📄 Editorial: Englishes in a globalized world: Exploring contact effects on other languages

Summary

This editorial introduces a Research Topic of sixteen studies on how English interacts with other languages, both as a source of borrowed material and as a language reshaped by its contact partners. To organize this varied work, the editors use Onysko's Language Contact Typology of Englishes (LCT), which classifies Englishes by the contact situation that produces them rather than by national labels. The LCT has five macro-types: Global Englishes, Englishes in Multilingual Constellations, Learner Englishes, Koiné Englishes, and English-based Pidgins and Creoles. Two refinements matter. A single named variety, such as Indian or Singapore English, can combine several types depending on the speaker or community. A variety can also move between types over time, as Irish English did when it shifted from a bilingual setting to dialect contact. The editors then place each contribution within these types. The studies draw on social media, corpora, elicitation, questionnaires and interviews. The editorial's main value is a single contact-based framework that makes very different case studies comparable.

Key points

• The LCT's five types are GEs, EMCs, LEs, KEs and EPCs. Most of the collected papers fall under Global Englishes, where English elements are absorbed by recipient languages.
• In interviews with young Belgian Dutch-speaking children, about 9.7% of words were English loans, and girls used significantly more of them (Schuring and Zenner).
• In a German-language newspaper in Australia, anglicisms in DPA agency copy were flagged more often than in pieces by local journalists. This may alienate a fully bilingual readership (Hunt).
• In Japanese soramimi (reinterpretations of English lyrics), the processes found were substitution, insertion, deletion and word-boundary crossing. Substitution was by far the most common (Scherling et al.).
• Kirk positions Irish English among world Englishes models using ICE corpora, handbooks, eWAVE and GloWbE. Westphal studies Nigerian English question tags through both corpus data and questionnaires.

Interesting ideas

• I1. Classifying language by the contact setting behind it, and allowing one text or speaker to mix several settings, is more realistic than a single variety label.
• I2. How often a foreign word is visibly marked depends on who wrote the text, so flagging tells us something about the source.
• I3. A small, closed set of edit types (substitute, insert, delete, cross a boundary) was enough to describe how foreign speech gets reinterpreted.
• I4. Pairing corpus counts with attitude questionnaires exposed a gap: people used invariant tags but rated prescriptive variant tags favorably.
• I5. Following one borrowed phrase across its whole life, as with "pimp my ride" through a construction grammar lens, shows how a fixed expression turns into a productive template.

For Mova

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

• I1 (tree): When choosing UD dialect conversion or parser settings, Mova could estimate a mixture of influences per document (learner, multilingual, global) instead of forcing one "English" profile.
• I2 (tokens): The tokenizer and tagger could record flagging cues around non-English items as a feature, and keep it as an explainable attribute rather than discarding it.
• I3 (learn): Self-induced repair rules could be labeled with this same four-way scheme, making error statistics comparable and showing which repair class dominates.
• I4 (learn): Evaluation could keep separate gold sets for attested usage and for normative judgments, so a model is not rewarded for preferring the textbook form when real text differs.
• I5 (frames): Recurring templates with open slots could be stored as constructions that carry their own frame, instead of being parsed word by word each time.