📄 Compound Nouns in the Old English Period: Ælfric's Lives of Saints. Functional and Pragmatic Approaches

Summary

The paper asks why Old English formed compound nouns so readily and what these compounds do in communication. It joins two frameworks. From functional grammar (Dik, Halliday, Downing & Locke, Quirk et al.) it treats a nominal expression as a tool for picking out a referent, and a modifier as a classifier that narrows the head noun. From Sperber & Wilson's relevance theory it takes the idea that speakers seek strong effects at low processing cost. Compounds pack information by dropping elements such as prepositions. The author argues that this fits an inflection-heavy language with relatively few prepositions.

The empirical base is two of Ælfric's Lives of Saints, St. Oswald and St. Edmund. Nominal formations from both texts are collected and sorted by structural type in an appendix. The main claims are:

• Compounding is a matter of degree rather than a yes/no category.
• A compound's meaning is not simply the sum of its parts.
• Compounding in Old English also applies to loanwords.

The work matters because it gives a communicative explanation for word formation in early English.

Key points

• The corpus is two texts from Ælfric's Lives of Saints: St. Oswald and St. Edmund, using the Skeat and Needham editions.
• The appendix groups the instances into these types:
  • noun/adjective + suffix (e.g. -nysse, -dom)
  • two nouns joined together
  • two nouns standing side by side in apposition
  • an adjective before a noun, optionally preceded by a determiner
  • genitive + noun
  • prefix + noun (only in Oswald)
• Following Downing & Locke, compounding is a cline of semantic cohesiveness between classifier and head. There is no sharp line between compound and non-compound.
• Dik's example pregnant women is used to argue against a purely additive reading of compound meaning.
• The paper is by Mª Ángeles Ruiz Moneva (University of Zaragoza), published in SELIM 8. It grew out of a project on compound nouns in functional grammar directed by Javier Martín Arista.

Interesting ideas

• I1. Treat "compoundness" as a graded score of how tightly modifier and head cohere, rather than a binary label; this avoids forcing borderline cases into one bucket.
• I2. A modifier in a nominal group acts as a classifier that narrows the head to a subclass, so many noun+noun compounds encode an is-a relation.
• I3. Compound meaning is constrained by world knowledge rather than computed by intersecting its parts; the pregnant-women case shows naive intersection can be redundant or wrong.
• I4. Compounds arise by deleting the prepositions that would otherwise state the relation, so each compound has a hidden prepositional paraphrase.
• I5. Relevance theory frames compounding as a trade-off: less material to process, but heavy premodification raises the effort again.

For Mova

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

• I1 (tree): The UD conversion step could keep a cohesion score on compound vs amod/nmod choices, so disagreements between witness parsers on such cases are explained rather than counted as plain errors.
• I2 (world): Noun-noun compounds in text could propose candidate subclass links (e.g. a kind of -ceaster, a kind of -man) to be checked against the WordNet/Wikidata ontology.
• I3 (frames): When building frames for compounds, the relation between modifier and head should be chosen using the knowledge base's constraints, not by stacking predicates as in a bag-of-features approach.
• I4 (back): The generator could expand a compound into its prepositional paraphrase, or contract it back, giving an explainable round trip in the text → graph → text compressor.
• I5 (back): During generation, a simple cost rule could limit premodifier stacking and switch to an of/prepositional phrase once a nominal gets too dense.