Eva Heller’s research confirms genetic semantics

Ferre Alpaerts, 1993.


Is language genetically encoded? The starting point of the theory of genetic semantics is the hypothesis that there is a universal genetic code, that gives an inner meaning to words, to the sounds of words as well as to colours. To answer this question, Alpaerts puts the results of the research Wie Farben Wirken (1990) by German sociologist Eva Heller, on the effect of colours, together with his genetic Semantic Dictionary (1992). If his classification, built up according to the mathematics of the DNA, does not show universal structures but a random arrangement of words, there can be no connection with Heller’s research results. The expectation, on the other hand, is that Heller’s respondents will be more or less guided by this hereditary code when making a connection between words and colours. Alpaerts shows in this comparative study that the connection made by genetic semantics between phonetic shapes and colours is indeed retrieved in Heller’s research.


The research of German sociologist Eva Heller* towards the effect of colours, is generally seen as a standard work. The basis of this work is a comprehensive study with 1888 German interviewees from young to old. The results of the responses and preferences are based on figures and make this study of great importance to anyone who is professionally applying colour.

Heller’s survey was in writing and anonymous. Each interviewee linked 40 keywords to 11 colours. With different questionnaires, a total of 200 keywords and their link to colours were asked. The colours presented to the respondents were: blue, brown, yellow, gold, orange, rose, red, black, silver, purple and white.

In his Semantic Dictionary **, Alpaerts classifies 3,941 Dutch keywords according to 8 main divisions, further into 64 (8×8) subdivisions and finally in pages on which words are grouped with a highly related meaning. The coding used to make the semantic distributions is displayed with colours. Alpaerts used eight main colours: blue, yellow, green, red, black, white, purple and brown as the base distribution. It is further refined to 64 divisions in which the 8 main colours are combined, supplemented with 64 colour tones that display the same inner, codic meaning. Thus, a comparison with the results of Heller’s research is possible.

The Semantic Dictionary contains 140 of the 200 words Heller incorporates, with three kinds of colours that, according to genetic semantics, show the inner meaning:

  • H-Colours: 64 colour shades that include the colour space
  • T-colours: The top colours of 64 colour combinations
  • B-colours: The bottom colours of the 64 colour combinations

In Alpaerts’s paper the similarities and differences between the Semantic Dictionary and the Heller wordlist are calculated. From this, statistical conclusions are drawn to the degree of similarity between the two, and the significance level of this similarity. I.e. are these similarities due to accidental fluctuations, or should we suppose a natural law?

For the calculations, an Excell spreadsheet, mathematical tables and the tables of Wijvekate were used.


Alpaerts shows that the connection made by genetic semantics between phonetic forms and colours is retrieved in Heller’s research. Words that receive a particular colour code because of their vowel combination, often get the same colours assigned by the interviewees of Heller’s enquiry.

On average, 17.74% of Heller’s respondents choose a solution that is also found in the Semantic Dictionary. I.e. for H-, T- and B-colours, this is 5.24% more than the normal average. Statistically, with 142 samples, this percentage constrains satisfaction. Consequently, the connection can not be interpreted as the result of accidental fluctuations. The probability value is less than 0.03% according to the calculations.

In addition, when taking into account the fact that it are 142 x 3 tests, because every time H-, T- and B-colours are selected, and that for each of these series, the averages are about 5.24% higher than normal, then the probability value will be even smaller and random can be statistically ruled out.

The similarity between the Semantic Dictionary and Heller’s wordlist is unmistakable but does not seem to be impressive: 17.74% of the keywords have the same solution instead of the normal average of 12.5% that we can expect by chance. However, keep in mind that 100% is not feasible for each of the three correct colours. After recalculation and after adjustments of the first wordlist a 31% similarity is reached.

As is done with the Semantic Dictionary, Heller’s wordlist is adjusted by leaving the most dubious cases out. To this end we introduce a factor Z that indicates how sure the respondents themselves are of their choices. It appears that the higher Z is, that is, the more Heller’s interviewees agree on the most suitable colours for a particular word, the more similarities appear with the Semantic Dictionary. This linear relationship can be graphically represented.

Percent H-, T- and B-colours compared to certainty Z.
Percent H-, T- and B-colours compared to certainty Z.

When Heller’s respondents would agree completely about the choice of appropriate colours, we would reach about 70% of the achievable ranges. The 30% difference between calculated (Semantic Dictionary) and chosen colours (Heller’s wordlist) can be divided into several cases:

  • Statistical bias, eg because Heller and the Semantic Dictionary use different standards. Heller lets respondents choose between 11 colours, while genetic coding assumes a distribution of 8, 64, etc.
  • There may be important cultural differences between German speakers (Heller) and Dutch speakers (Alpaerts). Perhaps each language draws only certain related options out of the one universal semantics, such as is the case with the generative grammar and the structural phonology.
  • Semantics is a linguistic science, ie calculating the codic expression of a keyword sometimes resembles more like a rebus than a mathematical problem. Different ‘correct’ solutions are possible and we may retrieve that in colour associations. For example, “laundry” is associated as well with the colour code white-on-blue (the cleanliness of the finished product) as with black-on-blue (the traditional impurity of this profession).
  • Undoubtedly, the Semantic Dictionary contains a percentage of errors. By introducing a point system, the dictionary will be improved.
*For this study the Dutch translation by Emiel Stevens was used.
**The Semantic Dictionary of Alpaerts is at the heart of KHNUM, the creative database that goes online by autumn 2017.


Heller, E. Kleur: symboliek, psychologie, toepassingen. Utrecht, Het Spectrum. – III. – (Aula-boeken; 189). 1990.

N. N. redactie. Wiskundige Tafels, Utrecht, Het Spectrum. Prisma 1267. 1977

Wijvekate, M.L. Verklarende statistiek. Utrecht, Unieboek – Het Spectrum. (Aula-boeken). 1972.