📄 On Problem of “Relict” Phenomena in American English Grammar

Summary

This review asks whether several American English grammatical features really are "relics" preserved from older British usage. That view goes back to Marckwardt's 1958 idea of "colonial lag" and to Shveitser's work on English in the USA. Goncharova pulls together historical and corpus findings, mainly from Hundt, Algeo, the Longman grammar (Biber et al.), Levin and Kjellmer. She examines four features:

• the participles gotten and proven;
• regular past forms of verbs like burn, lean, spell;
• the synthetic (mandative) subjunctive;
• singular verb agreement with collective nouns.

The evidence shows that most of these forms were uncommon in 18th–19th-century America too. They spread there later, so Hundt's label "colonial revival" fits them better than "lag." Collective-noun agreement looks like parallel development in both varieties. The review also observes that British English is now adopting several of these features. This matters because it replaces an inherited assumption with dated usage evidence. It also shows that the two varieties influence each other in both directions.

Key points

• Gotten was very rare in 18th–19th-century American usage. Today it marks dynamic meaning: acquiring something, or getting permission. Got covers static possession and obligation. Have got to is about a third more frequent in BrE than in AmE.
• Proven came from Scots in the 16th century. In AmE it has been attested since the 19th century as a low-frequency rival of proved. It now has no stylistic restrictions and is gaining ground in BrE.
• AmE has led the regularization of burn/lean/spell-type verbs since the later 19th century. Yet 18th-century American fiction used regular forms less often than English fiction did.
• Biber's register data show a split by register. Learned dominates fiction, news and academic prose. Conversation prefers the -t form (over 75% as participle). The -t forms also persist in adjectival collocations such as burnt toast and burnt sienna.
• The mandative subjunctive (I insist that this book be removed) was rare in both varieties in the 18th–19th centuries. It is now the AmE norm and is common in BrE, especially in the passive. Kjellmer links its survival to proverbs and to fixed phrases from the King James Bible.

Interesting ideas

• I1. Irregular and regular variants such as dreamed/dreamt can sit side by side in one passage, and Taylor found such doublets lasting four or more centuries, so a system should treat both as equally normal forms of a single lemma.
• I2. Have got and have gotten differ in meaning: the first describes a state you are in, the second an event that changed something.
• I3. Verbs like insist and suggest can be mandative (followed by a bare-form subjunctive) or simply reporting, and the subjunctive also appears after nouns like recommendation.
• I4. The register-based percentages show that which variant is preferred depends on register and dialect, not on how frequent the verb is overall.
• I5. The review argues for testing an inherited claim ("colonial lag") against dated corpus counts before accepting it.

For Mova

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

• I1 (tokens): Lemma tables should map both spellings to one lemma and record the variant as a feature rather than flag it as an error.
• I2 (world): The world builder can treat 've got as a current holding and 've gotten as an acquisition event, which updates holdings at a specific point in the order of events.
• I3 (frames): A bare be in a that-clause after these words can signal a directive speech act rather than a report, helping prag separate demands from assertions.
• I4 (back): Generation can choose learnt or learned, -t or -ed, according to a chosen register and dialect profile, keeping a whole text consistent.
• I5 (learn): The same habit can apply when checking rules Mova induces for itself, by measuring them against period- or register-specific slices instead of trusting the rule's original rationale.