Analyzing language and behavior in Top Performers
and its uses in the employment area

Patrick E.C. Merlevede, M.Sc

Although it may seem a bit dated in 2010, this paper lies the foundation for the Methodology based on Models of Excellence, which has been developed for This paper was presented at the International Colloquium on New Technologies based Learning and Employment Support, held in Belgium from 17 to 19 September 1997. This colloquium was a joint project from the Laboratory for Applied Epistemology of the University of Gent and the Laboratory for Cognitive Sciences from the University of Mons - Hainaut.


From someone's language in an interview, we can successfully predict how he or she will behave in a given situation. The main idea is to analyze the top performers in the company for the structure behind their excellence and to use this structure for the employment of other persons that have a similar job profile. Several series of tests have been developed in recent years for this purpose. This paper discusses the tests that are based on the Motivation Traits and Working Traits, coming from the field of Neuro-Linguistics. It will show how to use these tests to recruit people, to coach them and to train them, based on the knowledge from the company itself.

The theory behind the model

Neuro-Linguistic modeling is derived from the fields of Cognitive Science. It was started in the early 70's at the Santa Cruz Campus of the University of California, where scientists of the different fields worked side by side. Among other things, it applies research methodology from cognitive psychology and techniques from knowledge acquisition for building models.

In fact, a "problem" of AI & Cognitive Science is that they try to make abstraction of physics or biology, by mapping human thinking on machines. For building applications, we can use the results of research without this "abstraction", since we intend to apply to humans anyway! However, following the example of AI and Cognitive Science, we stress on the structure of the cognitive processes.

Neuro-Linguistic Modeling

Definition: Neuro-Linguistic Modeling consists in using tools that have their origins in AI and Cognitive Science research with the goal of making a model of excellent behavior, for transfer to other persons.

Its technology basis comes from combining linguistics, AI & Neurology, more specific:

Based on structure, not content

When companies hire people, manage them or train them, they mostly focus on the content. This content will be typically different from person to person, which complicates the situation. For instance, examples given in a training often do not apply to the particular situation the participant finds himself in, and it is difficult to find out how to convert the principle behind the example so that it can be applies. In stead, Neuro-Linguistics focuses on the structure that is the same. In the training, we will stress on the principles, and explore how to apply them with the participants. Below, we explain which structure elements are taken into account.

For analyzing language and for bringing the person closer to the actual experience, we use 3 types of questions, based on 15 language patterns discovered by Chomsky (1965) and Grinder (1973). The first type of questions helps us to question unspecified nouns, the second type serves for finding out what is behind unspecified verbs and a third range of questions explores the limit of the world-view of the person for a specific context (modal operators). The language-patterns used by persons on itself is a structure as well, and will sometimes be considered as being a part of the model.

For analyzing behavior we combine a set of models, such as the13 meta-programs (with sub-classes as Motivational and Working Traits), 6 eye-movements, 6 logical levels and the 4 perceptual positions. Explaining these models goes beyond the scope of this paper, and I'd like to refer the reader to the book of Joseph O' Connor: "Introducing NLP" for more details. The expertise is analyzed using all the dimensions these models offer, both for an expert and for a less experienced person. By comparison, we find the meaningful differences between an expert(s) and the other person(s).

The resulting significant parameters that are withheld are grouped according to the principles of the TOTE and SCORE models. These are 2 models for planning and problem solving. The TOTE model is based upon the feedback loop as worked out in 1960 by Miller, Galanter & Pribram in their book "Plans and the Structure of Behavior".

Sometimes content is taken into account. In that case, it is considered like filling out the structure-model.

The quality of a model is evaluated by the predictability of the results (will someone applying the model get the same results as the expert, within the scope of the model?) and by the simplicity of the model (can the model be explained in less tan 7 steps, does it only include the elements that make a difference, or does it involve complex steps?).

Given the number of parameters to be considered, the complexity of modeling is quite high. Ways to handle the complexity is by making video-tapes of the interviews and the demonstrations given by the expert and the other subjects. Training a modeler requires at least the equivalent of 300 course-hours of theory and practice.

In the rest of this document we will show some possible applications of modeling in the employment area, explain a typical project approach and give an example for programming. For the example we will only expand on some of the Motivational and Working Traits

Possible Applications for Employment Support

The basic instrument for deciding which patterns are withheld, is a contrast analysis between the successful persons and the less successful ones. This contrast analysis is based on the SCORE-model for problem solving, where we differentiate between symptoms, causes, outcomes and effects. The resources, used to get from the problem space to the solution space, show the patterns we are after. For this, we use the TOTE model to further analyze the person's actions. The quality of a model increases with the number of excellent examples taken into consideration, where practice has shown that 3 well-chosen examples are enough to get a good model of one person.

Recruitment: Personnel Selection

The principle of that we want to find similar patterns in the persons we want to recruit as the patterns the model(s) show us. The critical patterns are the patterns shown by all the experts and not present in the counter-examples. If critical patterns are not present in a candidate, this is a serious counter-indication to hiring him for this specific function, unless we have indications that it will take an acceptable amount of training and coaching for the person to use the desired patterns.

Training: Transfer of excellence

While training, we want to install similar patterns in the students as the patterns shown by the model. The contrast analysis teaches us which patterns have to be included in the training material. Often, the person already uses the patterns we want to install, only in other contexts than the work context we want him to use the patterns in. If this is the case, often exercises in which the patterns are used as a demonstration will be sufficient to switch on the patterns. Sometimes, some beliefs of the person will show up and help us find what stops him from using the patterns. If beliefs form a blockade, the severity of the blockade will help predict the chances this person has to become successful in the context.


Here we want to adapt the person so that he shows more of the effective patterns, in which we focus on his weak spots, compared to model. In the coaching process, we use the personís language and behavior structure as a "manual" on how to adapt the manager so that he will get the best results with this particular employee. If the coaching requires getting through some serious blockades, the line between coaching and therapy becomes a very thin one. In these cases, because of the boundary between professional and personal life, we would suggest the person to find therapeutic help or find a job better adapted to the patterns he is comfortable with.

Also, while managing a person, we check if the performance parameters, as well as the other criteria used to evaluate the person are compatible with the patterns we want to see in the person's behavior. For instance, if I am managing an airline, and I want a co-pilot to correct the mistakes a pilot makes, I'd better know that the typical line of command makes it difficult or the co-pilot to give "negative feedback" to the pilot, unless I add a specific evaluation instrument that judges the co-pilot on the detection of a number of "deliberate" errors the pilot was instructed to make.

Some companies will give the function of coach to another person than the manager. This often makes it easier to get through some barrier during the coaching process. I do advise that this kind of coaching is done with the same kind of ethical code one expects from a therapist.

Typical Approach

A typical project starts by identifying the context where the excellence is needed, together with the criteria used to decide upon the excellence of the behavior. The criteria of excellence play the role of an objective measurement instrument. This means that we often have to improve the criteria if objectivity is not sufficiently guaranteed.

Based on this information, we proceed by selecting one or several experts as well as a group of counter-examples. Once this is done, we can start the actual modeling of the excellent language and behavior patterns. Modeling gives the best results if we can observe the experts while they are active in the context to be modeled. We often observe a difference between the patterns of a person talking about a context and the same person in the context. When life observation is impractical or too expensive, we simulate the context or we try to get information that is quite accurate involving Ericksonian techniques. To increase the accuracy of the modeling, we want to get 3 successful cases and 3 unsuccessful cases from each subject participating in the modeling.

Use the results

We can use the results of a modeling project in several area's of the human resources field, such as recruitment, training and management, but its effect will be the best if we combine the applications in several area's. In this section, we describe how to apply the results of the modeling in 3 typical area's:

Work out profile & Use profile for selection

From the model of excellence, we know which structural parameters and patterns are important for getting results in the context. In stead of focussing on the experience of the person, we use these parameters and patterns as the "job-profile". Also, we can derive specific profiling tools. Especially if the number of candidates or the number of positions to fill is large, we can derive a specific questionnaire that will test for the profile. The persons in charge of the selection are trained to use the profile and the profiling tools.

Work out training material to pass on the knowledge of the model

Again we start from the model of excellence and the contrast analysis. The training design focuses on bridging the differences between the current behavior and the desired behavior. This is applicable to new hirees as well as to the average performer. The result of the training will be evaluated using the criteria of excellence, by measuring the criteria over a period before and after the training, thus given an objective measure for the result of the training.

Educate the manager and coach

We train the manager and the coach to use the model. Essentially, we teach the skills needed to detect the wanted patterns and how to get to them.

Example: Excellence in Programming

For this paper, we select an application of meta-programs (also called perceptual filters), one the current Neuro-Linguistic Models, to illustrate how it helps to make a good model for hiring, training and managing software programmers. Of course, as explained before, we will consider other parameters when building a complete model.

Note: what is the "best" programmer for a company, may differ based on the demands the company has about a programmer job-profile, and on how they rate the results. As such, what is a good programmer for one company may differ with the profile of a good programmer for another company. Also notice that the typical job content has its influence: does the programmer also do the analysis, does he/she do the testing, Ö ?

Modeling: find desired perceptual filters

Modeling will give the answer to the following questions: "How do the perceptual filters of the person work? How does a person address structure, context and process? How does he sort and attend to information?" Also, we will fund out which kinds of filters are the most advantageous for the job.

The way to find out which meta-programs are the best, is by testing some successful persons for their response to questions about these perceptual filters. From these answers, we can distinguish which filters are the most influential.

As an example, we selected 3 perceptual filters (from the 13 meta-programs to consider), demonstrate which kind of question to ask to get information about these filters, and explain what is the best value for this filter for a programmer.

Working style / Affiliation Filter

Question: Tell me about a satisfying work experience. What did you like about it?

Value "Independent": people that work best in situations were they get to work alone (good for computers, more difficult for management & sales). They like responsibility for their own work. (in work context: 20% population)

Working Organization / Preference Filter

Question: Tell me about a satisfying work experience. What did you like about it?

Value "Activity-Thing": you think that getting boxes out the door is your task. You organize your work tasks so that you work with things or ideas. (in work context: 55% population)

Stress Response / Basic Communication Mode

Question: Tell me about a work project that gave you trouble

Value "Thinking": responds to language not showing emotion, best for tasks not requiring emotion (in work context: 15% of population)

Use the model

Hiring : we set up questionnaire, that uses the same questions as described above, and looking for the same kind of answers: independent, things, thinking, ...

Management : the model is used as a "manual", whch means we explain the manager and coach which are the typical patterns of the employee at this moment, and the patterns we'd like to see more of. Also, the job is adapted to the expected patterns of a good employee. In this example, this means: independent work, limited people contact and performance evaluation based on "things".

Training : we transfer the missing pieces, so that the employees get closer to the patterns demonstrated to the exemplars. In this example, this training includes a course of time management, based on individual work, with time for planning. For the training in general, we put the focus on personal learning, putting the initiative with the programmer.


Better hiring:
Since the person hired has a profile that corresponds to the typical profile of an expert for that particular domain, we can expect that the person will have more job satisfaction and will show a better performance on the job. As a result, since the person is "the right man on the right place" we can also expect less staff turnover.
Better follow up:
Given that we can put together a kind of "user manual" in which we explain how to manage the person and how a manager can provide coaching adapted to the person, the follow up on the person and the way he obtains results will be better adapted to the task at hand. We avoid that the person is measured according to parameters that are not important for his function: e.g. it is not important to test a programmer for his ability to work in team, since he will typically be working alone.
Better training:
With the explicit model of expertise at hand, we can teach those tings that are said to be dependent of "experience". This is best done by comparing the current skill level of the trainees with the desired level, and by developing a course based on these differences.


In this paper, we have shown how the field of Neuro-Linguistics offers tools to help analyze behavior, make a model out of it and use it for recruitment, training and management. We illustrated this with an example which we restricted to the parameters of the Motivational and Working Traits.


NLP - The New technology of Achievement, Steve Andreas & Charles Faulkner, Eds., Quill, New York, 1994

Words That Change Minds - Mastering the Language of Influence, Shelle Rose Charvet, Kendall/Hunt Publishing Company, Dubeque, Iowa, 1995.

The Language and Behavior LAB Profile Manual - Hiring, Managing and selling for Peak Performance, Rodger C.Bailey, M.Sc, IEPublicaties, Nijmegen, 1996.

Models for Advanced Pattern Detection, Dr. John Grinder & Patrick Merlevede, M.Sc, paper to be published

About the Author

Patrick E. Merlevede studied at the Catholic University of Leuven. He has a background in Computer Science (Major within his Master studies of Applied Economics and Engineering) and obtained a Master of Science in Artificial Intelligence, Cognitive Science Option. Since then he obtained a Master and a Trainer certificate in Neuro-Linguistic programming from the NLP University (linked to the University of California, Santa Cruz).
After working as knowledge engineer he evolved more towards applied cognitive sciences. Since 2001 he is in charge of the network, and specializes in the area of Modeling Excellence and Emotional Intelligence.

Tieltsesteenweg 22 / B-9900 EEKLO / Belgium
Tel: +32(475) 87.08.52 / Fax: +32 (9) 378.48.88

Contact the author though or Go To the Author's HomePage