“I keep six honest serving-men
(They taught me all I knew);
Their names are What and Why and When
And How and Where and Who.
Rudyard Kipling 1865 – 1936,
Just So Stories (1902) “The Elephant’s Child
I want to thank all those people who have responded to my last article, especially Michael Hall and David Smallwood for sending me related material. I also appreciated comments made by Jo Hogg on the difference between techniques and modelling in training. I changed my own NLP training a number of years ago to create a modelling approach to training NLP. It is initially more challenging but the increase in effectiveness rewards the effort. Thanks again to Martin Roberts for his continuing questioning of modelling and NLP.
In my first article I briefly outlined my pursuit of the modelling methodology in NLP. My ‘stopgap’ approach that later became DBM has been created over a period of almost twenty years. In this and subsequent parts I will outline my re-modelling of NLP. I will try to present it with a sense of development. I do not work in a linear
fashion, preferring to have a number of inter-related projects at any one time. This facilitates the holistic nature of my work and the fertile cross-fertilisation of different areas and models being applied.
In this article I will outline some of my developments in modelling language. This is an area in which I have worked longest and probably found the most rewarding.
Language has been the main tool in creating DBM, which is similar to the central role which it has had in NLP.
For a number of years I was content to use the Meta Model (M. M.) questions. I was working therapeutically with families and adolescents and the M. M. proved to be very useful. They were by no means the only things I used. The stimulus for my re- modelling was initiated through my attempts to train others working in social work and child care. Teaching the M. M. highlighted a number of concerns for me. I have continued to explore language and the following account is a result of twelve years of development. I will summarise the main stages to give a flavour of the modelling sequence as well as to emphasise the recursive nature by which models are developed.
My response to my more serious concerns has been a continual developing of my modelling of modelling. I will outline the main stages below .I suggest also that you study the summary sheets provided and I would recommend you complete the exercise I have included in this article in order to give you an experience of the additional questioning possible with the Fractal Language Model.
Part One: Modelling and Re-Modelling
The original M .M. was the product of modelling so I will begin with a few point about models and modelling before examining the M. M. and my developments.
A model of the interaction between structure and process underlies much of the argument of this book, and it will be critical to understand the relationship between these notions and the problem of knowledge or description.
A model has several uses: first, to provide a language sufficiently schematic and precise so that the relations within the subject that is being modelled can be examined by comparing them with relations within the model. Occidental languages, in general, do not lend themselves to the discussion of relations.
We start by naming the parts and after that the discussion of relations between the parts appear as predicates attached usually to a single part – not to the two or more parts among which the relation existed. What is required is precise talk about relation, and a model will sometimes facilitate this. That is the first purpose of a model.
A second purpose of a model appears when we have a vocabulary of relations, for then the model will generate questions. One can then look at the subject which is being modelled with these specific questions in mind – and perhaps find answer to them.”
Bateson, Angels Fear, p.37:
In DBM we constantly E-valuate our models for further development. Our main E-valuations are firstly ‘is the model “effective”‘, if so then, is it “efficient”, then if so, is it “elegant”. The summary table below outlines some of the relevant issues. The life cycle of a model is given next..
DBM E-Valuating Models
|E-Valuation and Description||Performative Impact||Performative Effect||Performative Affect|
It does what it is designed for
It can be applied effectively
It is simple and beautiful
Modelling Life Cycle
|Antiquated||Obsolete||Create New Model|
Part Two: The Meta Model
I first came across the M. M. in 1979 through ‘Magic I’. I was already familiar with Chomsky’s model of Transformational Grammar, the source of the M. M.. Anyone who has read this book will be aware that the structures are not summarised clearly so I was fortunate in having to make my own summary. I say fortunate because if I had been given a clear summary I would not have needed to understand the differences between the structures in enough depth to classify them. For me the lost performative language structure was difficult to place. If it was as it states the performer is deleted as in “It’s important to study”, question, “Says who?”, then it should be a deletion. It was included in the distortions as there seemed to be more ‘going on’ than a simple deletion. Through the M. M. Bandler and Grinder aimed to offer a model for therapy. In the ‘forming’ of their model a certain information regarding transformational grammar and therapy was used to ‘conform’ to a chosen structure.
Since one of the main ways in which therapists can come to know and understand their clients is through language, and since language is also one of the primary ways all humans model their experiences, we have focused our work on the language of therapy. Fortunately, an explicit model of the structure of language has been developed independent of the context of psychology and therapy by transformational grammarians. Adapted for use in therapy, it offers us an explicit Meta-model for the enrichment and expansion of our therapeutic skills and offers us a valuable set of tools to increase our effectiveness and, thus, the magical quality of our own therapeutic work.
Magic I, p. 18-19
The M. M. was first published in 1975 in the ‘The Structure of Magic, Volume I. I am
assuming that readers are familiar with the basic model. The main distinctions are:
SIMPLE and COMPARISONS:
LACK OF REFERENTIAL INDEX
CAUSE-EFFECT (X CAUSES Y)
Some benefits of the Meta Model
- It details explicit questions based on the structure of language – there is always an answer.
- Works with the clients model rather than content..
- Draws attention to the benefits of explicit models.
- Can stimulate an interest modelling.
Part Three: Questioning the Meta Model
A. The Map
How was the M. M. formed?
The ‘format’ chosen for the M. M. included three so called universal modelling principles of deletion, generalisation and distortion. I could find no source of them apart from Bandler and Grinder. On closer reading it seemed that they were their own terms. Transformational Grammar used only deletion as a central term.
Deletion, generalisation and distortion are three different types of label. In other words the internal structure of the meta model was not well modelled in terms of ‘types’. A deletion is about what is taken out, a generalisation is a change in what was there and distortion a change from what is there. They are unrelated categories.
Again, we wish to point out that our categories do not impose any necessity on the structure of reality – we have found these categories useful in organising our own thinking and actions, both in presenting this material and in therapy; that is, in developing our model for therapy. We suspect that most readers will, if they think about the usual meanings of the terms, come to see Generalisation and Deletion as special cases of Distortion. Magic I, p. 20, Footnote 8
A model should be integrated if it is to be more than a list or classification (see my last article regarding levels of modelling). A list or classification can be useful but Bandler and Grinder claimed that the M. M. is both a ‘model’ and ‘meta’ – a model of the model known as language. This is indeed a grand claim to make. So ‘how specifically is it a meta model?’.
Fortunately, an explicit model of the structure of language has been developed independent of the context of psychology and therapy by transformational grammarians. Adapted for use in therapy, it offers us an explicit Meta-model for the enrichment and expansion of our therapeutic skills and offers us a valuable set of tools to increase our effectiveness and, thus, the magical quality of our own therapeutic work.
Magic I, p 19
How complete is the M. M.?
The Meta-Model we are presenting is in large part inspired by the formal model developed in transformational linguistics. Since the transformational model was created to answer questions which are not immediately connected with the way that humans change, not all portions of it are equally useful in creating a Meta-model for therapy. Thus, we have adapted the model, selecting only the portions relevant for our purposes and arranging them in a system appropriate for our objectives in therapy.
Magic I, p. 40
Bandler and Grinder would appear to have included only the portions that they thought were relevant for therapy. Why did they select the structures they did? They do not say. Did they have other structures? The simple answer to this is yes. In Grinder’s book on Transformational Grammar there are a number of additional structures. For example from page 80-84 six structures which were not included in the M. M. are described (The either argument, the Tag question, the ‘not even’ fragment argument, the ‘neither’ fragment argument, the ‘some any’ shift argument, the ‘until’ argument). A number of other patterns are given later in the book. Many of these patterns I consider to be great benefit in therapy and beyond. Clearly the M. M. could be extended. Bandler and Grinder were of a similar opinion at that time. In their very useful bibliography they twice identify potential material for extending the M.M.:
An excellent example of the General Semantics approach which we feel will contribute much to an enlarged Meta-Model for therapy. P 221
B. The Territory: The Wizards: Other questions asked by Erickson, Perls, etc.
Bandler and Grinder had other patterns available but what about the ‘territory’ they were modelling? Did the ‘therapeutic wizards’ whom they modelled use other patterns of questions? Again the answer is yes. Open any book with transcripts of Perls, Satir, Erickson you will find many other types of questions.
Perls : From: Gestalt Therapy Verbatim
1. M. I can’t go back.
Perls: You cannot go back. Who prevents you? p. 109
2. Perls: And now?
M: They go away.
Perls: And then?
M: Then I am Alone.
Perls: And you are safe? p. 112
Satir: From: Virginia Satir, The Patterns of Her Magic, Andreas
1.48.18: Linda: No, she couldn’t touch me.
48:20 Virginia: Why?
2.48:25 Linda: I was a wonderful little baby.
48:28 Virginia: “Wonderful little baby”. Where did you get that idea?
Erickson: From: Life Reframing in Hypnosis, M. H. Erickson, Irvington, 1985
1. Erickson: Is it rolled up enough? p. 83
2. Erickson: Why do you say yes? p. 239
It is not just the ‘wizards’ whom they modelled that used other questions, Bandler
also used other questions.
Bandler : From: Magic In Action
1. Bandler : “ You have a belief, right? Right? Now, is it a strong belief?” page158
2. Bandler: “For example, what are some of the differences?” page160
“Yet, while the techniques of these wizards are different, they share one thing: They introduce changes in their client’s models which allow their clients more options in their behaviour. What we see is that each of these wizards has a map or model for changing their client’s models of the world – i.e., a Meta-model – which allows them to effectively expand and enrich their client’s models in some way that makes the client’s lives richer and more worth living.” B and G., Magic I, p. 18
The wizards are using more than M. M. questions. How do Satir, Perls, and Erickson know when to use a M. M. type question or one of the questions above that do not fit the M. M.? The answer cannot be formed using the M.M. so clearly the M. M., however useful, is not a ‘Meta Model for Therapy’. It cannot be used to organise the therapists questioning in a MEANINGFUL way. It is not a model of meaning and purpose but a model for specifying the historical changes in the structure of the client’s statements.
C. The Territory: Some other language uses not included in the Meta Model
- From Transformational Grammar, Grinder lists a number of patterns not included in the later M. M..(Grinder 1973)
- Other NLP Language Models: Milton Model, Sensory Predicates, Metaphoric, Presuppositions, Sleight of Mouth patterns
- The M. M. is only for constructing questions. A complete model would include statements and commands (integrate with the Milton model).
- From my modelling experience: Concepts, Qualifiers, Concurrence, Judgements, rules, inferences, numbers, performative language. There are no M. M. questions for questions! The M. M. makes no distinction between ‘walk’ and ‘hurry’(both unspecified verbs). We can go for a walk, we do not go for a hurry. The higher functions of ‘love’, belonging’ are treated the same as sensory behaviours of ‘walk’ or ‘talk’. Try the M. M. question ‘how do you know?’ to each of these and experience the difference.
- The DBM ‘Epistemology Grid’ (see my previous article) uses a fractal structure for what – how –why which is very useful for relating different levels of understanding. This emphasises the great benefit of the question ‘why?’. This was actively discouraged in NLP. I have been given many different reason as to ‘why’ why shouldn’t be used. One reason was that parents ask this of children and so it will anchor a negative state. Another, more serious, reason was that the question ‘why?’ takes the client out of their model and so you will not get sensory descriptions. This reason is valid only if you are wanting to stay at the sensory level. When we are working with values, beliefs and identity or with strategies and modelling where sequence is important then it is necessary to know why things need to be in a certain order. With the Epistemology Grid we can ask a why level question without using why, for example‘how come?’ or ‘what is the reason for that?’. If you don’t know why you have not to use why and if there is a paramount reason (which I doubt) then there is no guarantee it won’t be asked with how or what question!
Part Four: DBM Re-modelling of Meta Model and further Language modelling
There has been little official development within the M. M.. The Milton model together with pre-suppositions were added. Meta Model III with some other guidance for the use of the M. M. was also added. Recent additions of questions from General Semantics and elsewhere by Michael Hall are a very useful add-on to the M. M. but they do not address the serious limitations listed below.
If we keep Batesons comments in mind when we study the relationship of the M. M. with the territory of language use then a number of issues become obvious.
1. The model is not an integrated classification but more of a structured list. If it was an integrated model then deletion, generalisation and distortion would be of the same type. They are not.
2. There are many important language structures not included
3. The individual structures do not relate to each other.
4. The model is only designed to retrace the historical ‘transformation’. It is not designed to study language use. This means it can not be a meta model for therapy as intended. The model does not give the required basis for directing questioning beyond what statements the client makes.
5. The M. M. does not relate to our use of language only to the production of surface
structures from deep structure. It does not relate to how these deep structures are created and changed. The main classification structures correspond to what is produced not how it is produced. Much of this is due to the use of filter as a working metaphor. As I pointed out in my last article filtering is the opposite of what happens. To model language use we require to identify what is happening in positive terms. What is the client doing that results in deletion, generalisation and distortion?
6. If everything is a distortion, as Bandler and Grinder suggest, then to use the term distinguishes nothing. Distortion presupposes a previous accuracy of form, something well formed. It is this ‘well formed’ in terms of models that the M. M. fails to specify.
7. The M. M. responds to the products of our subjective experience, not how we created them. It cannot be used therefore to model its own creation. It cannot be a ‘meta’ model of how we use language.
An improved model would need to respond to these concerns.
I am continually seeking ever more practical and integrated models of language that can be used for therapy, modelling and for any other application area. I will now outline some of the (overlapping) stages in my re-modelling of the NLP meta model..
The first stage (1979 – 1983) as I said earlier was putting my own summary together. I used this for over five years until I wanted to teach other professional how to gather quality information.
The second stage (1983 – 1986 ) was my putting the meta and milton model together. These were presented as the opposite though there was not a perfect overlap. This resulted in the ‘filling in of some gaps’. Part of this stage was to change the aggressive language used with the M. M.. I changed ‘violation’ and ‘challenge’ with ‘pattern’ and ‘specifier question’. This made teaching them easier and the use of them untainted by the over aggressiveness that often seemed to accompany the M.M..
The third stage (1986) was the creation of the ‘Integrated Language Model’ (see diagram). This resulted from a major insight that I had. I wanted to develop a holistic language model. I had puzzled over this for a while before realising that any whole model can be described using three related components of ‘detail’, ‘scope’ and ‘connection’ and that these components could replace the unrelated components of deletion, generalisation and distortion. creation of detail, scope and connection. to make it positive, what is modelled – a model of the modelling of language. I also realised that mind reading was also something we did to ourselves and not just others as in ‘I can’t do that’. This later led to my realising that all of these were examples of a larger category of ‘judgements’.
The fourth stage (1987 – 1993) involved integrating with other areas of NLP including Presupposition, Sleight of Mouth, Sensory Predicates, Metaphor. Detail – Scope, and connection prove to be very useful in re-modelling and integrating many diverse areas of NLP.
The fifth stage (1987 – 1999) involved integrating other models of language. There have been many other language models, each with something to offer. I wanted to make sure that I included the best of all of them. There had been a few specific models before the Meta Model. The General Semantics of Korzybski offered a number of structures. Erickson and Rossi identified many hypnotic language structures.
The sixth stage (1992 –1996) involved the creation of the basic fractal language model. This was an integration of what I had discovered and learnt in the previous stages. This model had detail – Scope –Connection at two levels for two categories giving 18 Language distinctions.
The seventh stage (1996 – 1998) involved the creation of the much extended ‘Fractal Language Model’ (see diagram). I created this by applying Fractal Modelling to the Basic Fractal Language Model. In Bateson’s terms the ‘relationships’ suggested themselves once I had the basic patterning. For the first time I was able to use the model to predict new language possibilities. In this model there are 81 related distinctions. This truly becomes an effective model to use for modelling in order to make useful distinctions and to assist clients in making precise changes.
Benefits of Fractal Language Model
1 It is more effective. It covers the territory more accurately with far fewer gaps and inaccuracies.
2 Many more questions are possible through having many more distinctions. This creates a greatly improved basis for modelling, therapy and all other applications.
3 It is practical. It is easy to start with general details that are bounded (scoped) and connected together in some way and then to develop more precision using the next two levels of distinctions. From a detailed exploration you can easily return to a simpler overview.
4 It integrates with Higher Functions through the Fractal Model Meaning and Fractal Model Thinking and to subjective processing through Life Grid and to other Fractal Models – Motivation and attainment Models; Sensory Processing; other Language Models – Presupposition, Erickson and Rossi’s Language patterns, Sleight of Mouth, etc. This greatly facilitated my integration of many diverse areas of NLP and will be evident in later articles.
5 By being integrated to all the clients processing it can be used to identify what to change (How the client has organised their model of the world and how to usefully assist them in changing). This is consistent with intentions of Bandler and Grinder for their M. M..
6 It can be used to model the processing of subjective experience as well as the products.
7 It is inter-related and holistic.
8 It is developmental. It can be applied to its own production.
Part Five: Application: Conversational Belief change
This is major DBM model which describes how, based on our knowledge of the world (know) we organise our behaviour (do). When we do not know and need to do things it is useful to explore the world (investigate).
The “know” part of this model is relevant here. In DBM we identify three levels of knowledge. The first, knowledge with evidence, we call knowing. The second is for when we don’t know for sure but we have some evidence – we call this believing. Beliefs will vary from 0 – 100% in our sense of commitment to them. That is why we can have a strong or weak belief – we never describe ourselves as having ‘strong’ knowledge (we talk about a lot or little knowledge not strength of knowledge). Beliefs then are our judgement tool for when we don’t know. They let us ‘know’ what might be or could be until we get to really know for sure. The third level is for when we need to know in order to do but have not got any evidence. This is conviction. Conviction is digital like knowledge, and is a necessity in an unknown world and as such it potentially has the most limiting consequences for us. NLP has generally used the label “belief” to describe this whole area of knowledge, beliefs and convictions.
Modelling Inaccuracies in NLP
We can use the three types of judgement to clarify inaccuracies in NLP (and elsewhere). I have chosen two examples. The first is from the sub-modality Belief Change developed by Richard Bandler. In this technique the client is asked to identify a ‘strong belief’. Richard often suggests the belief that “the sun will come up tomorrow”. I don’t know anyone who actually believes this! It is in fact a basic knowledge of the world. The technique in this instance uses knowledge to change a belief. If an NLP practitioner follows the procedure and the client actually picks a strong belief and not knowledge then the results are different from those that Richard gets! Maps and territory problems again!
The second example of how I have used the model to clarify NLP is from the work of Robert Dilts. In his book ‘Changing Belief Systems’ Robert uses the following example to demonstrate the difficulty in changing a belief. In the example a psychiatrist has a patient who ‘believes’ he is a corpse. The psychiatrist asks, “Do corpses bleed?” He then takes a needle and pricks the patients finger and of course the man bleeds. Dilts continues:
“Since the patient is a corpse there is nothing much he can do about it. So the psychiatrist sticks him with the needle and the man starts to bleed. The patient looks at it totally amazed and says, “I’ll be damned. Corpses do bleed!”
The point is that when you have a belief, even environmental and behavioural evidence won’t change it because a belief isn’t about reality. You have a belief in place of knowledge about reality. Beliefs are about things that nobody can know in reality. If somebody has a terminal illness he doesn’t know if he is going to get well. There is no present reality as to whether he is going to get well or not. He has to believe that he is going to get well precisely because nobody knows what the reality is.”P.8
With my more detailed model of judgements we can understand this situation differently. The client has a conviction or knowledge of self and not a belief. That is why the counter example strategy does not work. In my modelling and client experience counter examples tend to reinforce identity (self knowledge or conviction) and so can make things more difficult to change while genuinely trying to help! We will often express knowledge through saying that we believe or think something is the case. If you are asked if you belief that the sun will rise tomorrow you can easily say that yes you “believe it” because it is contained within your knowledge of the world. The success gained in NLP through taking language literally ( as in I ‘see’ what you mean) gets in the way here.
The potential limitation of success has been usefully highlighted by Robert Dilts in his book on beliefs and health (Dilts 1990):
“One of the claims that I make is: “Success is as much a limitation to creativity as is failure”. This is because when you remember a success, your memory often becomes really strong and you get a good feeling about it. You are likely to keep doing the same thing over and over again without exploring other options. You get to a point where you stop being creative and get stuck, because you’ve run into a new situation where your old behaviours don’t work and you don’t have new choices.” Page 40
The great success of the M.M. and literal language in NLP has indeed resulted in a failure – a failure to accurately model and to develop further the tools that will assist all of us to develop more effectively, efficiently and elegantly.
Exercise: Changing Judgements
This exercise sequence is one way to assist clients to develop a more accurate self knowledge. Please refer to the Fractal Language Model for more detail.
Step One: Judgement
Identify a limiting judgement, eg, “I can’t learn mathematics.”
Step Two: Equating
Question this to identify evidence
- How do you know?
- What else?
This will result in either an ‘equating’ structure or else ‘I don’t know’ if the subject has taken on the judgement without understanding.
Step Three: Inferring
Question to get the underlying thinking (inferring) that produced the judgement.
- When did you 1st decide / conclude / realise this? (Each of these three will get a different answer and come from the Fractal Thinking Model which also has 81 related distinctions).
- How did you decide / conclude / realise this?
- What else did you do?
- What else do you know (or could know) now?
- What would you decide / conclude / realise differently NOW?
In this step the subject might also answer ‘I didn’t’ if they haven’t thought it through themselves but only accepted someone else’s reasoning. If the answer at any stage is ‘I don’t know then a new judgement can be formed using question 5 to investigate.. (Further details and exercises are available on the Sensory Systems web-site)
In my experience we don’t just retrieve past ‘transformations’ of language when we answer questions – we can also change our sensory experience, our thinking and change our language. What is worth changing will be dependent on what we are aiming for in terms of a usefully ‘well-formed’ model of the world. Bandler and Grinder proposed:
“…… that there is a subset of the well-formed sentences of English which we recognise as well formed in therapy. This set, the set of sentences which are well formed in therapy and acceptable to us as therapists, are sentences which:
- Are well formed in English, and
- Contain no transformational deletions or unexplored deletions in the portion of the model in which the client experiences no choice.
- Contain no nominalisations (process → event).
- Contain no words or phrases lacking referential indices.
- Contain no verbs incompletely specified.
- Contain no unexplored presuppositions in the portion of the model in which the client experiences no choice.
- Contain no sentences which violate the semantic conditions of well-formedness.” Magic I, Page 54
These conditions mean that Bandler and Grinders well formed in therapy is practically impossible as there will always be something unexplored. In terms of a well-formed outcome all of them apart from the first one is stated negatively! What are we to go for that is well formed and how will we positively recognise it? Well formed in therapy needs to be about the function of the language not the structural content of what is said- the grammar. The function of language is to organise and communicate meaning. Well formed in therapy should be well formed in terms of the ecology of the meaning. This can be done by E-valuating the clients model for its effectiveness, efficiency and elegance for their life as a whole – an ecology of being.
Modelling developmentally is an on-going process. The products of modelling are static and not easily open to change as evidenced by the twenty-five years of unchanging meta model. And so we must always return to the world and ask how we can make our models more effective, efficient and elegant. There is always more!
“We Shall not cease from explorationAnd in the end of all our exploringWill be to arrive where we startedAnd know the place for the first time.”T. S. Eliot, Little Gidding, from Four Quartets 1943
Andreas, Steve, Virginia Satir The Patterns of her Magic, Science and Behaviour Books, Inc. 1991
Bandler, Richard & Grinder, John, The Structure of Magic vol. 1, Science and Behaviour Books, Inc. 1975
Bander, Richard, Magic in Action, Meta Publications 1984
Dilts, Robert, Hallbom, Tim & Smith Suzi, Beliefs: Pathways to Health & Well-Being, Metamorphous Press 1990
Dilts, Robert, Changing Belief Systems with NLP, Meta Publications 1990
Grinder, John & Elgin, Suzette Haden, Guide to Transformational Grammar, History Theory Practice, Holt, Rinehart and Winston, Inc. 1973
Lewis, Byron & Pucelik, Frank, Magic of NLP Demystified, Metamorphous Press 1990
Perls, Frederick S., Gestalt Therapy Verbatim, Real People Press 1969