Steve Andreas on Symbolic Modelling

From Six Blind Elephants: Understanding ourselves and each other
Print Friendly, PDF & Email

In Six Blind Elephants: Understanding ourselves and each other, Volume I  Steve Andreas (1935-2018) devotes a section to analysing the questions and methodology of Symbolic Modelling.

Before the book was published Steve kindly sent Penny Tompkins and myself the manuscript of the relevant chapter for our comments.

We thought Steve had captured as much of the essence of the approach as could be expected in four pages and had contributed a valuable and novel perspective. After a short email dialogue Steve made a few adjustments to the final version which he has generously allowed us to reproduce below.

Following the extract you will find my (17 September 2005) comments on Steve’s original manuscript. Steve’s analysis raised some fascinating questions and in my reply I attempted to make some fine distinctions about clean questions and a clean methodology.

Extract rrom Six Blind Elephants, Volume I (Real People Press, 2006) by Steve Andreas, pages 130-134:

Scope and category in psychotherapy

Most therapies are pretty vague about exactly what questions to ask or what statements to make in order to help people change. General methods, like “paraphrasing,” “reflecting feelings,” or “empathizing,” are taught without specifying how they can be used systematically to change scope or category. Accordingly, we would have to study extensive transcripts of sessions in order to find out how most therapies change scope and category.

However, there is a very systematic approach developed by James Lawley and Penny Tompkins called “Symbolic Modeling” (40) using “Clean Language,” based on the work of David Grove. In addition, the book describing this approach is exceptionally clear and specific, and the authors have also produced an unedited videotaped demonstration, (53) something that very few therapists are willing to do. A videotape makes it possible to review exactly what the practitioners of an approach actually do in a session – which is often very different from their description of what they think they are doing. A videotape also provides a wealth of nonverbal information, which is a very important aspect of communication, often even more important than the verbal.

In very condensed outline, in Symbolic Modeling the therapist begins with “a short personal history and asks them to define an overall outcome.” After this, the “self-modeling” process starts with the “basic opening question,” “And what would you like to have happen?” The therapist listens carefully to the words in the client’s response, and also for the symbolic or metaphoric aspects of the client’s communication. The therapist then asks a question that includes some or all of the exact words just used by the client. These questions are also sometimes asked in response to nonverbal postures or gestures, including them in the client’s scope of experience.

The question is usually taken from a list of nine “basic developing questions,” (40, p. 54) and less often from another list of 21 additional “specialist” questions, (40, p. 283) which are asked less than 20% of the time.

    In the list of nine “basic developing questions” below, three dots (. . .)  indicate where the therapist repeats the client’s words, which “can be a single word, a phrase, or everything the client has just said, depending on where you want to direct their attention.” Read these questions and decide whether each one elicits a change in scope or category.

  1. And is there anything else about . . . (client’s words)?
  2. And what kind of . . . is that . . . ?
  3. And that’s . . . like what?
  4. And where is . . . ?
  5. And whereabouts . . . ?
  6. happens?
  7. And what happens next?
  8. And what happens just before . . . ?
  9. And where could . . . come from?  

Question 2 asks the client for an explicit categorization of their experience at a more specific logical level, a subcategory of the existing one. Question 3 asks the client to make a comparison, and invites a metaphoric response, which requires an implicit categorization at a more general logical level.

All the other questions ask for scope. Question 1 asks for more detail, while 4 and 5 ask about scope in space. Questions 6 and 7 extend the scope of time into the future, and 8 and 9 extend the scope of time into the past. There is a good deal of redundancy in the questions; 4 and 5 are nearly identical, as are 6 and 7.

Of the 21 additional “specialist” questions, which are asked less than 20% of the time, two ask for a comparison, which implicitly ask for category; one of these asks “same or different?” and the other asks about a relationship between the two elements of experience.

Again, all the other specialist questions (19 of 21) ask the client to clarify some aspect of scope. Nearly half ask about some aspect of space (“where,” “distance,” “direction,” “far,” etc.), and six have to do with motivation or intention (“like,” “drawn,”). This emphasis on developing awareness of scope is parallel with many meditative and spiritual approaches that say the deep and full awareness of a problem is all that is needed to bring about movement and change.

The repeated use of “and,” and “as” connects each question, and the response to the question, to the client’s preceding experience. The repetition of the client’s preceding words also connects each response and question in the same way, so that each new scope or category is added to what is already in the client’s experience (rather than replacing it). This expands and enriches the client’s experience of the problem or outcome.

To summarize, this psychotherapeutic method is based on asking the client to change the scope of their experience in a variety of ways, occasionally asking them to categorize it metaphorically. Expanding scope always results in more information that can be useful in solving problems. Asking the client to recategorize their experience is also useful, because that associates it with other examples in the new category.

This process continues until the client comes to some kind of resolution. The therapist never makes a statement, and only responds from the lists of questions. This is called “clean language” because the therapist does not introduce new content, but simply asks open-ended questions about what the client has already said. Asking the client to change scope is a content-free intervention, and asking them to recategorize their experience is also content-free, since the recategorization emerges from the client’s experience.

However, this protection from content introduced by someone else is a double-edged sword, because it also protects the client from a content intervention that might be very useful, and which could save a lot of time. If the client has to come up with a new categorization of the problem in which they are stuck, they will often be unable to think of a useful one that might be very obvious to the therapist or someone else. “When you’re inside a box, it’s hard to think outside the box.” Input from outside the box can sometimes speed up the process of reaching resolution immensely.

For instance, in dealing with objections to reaching forgiveness (10), many people say angrily that the other person doesn’t deserve to be forgiven. If this objection is not resolved, it will prevent the person from reaching forgiveness and resolution. I know from experience that this objection can be resolved very quickly and easily by a simple change of scope.

“You may be absolutely right that they don’t deserve forgiveness. However, forgiveness isn’t for them, it’s for you – so that you can be free of resentment, obsessive thoughts about revenge, etc. As Nelson Mandela said after spending 27 years in prison, ‘Resentment is like drinking poison, and waiting for your enemies to die.’ ” It could take most clients months or years to find this essential understanding on their own. While this is certainly an example of the therapist offering content to the client, it is one that can help the client become open to a very useful resolution.

When someone is grieving over a loss, they often make the mistake of recalling the ending of the relationship, rather than the treasured experiences that are no longer available to them. The first step in resolving grief (6) is to replace this image of the ending with an image of the relationship at its best. Without this change of content, it is impossible to resolve grief.

Of course, introducing content in this way is also a double-edged sword. While an appropriate content intervention can speed up a process, an inappropriate content intervention can divert the client’s attention from the problem to be solved, and even make the problem worse. “Clean Language” is an evaluative term that categorizes not introducing content as “clean,” and by implication, any content intervention as “dirty.” Either can be useful, depending on the goal and the results, so I think it is inappropriate to use an evaluative term for one or the other except as it is applied in a particular context. Perhaps “content-free” response would be a more descriptive term than “clean” language.

Despite the specificity of this approach, and the emphasis on Clean Language, there are two elements that are uncontrolled, and can still allow the bias of the therapist’s beliefs and orientation to influence the client. The first is the choice of which words of the client to repeat back to them. “A single word, a phrase, or everything the client has just said, depending on where you want to direct their attention,” allows a lot of opportunity for bias.

For instance, a therapist with traditional psychiatric or psychodynamic training would be likely to choose words or phrases hinting at disease, pathology or problems. Their nonverbal responses will also have the same bias, leading the client to explore past events and causes. Someone with Solution-Focused training would likely choose words that indicate competence, future possibility, and solutions, with accompanying nonverbal signaling. This bias is much more likely to result in exploring future solutions.

    A second source of potential bias is in the choice of which questions to ask. Again, someone with psychodynamic training would be more likely to ask questions about the past, despite the fact that more of the questions are directed toward future possibility (“could,” “would,”) while someone with Solution-Focused training would tend to ask even more questions about future possibility.


6. Andreas, Steve. “Resolving GriefAnchor Point Magazine, Vol. 16, No. 2, Feb, 2002.

10. Andreas, Steve. “ForgivenessAnchor Point Magazine, Vol. 13, No. 5, 1999.

40. Lawley, James; and Tompkins, Penny. Metaphors in Mind; Transformation through Symbolic Modelling. London, Developing Company Press, 2000.

53. Tompkins, Penny; and Lawley, James. “A Strange and Strong Sensation.” (DVD) London, The Developing Company Press, 2003.

Steve Andreas, with his wife and partner Connirae, has been learning, teaching, and developing patterns in Neuro-Linguistic Programming (NLP) since 1977. Steve and Connirae are the co-editors and/or authors of many NLP books (both classics from the early days of the field, and new innovations) and over fifty NLP articles. They have also produced over fifty videotaped and audiotaped demonstrations of NLP patterns for personal change.  For information about their work go to:

James Lawley comments

Dear Steve,

Thanks for the opportunity to comment on your manuscript.  Your summary of Clean Language seems as accurate as you could get in a few pages.  Penny and I have no concerns about you publishing ‘as is’, with one exception – we request that you credit David Grove as the originator of Clean Language.

Your chapter helped me to improve my ability to distinguish between scope and category shifts and how to apply that to the Clean Language. Although I had never thought of it before, it now seems obvious and appropriate that the vast majority of Clean Language questions request information about scope.

As you say, a number of the original 9 basic questions ask the client to attend in similar ways. We noticed this too and so we revised the model and published it in an article Clean Language Revisited in 2004.

Reading your manuscript prompted a whole number of thoughts which I have taken the opportunity to highlight  below.

1.  We consider Clean Language the ‘means’, modelling the ‘method’, and metaphor the ‘medium’ in which we work.  We feel it’s important to keep in mind that these three aspects operate at different logical levels. Clean Language is a set of questions and a behavioural description of how to ask them devised by David Grove. People have used these questions in a whole variety of ways, many of which have got nothing to do with modelling, symbolic or otherwise. It is even possible to be downright dirty with Clean Language. Symbolic Modelling is a way of working with human perception. It  can be applied to a therapeutic setting, and with subtle changes, to many other contexts.

2. We maintain that ‘being clean’ is good practice when modelling or facilitating self-modelling. Of course there are times when it’s valuable to ‘be dirty’ or leading but does the modeller have a very good purpose for doing so, and are they aware how much of their own map (metaphors, assumptions and presuppositions) they are adding into the modelling process? Having observed many hundred people (many of whom are NLP trained) attempt to model we have noticed that most simply have no idea when they are introducing their metaphors into the modelling process. At worst, the modeller ends up with a model that is as much about them as it about the exemplar. At best, the exemplar either has to keep correcting the modeller or translating the extraneous metaphors into their own. Either way it can be inefficient and frustrating for the exemplar.

3. Given that you classify the ‘Anything else?’ question as asking the client for more detail and the ‘What kind of?’ question as asking the client for an explicit categorization of their experience, perhaps you could comment on how you’d classify the following examples:

C:    It’s in my heart.
T:    And is there anything else about that heart?
C:    It’s red.

C:    It’s in my heart.
T:    And what kind of heart is that heart?
C:    It’s red.

C:    It’s in my heart.
T:    And is there anything else about that heart?
C:    It means I can be all of who I am.

One of the features of most CLQ’s (as the examples above demonstrate) is that they give the client wiggle room to respond in their natural manner.

4. I feel that one of the most important things to appreciate about using Clean Language is that it was originally designed to be used within the symbolic or metaphoric domain of experience.  David Grove found, after many years of clinical experience, that to work in the symbolic domain requires a special approach. Until a person has experienced, say, 30 minutes of working in their own symbolic domain, they simply cannot appreciate what that is like. I’ve lost count of the number of people who have told me that it is such a different experience to be asked the questions than to watch someone answer them. So far we have not discovered any other questions that encourage and maintain a symbolic perspective as elegantly as does Clean Language.

5. You give the example of a therapist being able to introduce a new category of thinking to someone who is ‘stuck’. As you say it is a “double-edged sword” to be “protected” from introducing content. Equally, introducing a new category is double-edged sword as doing so may undervalue the function of stuck, or some useful aspect of it, as well as the client’s own ability to get unstuck.

In Symbolic Modelling we assume that “stuck” is serving a purpose or function (not necessarily a ‘positive’ intention). Therefore in Symbolic Modelling we will facilitate the client to explore their subjective experience of the metaphor “stuck”. If nothing else this is an opportunity for the client to learn about how they do “stuck”.

In most cases, however, the exploration leads the client to make new connections, take a new perspective, reframe, or some other useful response that may not have been available had we offered them a new categorization. Of course there are times when it is valuable to offer content. Teachers, for example, are paid to do just that.  And, it is my experience through training therapists from dozens of different schools that for many the desire to come up with a new categorization or solution is akin to an addiction. Many coaches, for example have said “But what’s my purpose if I don’t offer solutions?”, “Good question,” I reply.

6. In your example about forgiveness, you say that “It would take most clients a very very long time to find this essential recategorization on their own.”  In Symbolic Modelling I would likely give the client a good opportunity to notice what they do when they can’t find a recategorization, and to ask them to attend to that dilemma.  I might well say:

“And when you want to forgive and they don’t deserve forgiveness, what would you like to have happen?”


“And when you want to forgive and they don’t deserve forgiveness, then what happens?”

At first some clients will either not understand these questions or find a way to not answering them (especially the first one). However if I continue to repeat the question, say four to six times, the client eventually attends to their current reality: that they both want to forgive and continue to believe that the other person does not deserve forgiveness, and that they do not know how to do both.

When this place/state is accessed, very often a new way of seeing the situation or themselves emerges spontaneously. If it doesn’t, as a last resort, I can still make a suggestion about what I know about forgiveness (which they may or may not find helpful). In my experience, some clients need to fully appreciate that their tried and tested methods of solving problems will not work with this class of experience, before they become open to a new approach. It may also be that the forgiveness dilemma is but one example of the client experiencing other, structurally similar, dilemmas. If this were the case I would refrain from adding any suggestions as to how to resolve the forgiveness issue because this might reduce the opportunity to address the larger question: How to deal with such dilemmas.

7. This leads me on to my next point.  Clean Language is anything but “non-interventionist”. The examples I give above are hugely interventionist because they attempt to define the context in which the client attends. The difference between a clean intervention and a leading intervention is not just one of content, but also of the closeness with which the intervention stays to the logic of the client’s map.  Clean interventions attempt to work with the logic.  Many other interventions attempt to work against, outside, or to defeat the client’s logic.  Again, this can be very useful.  However, we say to our trainees that they have to earn the right to be dirty, and they do that by demonstrating that they have the flexibility to stay clean, even when the going gets tough.  

8.  The Where? and Whereabouts? questions might seem almost identical, but that’s not what people who have attempted to translate these question into other languages say.  For example in French there is no equivalent of ‘Whereabouts?’.  In English ‘Whereabouts?’ can either mean ‘where more specifically?’ or ‘where more generally?’ depending on the context, and because of that it is a very useful question.

9. I see that the distinction you make between “useful” and “not useful” to the client is “depending on the results”.  That’s fine for evaluating an intervention after the results are known, but it’s not much help before making the intervention because (a) you cannot know the result in advance and (b) once you’ve made one intervention it’s impossible to go back and find out what would have happened if you had made another (clean or not) intervention.

10. I heartily agree that the art of Symbolic Modelling is in the choice of which question to ask of which words.  And that therapists with particular training are drawn to ask about particular classes of experience.  In addition, there is an almost irresistible tendency for novices to ask questions about metaphors and symbols which have significance for them. This is very subtle and out-of-awareness because most therapists have no idea which symbols and metaphors will unconsciously activate and bias their attention.  As part of our training we insist therapists become intimate with their own favoured metaphors so that they can consciously counter their magnetic effect.  It’s a bit like interviewers being trained to recognise their own prejudices so that they can take steps to make the interview fairer for the candidate.  

11. It is worth noting that when we are using Symbolic Modelling in a therapeutic setting we have a different intention to what might be called ‘standard’ NLP modelling.  In the latter, the aim is to find a generalised model that has wide applicability.  In Symbolic Modelling we are looking for the unique, idiosyncratic aspects of the client’s map – the bits that make them them and nobody else.  This means that while each client’s metaphor landscape may have common features it will have elements that do not conform. And the latter will evolve in a way that is within the logic of the Metaphor Landscape particular to the client.  Symbolic Modelling seeks to work alongside this personalised “arrow of evolution” recognising (as complexity theory teaches) that a change in the initial conditions may lead to all sorts of unpredictable outcomes.  To use another metaphor, because a human is a self-organising system, the unfolding of their story emerges through its telling.

A further question from Steve

I also have a question for you. Are there any sorts of problems that you think Symbolic Modelling is inappropriate for, or that you don’t get satisfactory results? For instance, I would think that it would not work well for a phobia – at least without some careful preparation.

James replies

There are areas where we think other approaches are probably more applicable than Symbolic Modelling.  For example:

  • We have successfully used Symbolic Modelling with phobias (the client gets the ‘distance’ by examining a metaphor for the phobia or symptoms) but unless there is something unusual about a phobia then techniques such as the fast phobia cure and EFT are more efficient.
  • Unless the practitioner is experienced in working with clients who have a poor grasp on consensual reality, we do not recommend Symbolic Modelling is used with this client base.
  • There are probably better approaches for working with people who have a very low level of self-awareness or who have a low-capacity to learn how to be self-reflective.
  • Clients who want/need advice or external feedback should look elsewhere.

Otherwise, we and our colleagues have used Symbolic Modelling successfully in settings as diverse as maximum security prisons and kindergartens. We have found that Symbolic Modelling is particularly suited to those areas not easily resolved by traditional techniques, e.g.

  • The big or highly generalized issues of life.
  • A vague or ill-defined sense something’s wrong, fearful, unsafe, missing, etc.
  • Intractable and double-binding problems.
  • High-level identity and spiritual stuff.
  • Obsessive and addictive behaviour.
  • Subtle and pervasive dissociations/splits of perceiver.
  • Misaligned, distorted perceptions of epistemology (this could be time, space, and in particular, scale).
  • Situations where a lot of transference and counter-transference may be operating, e.g. In Northern Ireland some counsellors work with clients who directly or indirectly have had a hand in physical violence against a family member or neighbour of the counsellor.  In these situations the discipline of Clean Language and the use of metaphor really helps the counsellor to keep their personal stuff out of the therapeutic relationship.

Warm regards,
James Lawley
17 September 2005

Print Friendly, PDF & Email
body * { color: inherit !important; }