‘Impersonal’ and ‘reflexive’ constructions: Verb features peculiar to Old and Middle English

Summary

Ogura asks why Old and Middle English relied so heavily on two kinds of constructions. The first is the 'impersonal' type, with a dative or accusative experiencer and an optional (h)it. The second is the 'reflexive' type, with a coreferential object pronoun and optional self. She argues that both, along with 'be + past participle', took over functions that the language had lost before it was written down, chiefly the middle voice.

Counts drawn from her earlier work show no steady decline in either construction across Old and Middle English texts. She then traces individual verbs through OED3, DOE, DOEC and MED. Native verbs such as sceamian, þyncan, forþencan, ofþyncan and becuman stay in use. Loans such as plesen, remembren, repenten, anoien and paien enter and take on the same three patterns.

Her conclusion rests on three mechanisms:

• synonyms influence one another's syntax;
• forms that merge in sound or shape also shift syntactically;
• loan verbs join existing rivalries among native synonyms.

The study matters for understanding how English handles experiencer verbs.

Key points

• The definitions rest on alternation. An 'impersonal' (me licaþ, him þinceþ) can turn into a personal construction, while a true impersonal like it rains never does. 'Reflexive' (he him ondred) is separated from true reflexives such as behave yourself.
• Sample counts (impersonal / reflexive):
  • West Saxon Gospels 183/187 and Wycliffite EV 185/187;
  • ÆCHom 276/637;
  • OE Poetry 442/665;
  • GD(C) 583/483;
  • Laʒamon (C) 247/345.
• Examples are taken as closely as possible from OED3, DOE, DOEC and MED, checked against BT. Parallel versions are compared, including Lindisfarne, Rushworth, WSCp and the AV, and the C vs O manuscripts of Laʒamon.
• Approximate borrowing dates given for loans:
  • paien, as early as a1200;
  • anoien, c. 1250;
  • repenten, c. 1300;
  • remembren, c. 1350, displacing geþencan, beþencan and munan.
• Two survival paths stand out. Semen shared the impersonal role of þyncan, which competed with personal þencan, and lives on as seem. Plesen partly took over the impersonal use of lician.

Interesting ideas

• I1. An "impersonal" is defined by whether it can alternate with a personal version, so the test is behavioural rather than formal.
• I2. The paper treats the impersonal, the reflexive and 'be + past participle' as near-equivalent ways to express one experiencer state; "him wearies" and "he is wearied" both amount to "he is weary".
• I3. Synonyms and new loan verbs tend to pick up each other's argument patterns.
• I4. Comparing parallel versions of the same passage (different translations or manuscripts) shows which constructions can stand in for one another.
• I5. Old-style reflexives use a plain object pronoun that refers back to the subject, as in "he bethought him" or "it repents me".

For Mova

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

• I1 (frames): when filling roles, check whether a construction has a personal alternant, so that the it in "it pleases me" is never made an Agent and the oblique pronoun becomes the Experiencer.
• I2 (world): collapse "it annoys me", "I am annoyed" and "annoy oneself" into a single emotional-state record for the experiencer, so later reasoning sees one fact.
• I3 (world): if a verb class lacks frame data for one member, offer frames borrowed from its VerbNet/WordNet synonyms, marked as hypotheses until the text confirms them.
• I4 (learn): treat paraphrase or translation pairs as extra witnesses, and turn the points where their parses disagree into candidate alternation rules.
• I5 (answer): archaic phrasing in fables and tales should trigger reading the pronoun as coreferent with the subject (or as the experiencer), so questions about who regretted or feared get answered correctly.