Breadcrumb

  • Home
  • Imaginary Friends 4 - An Uncommon Language





Amazon Prime Trial

Search form

Main navigation

  • Home

HOME | ALL TECHNOLOGY

Imaginary Friends 4 - An Uncommon Language

Tuesday, 1 April, 2025 - 06:00

″You can’t have standardised understanding if you don’t have standardised language.″

The Four Imaginary Friends, A, B, C and D continue to develop fundamental questions of whether we can expect machines to relate to the real world when the real world communicates in human language. 

Or

Two computers divided by a common language: it matters, even if some people think they could[n't] care less. 

A. Last time we looked at a specific example of how one letter in one word of one line in one stanza of a song can determine the whole sense of the song. I want to use that as a baseline for further discussion. Let’s try to stay with the written word for the time being. 

C. I think that that example was one of the most important aspects of this subject. It’s more than just whether a letter is a capital or lower case: it’s also where typos and even variations in spelling make a difference. 

D. I’ll start then but to do so I have to go back to one of the quotations we had previously not to requote it but as a starting point. If everything is based on language, then either machines have to either work in a common and standardised language. That’s not a computer language, it’s actual words and phrases, vocabulary.

A. Well, that idea falls at the first hurdle, doesn’t it? It’s strange that we’ve previously mentioned George Bernard Shaw as a philosopher without mentioning the thing that, in this discussion, becomes utterly central: the statement often (but perhaps not accurately) attributed to him that ″Britain and the USA are two countries divided by a common language." In fact, it’s amusing in this context that the statement is often ″quoted″ in more than a few different formats. For example, the New York Times has it as ″England and America are two countries separated by the same language." Why, one has to ask, does the New York Times feel it necessary to use ″the same″ instead of ″common.″?

A. Let’s look at English, for that is what Turing specifically spoke about. And let’s look at English in England. Let’s look at two phrases that are exactly opposite but depending on which part of English they are used have the same meaning. ″It’s up to you″ in the north becomes ″it’s down to you″ in London. They both mean ″it’s your choice″ or similar. However, the London expression has an alternative meaning: it also means ″it’s your responsibility.″

C. So that’s like ″I couldn’t care less″ in English v ″I could care less″ in American.

A. No, it’s entirely different. In that example, the American version is complete nonsense. It specifically says that there remains capacity for care which is exactly not what the American is trying to communicate.

D. Let’s break that down another level: the use of the word ″care.″ How many different uses can we think of for that?

A. Obviously ″to take care of″. But what does that mean? If someone says ″can you take care of him″ does he mean to be kind and, er, careful or does he mean ″dispose of him″? Is it to pat him on the head or fit him for concrete boots? 

C. The ″couldn’t care less″ means that the consequences are absolutely immaterial so ″care″ here means something different to both of A’s ″cares″.

D What about homonphones – isn’t this where the spoken word and the written word collide? You can’t tell from a person’s voice whether the word is ″licence″ or ″license″. In English, ″licence″ is a noun and ″license″ is a verb. For example, ″A was licensed to issue licences.″ There has long been the use of automation to identify potentially fraudulent insurance claims from the voice of the person making the claim. They claim success. So if two entirely different words sound the same, the programming of apparent comprehension is more complex.

A. Bow and bough. Threw and through. Their, their, they’re. Our readers will see the difference but as I speak, they are identical. And that’s before we start to look at homonyms such as pen, band, spring. Or how about words that we use in ways that are incorrect but we all know, or think we know, what the speaker means such as ″average″ which means neither good or bad but we take it to mean sub-par. 

B. The systems that are designed to identify fraudulent claims from telephone calls can say they have success because they literally don’t see their failures. For example, they might say ″We identified 100 possible frauds and on investigation about 80% were in fact fraudulent claims.″ So they boast a 20% false positive rate. But the data is selective because they have no idea how many fraudulent cases i.e. false negatives, they did not identify. In fact, they cannot know unless they individually review every claim that the system said was not fraudulent to see if the software got it right.

D. True and we should return to that example. But for now, let’s look at the words ″practice″ and ″practise.″ They represent different parts of speech. We don’t even need to go into the complicated grammar here: ″practice″ is a noun, ″practise″ is a verb. ″Defence″ is a noun, ″defense″ is a verb. We can see that in writing but we can’t tell when we hear the word unless we can analyse the context.

A. We can’t tell it in writing, either, if we are reading American.

C. Let’s stick with spoken English English. I once heard of a chap in prison who wrote to his solicitor ″you fink Im thuckin’ fick.″ He wrote exactly as he speaks. The strange thing is that somewhere inside his mind, he knew that he should replace the f in ″think″ with a th so he extended that and replace the F in, well, F-ing. But didn’t replace it where TH was in fact needed.

D. I bet he didn’t pronounce Gs on the end of words.

A. Haha. And then there’s Estuary English as it was called when all the Sloan Rangers decided to sound like they came from Essex and choked on their Ts.

B. That got so bad, it was like they needed the Heimlich manœuvre to release all the Ts they collected in their throats. So even within a small part of England, words that had been clear became unclear. 


Dialect v accent

C. Haway, Hinny.

B. What?

C. It’s English. It’s Geordie. The Tyneside dialect. There’s a song, ″On Ilkley Moor bah tat″ in North Yorkshire dialect. In that part of Derbyshire next to Yorkshire, the use of ″thee″ and ″thou″ was common not long ago and still is amongst the elderly. Dialects are different to accents. A dialect can almost be seen as a different language. For example, in the North East of England, in just forty miles, there were four absolutely discrete dialects. Within a generation starting in the 1960s, they almost disappeared but accents remained. 

B. Do we think that American is an English dialect?

A. I think there are probably many views as to this and part of that is because English speakers that went or were sent to The American Colonies at a time when there was very little homogenisation of dialects in England – and of course many went from Ireland where, surprise surprise, there are many dialects – and a completely distinct language. The difficulty we now face is that American has become homogenised but it is not homogenised with English. I’ve a friend who says ″just because you are using English words doesn’t mean you are speaking English.″ 

C. I have another question, and I’m still talking about the spoken word. We’ve seen, over the past couple of decades, a massive change in the way that emphasis and intonation are used. Sometimes it happens with individual words for example ″harassment″ where the correct emphasis is on the first syllable but it is often placed on the second. So machines have to be trained to recognise both. Here’s the interesting thing: that mispronunciation is, to a degree, regional and generational. So it can be used to cross-refer to other information already held about a person. But it’s not reliable: people move around, they mix with others, they pick up each others’ mannerisms. And some of them stick: how many English people in their sixties use ″cool″? That arrived in England from the USA in the late 1950s via popular culture i.e. broadcasting. It was reinforced through cinema and the surf culture that crossed from California. Interestingly, it fell out of favour until, in the 1990s, a twisted version of it appeared and young people, at first as a joke, started saying ″kewl″ and yes, it even got a spelling. That seemed to reactivate ″cool″ and those that had used it before now found it embedded in their everyday language and it hasn’t gone away. It has replaced several English expressions. But it’s not ″cool″ in the sense that ″cool″ relates to temperature and its linguistic sibling ″neat″ doesn’t mean that something is orderly, as it would in the usual course of English.

D. Are we saying that we can’t assume that any machine can learn a language? That any claim that an algorithm that works on language is bound to fail? Are we saying that all machine learning that depends on language is falsely represented?

A. I don’t know what you three think but I’m certainly not saying that. But I am saying that any representation that such tools must be comprehensively programmed to analyse and produce reports based on both spoken and written English (if we are using English as the common language) and that it must have a databank of all variants of English, their use, their inconsistencies and, of course, their dialects and pronunciation variances. It must be capable of identifying the context and, of course, if it is to be used to identify intent, it must be able to identify what the speaker or correspondent meant. That’s where it all breaks down because in the current state of affairs, programming is not sufficiently comprehensive and nor are the analytical tools. If it’s programmed in Palo Alto, it’s all too often programmed in Palo Altese and, worse, it’s assumed that all language is current. And it’s not. People don’t all follow California’s lead and Hollywoodisms are just that, artificial expressions that script writers hope will become catchphrases. 

D. Examples?

A. Several US medical dramas have started to use the word ″Emergent″ when they mean ″Emergency.″ As in ″he needs emergent surgery.″ In one series it was annoying; now it’s in more and spreading. Will it spread across the world? Maybe. Programmers need to be aware of it. They can’t form the logical view that it’s stupid and they should ignore it because people will start to use it as they have with e.g. Down’s Syndrome / Down Syndrome and Tourette’s Syndrome / Tourette Syndrome to say nothing of doll house, butcher block and so on. It’s plainly grammatically incorrect and is actually nonsense but with use it has found its way into the language of the otherwise educated: in the UK, recent legislation missed out the apostrophe S in Down’s Syndrome. So something which is destructive to sense becomes some form of sense over time, even if most of the world rejects it.

B. I agree. We are seeing a ridiculous trend to make plurals out of words that are singular such as ″damage.″ When something is broken, there is ″damage.″ The word ″damages″ has a very specific meaning: it’s the award of money ordered by a court to be paid by one party to another in respect of a harm. To use it about a kicked-in door or a bent panel on a car is a nonsense. These mistakes make it difficult to set things in context.

C. I’m hungry. My dinner’s roast chicken. What are you going to have tonight?

D. I’m having faggots.

B. See, context is everything.


Note: In English, a ″faggot″ is a bundle of sticks or, as in this case, a meatball in gravy. Other interpretations are available but, in this case, wrong.


An edited version of this article was published in the-yuan.com in August 2022
 


Imaginary Friends 1 - Yes, No, Maybe 

Imaginary Friends 2 - Of Cogs and Golfballs 

Imaginary Friends 3: ″Capital Punishment? 

Imaginary Friends 4 - An Uncommon Language 

Imaginary Friends 5 - ″The first thing we do, let's kill all the lawyers" 

Imaginary Friends: 6. The circles of an imaginary mind…...

About this section

From FinTech to RegTech, from "AI" to security, from the terraverse to the metaverse, if there's a technology element, we're interested. But this is not an area for PR. It's an area for considered, structured articles that advance arguments. Think Op-Ed with a purpose.

Opinion pieces or "Op-Eds" are the home-made bombs of the publishing world. So long as they meet editorial standards, are not intentionally offensive with a view to causing hurt or insult and are relevant to our field of endeavour, we will look at submissions.

We like contentious, we like contrarian views. We don't like pretty much any -ism . We recognise that Opinion pieces are one person's view and are not balanced (if they are balanced and reach a reasoned conclusion, they are probably more suited to the Articles section). We do not like empty expressions (reaching out, going forward, circling back etc), acronyms and buzzwords. English, only please.

To submit an Opinion / Technology piece, please complete the Contact form.

Footer menu

  • Weekly Digest (opens in new tab)
  • Images attribution (opens in new tab)
  • Corporate, privacy, intellectual property and access (opens in new tab)
  • Advertising and Recommendations (opens in new tab)
  • Promote your business (opens in new tab)
  • Enquiries (opens in new tab)


 

BOT AND SCRAPER ACCESS DENIED

 


 

Built with Drupal     |     Hosted by Siteground     |     Template by Alaa Haddad     

Design by Vortex Centrum Limited    |     Some services provided by Google Workspace    

Posters and other merch by ProjectLXX   |   Privacy and security services by Surfshark and Firetrust. 


Nothing in this website is intended to be or shall be taken as legal advice. 

You should always seek advice from a practitioner experienced in this area. 


Everything on this website is copyright Nigel Morris-Cotterill and/or Vortex Centrum Limited for itself or one of its business units. No downloading, printing or other means of replicating or reusing is permitted. In particular, all bot access is denied and all scraping of content will result in the legal action set forth in the terms and conditions in this site. For legal, cookies and privacy see vortexcentrum.com.

Copyright 1999- © 2026 Vortex Centrum Limited - All rights reserved. Bot access denied.