model — its files, stages and tests:

• I1 (frames): Frame output could be summarised as a histogram of VerbNet classes per text, used to detect whether a passage is narrative or instruction and to adjust frame expectations accordingly.
• I2 (world): Animacy of arguments could set a default world model, with agents and transfers for narratives and substances undergoing state changes for procedures.
• I3 (world): Imperative clauses could be modelled as planned steps rather than past events, so that event ordering treats a recipe as a plan and a chronicle as a history.
• I4 (frames): Nominal compounds and agent nouns could be expanded into implicit frames with roles, so that hidden events become explicit and explainable.
• I5 (learn): A shift in predicate-type distribution within a document could serve as a segmentation signal, and as a negative control when evaluating genre-sensitive components on mixed texts.