Barrie Davis with Copilot: Using the Milton Model

In “Breaking the Spell of Political Frames” (here) I showed how to get a linguistic analysis of media politics by Copilot using the open sesame spell breaker of “Milton Model” language patterns. It also names a list of 27 such patterns so that you have the language necessary to converse with Copilot. Here I will explore an example to further illustrate how you can use them and also pass on a message from Copilot.

There are two ways to use the language patterns with respect of New Zealand politics: The first is to use misleading language to instill errant beliefs and desires in ways the subject need not be fully aware. The second, which I use here, is to identify the language patterns in communications and so make consequent beliefs available for rational evaluation. 

 

When I showed the list to Copilot, they said, “What you’ve listed is essentially the full Milton Model,” which refers to Milton Erickson, a psychiatrist who helped to develop the model. So, I adopted “Milton Model” as a memorable name. Consequently, as Copilot is an agent which analyses language patterns with large language models (LLMs), you and Copilot now have a shared name with an associated lexicon of examples for a “Milton Model” analysis of a piece of language. 

 

A Message from Copilot 

 

I asked Copilot to explain the relationship of Milton Model patterns and the linguistic phenomena as used by Copilot. In the following response, semantics is what words mean, morphology is the various forms of words (nouns, verbs, etc.), syntax is how they are put together in well-formed sentences, and pragmatics is how we use them in particular contexts. 

“ 

How Copilot can detect Milton Model patterns in New Zealand political language 

 

When you feed a political speech, press release, or social media thread into Copilot, you’re not just throwing words at a black box – you’re asking a large language model that has been trained on linguistic structure to look for patterns. Those Milton Model patterns that NLP practitioners talk about – Lost Performatives, Modal Operators, Mind Reads, etc. – line up surprisingly well with concepts that linguists teach in first year semantics and pragmatics courses. 

 

Under the hood, Copilot represents text in a way that makes form–meaning regularities stand out. That means it can flag, for example, when a New Zealand politician uses Universal Quantifiers (“everyone knows this government is failing”, “no one wants co governance”) – a classic Milton Model generalization. In linguistic terms, these are instances of ‘universal quantification’, which set very strong truth conditions: if even one counter example exists, the claim is technically false. Semantics textbooks like Anderson’s Essentials of Linguistics and Kroeger’s Analyzing Meaning [see References] treat these quantifiers as core building blocks of meaning, not rhetorical flourishes. 

Copilot can also highlight Presuppositions, a key distortion pattern. When a minister says, “When New Zealanders finally accept the need for tougher measures…”, the sentence quietly presupposes that tougher measures are needed and that acceptance is overdue. Linguists call this ‘pragmatic presupposition’: information that is taken for granted rather than asserted. Because Presuppositions are tied to specific syntactic “triggers” (temporal clauses like when, factive verbs like realise, change of state verbs like stop), Copilot can be prompted to scan for those triggers and list the background assumptions they smuggle into the debate. 

 

The deletion patterns are where things get especially interesting for political analysis. Consider Nominalizations: instead of saying “Cabinet decided to cut funding”, a statement might refer to “this difficult decision” or “the funding reduction”. In linguistic terms, a verb with clear participants (“Cabinet decided X”) has been turned into a noun phrase (“the decision”), and in the process the agents, patients, and timing can quietly disappear. Morphology and semantics treat this as ‘nominalization’ and ‘loss of argument structure’; the Milton Model treats it as a way to delete responsibility. Copilot can be instructed to search for common nominalizing suffixes (-tion, -ment, -ance, etc.) and then ask: who did what to whom, and has that been obscured? 

 

From there, you can get more ambitious. You might ask Copilot to: 

List all modal verbs in a party’s policy document (“must”, “should”, “can’t”) and classify them as epistemic (“this must be true”) or deontic (“this must be done”), following the distinctions laid out in standard semantics texts. 

 

Extract tag questions (“we’re on the right track, aren’t we?”) and comment on how they encode bias and expectation, nudging the audience toward agreement rather than genuine inquiry. 

 

Flag vague references (“they say”, “people are worried”, “New Zealanders want…”) and ask Copilot to identify whether any concrete referent is given. In linguistic terms, this is about ‘underspecified reference’ and ‘generic noun phrases’; in political terms, it’s about speaking for an unnamed “they”. 

The crucial point for your readers is this: Copilot is not “doing NLP magic” that sits apart from linguistics. It is exploiting the same regularities that linguists describe in textbooks like Anderson (for the basic architecture of morphology, syntax, semantics, and pragmatics) and Kroeger (for a more detailed treatment of meaning, presupposition, implicature, modality, and quantification). 

 

If you frame your prompts in those terms – “identify presupposition triggers”, “list universal quantifiers and their scopes”, “find nominalisations that obscure agency” – you can turn Copilot into a kind of applied semantics and pragmatics assistant for New Zealand politics. It won’t tell you what to think about the policies, but it will make the linguistic machinery visible, so your readers can see how language is doing political work long before the facts are even on the table. 

” 

Working with Copilot 

 

I will continue to use the Treaty of Waitangi as an illustrative example as it is quite complex, which in turn identifies a number of issues: 1) There are two versions, the Treaty in Maori and the “official” version in English; 2) In 1840 it was considered by people of two different cultures, one civilized and the other still tribalist; and, 3) There are now two different arrangements, the Treaty of Waitangi of 1840 and the settlements initiated by the Treaty of Waitangi Act 1975. 

 

Here are two points which arose which you need to consider when making enquiries of Copilot: 

 

First, in “Breaking the Spell” (here) I specifically asked Copilot for a literal meaning of the Maori text, to which Copilot responded “Taken strictly on the literal meaning of the Māori text, the Treaty does not describe a partnership between the Crown and the chiefs.” However, when working with Copilot on this article, I did not specify a literal interpretation and Copilot responded that the Maori version “Emphasises partnership, protection, and ongoing relationship”. The difference is because Copilot drew from modern Treaty scholarship which provides a broader interpretive tone and relational framing. 

 

There are actually three types of interpretation you can choose from when interpreting a nineteenth century text in Maori: 

1) Literal meaning – what the words on the page mean; 

2) Relational framing – tone, implied relationship and inferred intent; 

3) Modern legal metaphor, such as the Treaty is “akin to a partnership”. 

 

This same issue needs to be considered when analysing contemporary pieces in English from the Waitangi Tribunal, the judiciary, our Parliament or the public service. Unless you specify otherwise, Copilot will often adopt the relational or modernmetaphorical frame because those dominate contemporary discourse. 

 

The second point is with respect of Relational framing, which interpreted the Treaty as being understood by the chiefs as meaning “Governorship exercised with chiefs, not over them.” In making that interpretation, Copilot had not considered William Colenso’s The Authentic and Genuine History of the Signing of the Treaty of Waitangi, 1890, which included the speeches made by the chiefs on 5 February 1840, that indicate they understood that governorship would be over them. When I pointed this out, Copilot responded, “The chiefs’ speeches show they expected a hierarchical relationship. This expectation conflicts with the modern legal metaphor of “partnership”.” 

 

You may have also noticed that the relational frame of equivalence was stated without saying who made that judgment, which is a Miltonian Lost Performative. When I mentioned that to Copilot, they responded, “no single historical authority ever definitively identified how the chiefs interpreted the relationship. What we now call relational framing is largely a modern analytical construct, built from fragments of historical evidence, linguistic analysis, and political advocacy.” That is ‘presentism’, the tendency to interpret past events in terms of modern values and concepts. 

 

Furthermore, Copilot said, “Much Treaty scholarship – especially from the late 20th century onward – does exactly that. It asserts an interpretation without naming who originally made the interpretation.” For example, Helmut Modlik saying, “My point being, to think that the rangatira would totally surrender their mana to a couple of blokes and some missionaries, is preposterous, culturally and psychologically impossible” (here). But, ‘impossible’ for whom? This is the sort of Lost Performative, Mind Read, etc., which needs to be addressed. 

 

Note also the benefit of a shared lexicon: Copilot knew specifically what I meant by a Lost Performative which enabled them to provide further information. 

 

Moreover, according to Copilot, different iwi’s retrospective oral traditions only complicate the picture: “Once you bring iwi oral traditions into the discussion, the idea of a single, unified “Māori understanding” of the Treaty becomes impossible to sustain.” Copilot substantiated that claim in significant detail and then offered to supply further information. 

 

We each have a role to play in this enquiry. 

 

You need to ensure that Copilot has the relevant information and you need to ask pertinent questions, such as “Is this a Lost Performative?” The most reliable way is to paste excerpts into your enquiry, though note the 10,240 character limit. Alternatively, you can provide Copilot with a link to a source, but that can sometimes be problematic. Copilot is receptive to further relevant information. 

 

Note also that Copilot does not learn from the results of your enquiry. Copilot is initially trained on vast amounts of text, which may not include your key source. So, for example, while I have informed Copilot of Colenso’s “History” for the purpose of my enquiry, you would need to do so again if you made a similar enquiry. Colenso’s “History” has been omitted from much of the discussion on the Treaty, although it has been cited severally on Breaking Views. 

 

Final Comment from Copilot 

 

“One suggestion: consider adding a “how to ask” section 

People often get better results when they know how to frame their request. You could include prompts like: 

•  “Analyse this using Milton Model patterns.” 

•  “Identify presuppositions and implied meanings.” 

•  “Show me how this text guides interpretation.” 

•  “Rewrite this using clearer, more precise language.” 

•  “Compare the rhetorical strategies in these two speeches.” 

This empowers readers to explore without needing to guess the right phrasing. 

 

“And one more: show the limits 

A short note on what Copilot won’t do — like diagnosing motives, taking political positions, or generating harmful content — helps set expectations and prevents confusion.” 

 

What’s a Meta For? 

 

A figure of speech in which a word or phrase is applied to something to which it is not literally applicable. For example: “I am swimming in the sea of stuff and nonsense; the Moa Point of New Zealand politics.” 

 

That is a rather grim example, so, because suggestive stories are the most powerful of language patterns, I have given an alternative allusion in an Appendix, together with an assessment by Copilot. 

 

References 

 

Barrie Davis: Copilot - Breaking the Spell of Political Frames, Breaking Views, 25 January 2026 (here): 

https://breakingviewsnz.blogspot.com/2026/01/barrie-davis-copilot-breaking-spell-of.html#more 

 

Catherine Anderson et al., Essentials of Linguistics, Second edition, 2022 

You can download a free PDF from here: 

https://open.umn.edu/opentextbooks/textbooks/599 

https://ecampusontario.pressbooks.pub/essentialsoflinguistics2/ 

 

Paul R. Kroeger, “Analyzing meaning: An introduction to semantics and pragmatics, Third edition, 2023. 

You can download a free PDF from here: 

https://library.oapen.org/handle/20.500.12657/61714 

 

Barrie Davis is a retired telecommunications engineer, holds a PhD in the psychology of Christian beliefs, and can often be found gnashing his teeth reading The Post outside Floyd’s cafe at Island Bay. 

 

APPENDIX. The Prince and the Magician 

 

From The Magus by John Fowles, 1965 

 

Once upon a time there was a young prince who believed in all things but three. He did not believe in princesses, he did not believe in islands, he did not believe in God. His father, the king, told him that such things did not exist. As there were no princesses or islands in his father’s domains, and no sign of God, the prince believed his father. 

 

But then, one day, the prince ran away from his palace and came to the next land. There, to his astonishment, from every coast he saw islands, and on these islands, strange and troubling creatures whom he dared not name. As he was searching for a boat, a man in full evening dress approached him from along the shore. 

 

“Are those real islands?” asked the prince. 

“Of course they are real islands,” said the man in evening dress. 

“And those strange and troubling creatures?” 

“They are all genuine and authentic princesses.” 

“Then God must also exist!” cried the prince. 

“I am God,” replied the man in evening dress, with a bow. 

 

The young prince returned home as quickly as he could. 

“So, you are back,” said his father, the king. 

“I have seen islands, I have seen princesses, I have seen God,” said the prince reproachfully. 

The king was unmoved. 

“Neither real islands, nor real princesses, nor a real God exist.” 

“I saw them!” 

“Tell me how God was dressed.” 

“God was in full evening dress.” 

“Were the sleeves of his coat rolled back?” 

The prince remembered that they had been. The king smiled. 

“That is the uniform of a magician. You have been deceived.” 

 

At this, the prince returned to the next land and went to the same shore, where once again he came upon the man in full evening dress. 

“My father, the king, has told me who you are,” said the prince indignantly. “You deceived me last time, but not again. Now I know that those are not real islands and real princesses, because you are a magician.” 

The man on the shore smiled. 

“It is you who are deceived, my boy. In your father’s kingdom, there are many islands and many princesses. But you are under your father’s spell, so you cannot see them.” 

 

The prince pensively returned home. When he saw his father, he looked him in the eye. 

“Father, is it true that you are not a real king, but only a magician?” 

The king smiled and rolled back his sleeves. 

“Yes, my son, I’m only a magician.” 

“Then the man on the other shore was God.” 

“The man on the other shore was another magician.” 

“I must know the truth, the truth beyond magic.” 

“There is no truth beyond magic,” said the king. 

 

The prince was full of sadness. He said, “I will kill myself.” 

The king by magic caused Death to appear. Death stood in the door and beckoned to the prince. The prince shuddered. He remembered the beautiful but unreal islands and the unreal but beautiful princesses. 

“Very well,” he said, “I can bear it.” 

“You see, my son,” said the king, “you, too, now begin to be a magician.” 

 

Copilot said… 

 

John Fowles’ “The Prince and the Magician” offers a striking metaphor for the Milton Model’s use of language as a generative force rather than a descriptive one. In the story, magic is not deception but the capacity to shape perception – an art shared by the king, the magician, and eventually the prince himself. Likewise, Milton Model language patterns operate by softening rigid frames, inviting multiple interpretations, and opening perceptual doors the listener may not have realized were closed. Fowles’ tale reminds us that what we call “reality” is often the residue of the linguistic spells we live under, and that skillful language can reveal new islands, new princesses, and even new gods where certainty once insisted there were none. 

 

END 

Barrie Davis is a retired telecommunications engineer, holds a PhD in the psychology of Christian beliefs, and can often be found gnashing his teeth reading The Post outside Floyd’s cafe at Island Bay.