Remember that funny video on Scottish accent in the elevator with a voice recognition? Eleven, eleveeeeen, elvn — that would be funny, if… it doesn’t happen so often in the real world.
The challenge with English dialects is ubiquitous. Pronunciation, spelling, grammar, vocabulary, local jargon. Is there really general English which you can speak and don’t look weird in Canberra, Victoria, New York or London? Frankly, no.
Britons say chap, Australians say mate, Americans say bud or pal. Canadians will call you buddy, guy or friend.
For people, it’s a big task to remember all these differences, but basically they shouldn’t. During a conversation, most likely, you will be corrected or prompted. How to deal with online communication? Dictionaries, personal assistants or, maybe, text checkers?
Now, there are digital solutions speaking different English dialects that can back you up, at least, for emailing.
Therefore, you won’t ever rack your brain thinking what is better to say analyze or analyse when chatting with an Australian client.
Do all digital proofreading solutions correct English dialect nuances equally well? Spoiler: NO. And we’ll tell you why.
All the variety of English dialects
According to stats, more than 38M people speak Canadian English (CanE), about 30M — Australian (AusE) and New Zealand English (NZE) and more than 5M speak the South African version of the language.
Certainly, these numbers can hardly compare with 332M of American English (AmE) native speakers and 68M of Britons chirping the language of Shakespeare (BrE).
Nevertheless, the population of English dialects is growing and people in Canada and Australia claim that their lingoes deserve the status of separate languages.
How, in general, do English dialects differ from each other?
Well, there are literally thousands of articles on this topic discovering the nuances of American, British, Canadian, Australian, New Zealand, and South African dialects.
By default, the peculiarities of different dialects include spelling, grammar, and vocabulary.
Additionally, dialects may have different slang, idioms, and patterns originating from the historical roots of each language. That’s a burn! But that’s not the hardest part.
Don’t forget about vocabulary lists.
So, Canadians gravitate more towards American English.
However, this dialect is a mix of British English, Western American English, the language of Aboriginal peoples in Canada, and French.
Canadians fancy calling themselves as Canucks, drinking pop, and paying with loonies or toonies.
Australians or Aussies tend to use British English spelling and grammar patterns. But they also have many words derived from the Indigenous Australian languages, Polynesian languages, and even Aboriginal languages.
So, don’t be surprised if you hear dinkum, bludger or Sheila in Sydney or Melbourne.
In both dialects you may come across spelling options of the both.
New Zealand English is based on the language of English-speaking colonists and the vocabulary of the Māori language.
South African English (SAfrE) is actually a set of dialects, which slang is so catching, that lekker has already become widely used in other regions.
WProofreader browser extension
Free for individual users and 14 days-trial for business users.Explore more
Certainly, when a person is physically in a zone speaking a certain English dialect, it’s much easier to grasp and digest all the nuances.
In the online business environment, things are much more complicated.
Today, you may chat with a teammate from Chicago, tomorrow — email a client from Canberra, and the day after — call a partner from Liverpool.
Should you use one spelling by default or do you need to switch between them to be on the same page with your opponents? What if you don’t know all the dialects?
Some companies have a style guide establishing a default English dialect spelling for corporate communication, marketing and content promotion. In many cases, it depends on the target geo zone the majority of clients come from.
On the one hand, using one dialect is convenient for the staff, on the other — may cause misunderstanding with clients. Brits, for example, prefer their dialect over others.
Switching between lingoes is a tough task, and it’s great if you have a helper. Digital proofreading solution supporting English dialects may be a go-to, at least for spelling check.
How well do text correctors speak English dialects?
Companies and businesses serving globally obviously need a multilingual spelling and grammar checker proofreading in different English dialects.
Before selecting a solution, we recommend doing a performance check and see how well a tool can cope with spelling, grammar, and vocabulary issues.
For this research, we looked at LanguageTool, Grammarly, Sapling AI, ProWritingAid (PWA), Linguix, Outwrite and checked how well these solutions recognize the nuances of American, English, Australian, Canadian, New Zealand and South African languages. We added our in-house solution, WProofreader, to the sampling.
First, we checked all solutions on the dataset of spelling, grammar and vocabulary cases for American and British English dialects.
The list of cases was prepared by our in-house linguist and included common rules.
Spelling: suffixes and endings -ise and -ize, -ogue and -og, –or and -our, -re and er, dropped -e, doubling l, context-based spelling, etc.;
Grammar: gotten/got, have/take in collocations, the past participle ending with -ed/-t, collective nouns with are/is, use of adjectives instead of adverbs, shall/should and will/would etc.;
And vocabulary word pairs: autumn–fall, bill–check, biscuit–cookie, film–movie, flat–apartment, holiday–vacation, etc;
We checked whether or not a solution recognizes a lingo’s nuances, gives false positive or negative suggestions.
Second, we analyzed how text checkers cope with vocabulary in other dialects: Australian, Canadian, New Zealand, and South African. Grammarly, Sapling AI, and ProWritingAid do not support these dialects, so we removed these solutions from the sample. We didn’t test proofreaders on spelling and grammar errors, because there are no available public datasets. Besides, for the purpose of the research, that’s enough to see whether a text checker recognizes the vocabulary nuances of extra dialects.
For analyzing the performance of text checkers, we used a confusion matrix by Binary classification where:
- True positive means there’s an issue and a text checker recognizes it.
- True negative — there’s no issue and a text checker doesn’t recognize it.
- False negative — there’s an issue, but a text checker doesn’t recognize it.
- False positive — there’s no issue, but a text checker flags it.
For example, when we say a text checker offers true positives for spelling comparison (AmE–BrE), it means it detects almost all the differences in spelling in these two dialects. False negatives means a text checker fails to see spelling differences in both dialects.
A proofreader gives true negatives in checking vocabulary, e.g. AusE, when it doesn’t detect a dialect’s slang as a typo. And quite the opposite, false positives occur when a proofreader doesn’t recognize a real word and marks it as a mistake.
Checking spelling, grammar, and vocabulary in different English dialects
|Dialects supported||AmE, BrE, CanE AusE, NZE, SAfrE||AmE, BrE, CanE, AusE||AmE, BrE, CanE AusE, NZE, SAfrE||AmE, BrE, CanE AusE||AmE, BrE, CanE, AusE||AmE, BrE, CanE AusE, NZE, SAfrE||AmE, BrE, CanE AusE, NZE, SAfrE|
|Spelling English AmE–BrE||True positives, Oxford spelling||True positives, no Oxford spelling||True positives, no Oxford spelling||False negatives||True positives||True positives||True positives|
|Grammar AmE–BrE||Rare true positives||N/A||N/A||N/A||Rare true positives||Rare true positives||N/A|
|Vocabulary AmE–BrE||True negatives||True negatives||Rare false positives||N/A||False positives||N/A||N/A|
|Idioms English AmE–BrE||N/A||N/A||N/A||N/A||N/A||N/A||N/A|
|AusE vocabulary||False positives||False positives and false negatives||True negatives/False positives||True negatives||True negatives||False positives||True negatives|
|CanE vocabulary||True negatives||False positives||True negatives||True negatives||True negatives||False positives||True negatives|
|NZE vocabulary||False positives||N/A||Rare false positives||N/A||N/A||False positives||True negatives|
|SAfrE vocabulary||False positives||N/A||False positives||N/A||N/A||False positives||True negatives|
From the research, we see that:
- The majority of grammar checkers recognize the nuances between American and British English spelling. Only some like LanguageTool, ProWritingAid, Linguix detect grammatical errors (verb endings, have/take in collocations). Vocabulary differences are also important, however, only some solutions like Grammarly and LanguageTool have dictionaries which include them.
- Other English dialects are less developed compared to American and British lingoes. This is a great success if a text checker proofreads vocabulary equally well in Canadian, Australian, New Zealand, and South African languages. Some words are detected as unknown, others — as misspellings.
- Among all the solutions, Outwrite seems to cope the best with extra English dialects. It recognizes almost all vocabulary nuances. However, Outwrite doesn’t detect grammar and vocabulary issues in American and British English dialects.
- WProofreader supports American English, British English, Australian, New Zealand, South African English spelling. Importantly, it doesn’t offer Oxford spelling as LanguageTool. As for extra dialects, WProofreader sees spelling differences, recognizes unique vocabulary cases in Australian, Canadian, and New Zealand dialects. We keep improving the proofreading mechanism to give fewer false positives.
You can use our survey results for preliminary testing. To take more informed decisions, we suggest going deeper and comparing solutions on large datasets of texts written in different English dialects.
WProofreader — multilingual grammar checker and spelling corrector
If you haven’t had any opportunity to test WProofreader before, let us briefly describe what the solution offers and how it works.
So, WProofreader is an in-browser proofreading extension supporting many English dialects, including American, British, Canadian, Australian, New Zealand and South African.
By the way, WProofreader has AI engines for English, German, and Spanish. AI models for American, British, Canadian and Australian English dialects are coming. Stay tuned!
Additionally, WProofreader offers many helpful features:
- Language auto detect to proofread combo texts
- Spelling autocorrect and autocomplete suggestions
- Company-wide and user dictionaries
- On-premise deployment
- Personalization options
- Integration with online business tools and software
For those companies and developers who need predefined modules to integrate proofreading functionality to their custom systems, we offer WProofreader SDK.
Let’s sum up
- There’s no ultimate text corrector with a perfect proofreading performance for all the English dialects. So, before purchasing the solution, you need to run a basic comparison сheck. It’s recommended to proofread differences for all the types of mistakes, including but not limited to grammar, spelling, vocabulary. The quality of suggestions should also be taken into account.
- The more dialects are integrated, the more challenges the provider may experience. Integrating new dialects requires sustainable data to meet the requirements of check. The AI models are a way out. However, they demand either a lot of data to be trained or custom algorithms to filter and correct the behavior of AI with dialects.
- WProofreader offers a good spelling check for American and British dialects and recognizes some vocabulary nuances for extra dialects. We keep improving the quality of the checks to take it to the next level.
Any comments and ideas on this topic are welcome. Don’t forget to subscribe to WProofreader Blog not to miss a post.