Joris Dormans (2004)
So far I have been discussing visual language from a theoretical point of view. I have build a theoretical model of visual language with the Modular Parallel Architecture (MPA) and a set of grammatical rules at its core. Although I have illustrated many of the discussed notions with examples, I hope to found this theory with a brief research I have conducted.
I have taken a collection of various posters, and for each poster I rated the occurrence and importance of the grammatical rules presented in chapter 4 (R7 R22), its relative complexity and the orientations of the four Modality Markers presented in chapter 3. This results in 21 rates for each posters. In addition, I have recorded the author, the country of origin, the year of creation and classified the genre of each poster. Together the data of each poster formed a record in a database table on which various statistical operations have been conducted. This has led to an interesting collection of data, covering 406 posters.
The research has two objectives. The first objective is to check for correlation between the grammatical rules. This objective serves as a falsification test for the grammatical rules. If many and strong correlations can be found between the grammatical rules then something is wrong with the theoretical frame work of my visual grammar or the way I worked out the rules themselves.
The second objective of this research is to establish a correlation between grammatical structures (or the usage of particular grammatical rules) and different genres of posters. If this correlation is found, than the visual grammar I have devised is a good tool to analyse the difference between visual languages (see section 3.5).
I will start by explaining the method and the rates used in this research in more detail. Then I will show how the method works by discussing two examples. After this I will discuss the results of the research and draw some conclusions from these.
The research was conducted on all the posters in the book A Century of Posters (M.F. Le Coultre & A.W. Purvis, 2002). This book presents the poster collection Martijn Le Coultre. He has collected posters for their power to represent ages past and by the way they can make people do things: vote for certain parties, buy that product, or see that film (ibid. 6). Many of the posters in this collection have been chosen for their effectiveness, instead of their artistic value.
For each poster I entered the title, artist, country of origin and year of creation in a database. In addition I recorded the genre of the poster (political, cultural or commercial) and subgenre. The latter is used to distinguish film posters from theatre posters that both are classified as cultural posters, or indicates the type of product being sold by commercial posters. In the appendixes you can find a complete list of the genres and subgenres I have encountered.
For each poster I also entered 21 ratings. 16 ratings are used to represent the use and importance of the various grammatical rules (R7 R22) discussed in chapter 4. The next 4 ratings cover the 4 modality markers discussed in chapter 3. The last rating is an index of the posters complexity. The rules ratings are on a scale from 0 to 3, where 0 means that the rule in question does not occur in the poster, 1 indicates the rule does occur but is of little importance, 2 indicates the rule is present and plays an important role, 3 indicates the rule is present and plays a crucial role. Having a scale of 4 possible values is rather unconventional, but I think also necessary in the light of this research. A value of 3 helps to distinguish the impact of grammatical rules in posters that depend on only a few (1 or 2) grammatical structures as opposed to the impact of grammatical rules in posters several (3 or more) structures have an important contribution to the meaning.
The modality markers are also rated form 0 to 3. Where 0 stands the lowest possible position on the accompanying scale and 3 to the highest possible position. To get a 'pure' natural coding orientation all markers should be rated 2 (see section 3.2 and figure 3.1). The complexity index corresponds to the number of high-level participants present in the poster and the complexity of the grammatical rules invoked by the image. Again results to a rating on a scale of 0-3. In this scale 0 indicates a very simple and straightforward picture; there is only few (three or less) participants and usually only one distinct grammatical patterns. A 1 indicates a rather simple picture with four or five primary or secondary participants and a few distinct grammatical patterns. Figures with six to eight participants and several grammatical patterns will have a rating of 2 on this index and anything that is more complex will have a rating of 3.
I choose to use a pretty rough scale because 'importance of a certain rule' can be quite subjective. A scale from 0 to 3 and the amount of posters processed (400) roughens out the possibility of error, giving more significance to the statistical data. Figure 5.1 shows data being recorded for every poster.
Figure 5.1 - Template record for each poster
Then these records were statistically processed. In order to test for significant differences between various groups of poster the Mann-Whitney Test was used. This test can be used to test for difference between ordinal data of two nominal groups. I have used different groupings (see below) and tested for differences between the group and the rest of the collection. Although I have used numeric ratings these must be interpreted as ordinal data. The distance between 'crucial' and 'important' (ratings 3 and 2) cannot be expressed by numeric means nor can it be compared to the difference between 'present' and 'important' (ratings 1 and 2). The Mann Whitney Test uses rank instead of the numerical data. The rank of all entries with a value of 0 can be calculated by the following formula (where n0 is the number of entries with value 0):
|R0=(||n0(n1+1)||) / n0|
The rank of all entries with value 1, 2 & 3 can be calculated with the following formulas:
|R1=(||(n0+n1)(n0+n1+1)||-||n0(n1+1)||) / n1|
|R2=(||(n0+n1+n2)(n0+n1+n2+1)||-||(n0+n1)(n0+n1+1)||) / n2|
|R3=(||(n0+n1+n2+n3)(n0+n1+n2+n3+1)||-||(n0+n1+n2)(n0+n1+n2+1)||) / n3|
When the average rank of the entire set is calculated the value is 203.5 for all ratings. This is because there are 406 posters in the collection.
When a group is examined the following formula is used:
|U = n1 n2 + (||n1(n1+1)||) - R1|
In this formula n1 is the number of members of the group, n2 is number of non-members, and R1 is the sum of the ranks of the group members. The value U is used to determine the probability p that the difference in ranking is significant. When the R1 is higher than average the deviation is said to be significant when p>=0.95 and when R1 is higher p<=0.05 indicates a significant result (Baarda & De Goede, ????: 223-225). I used an online computer procedure to calculate the p-values. Finally, to produce the bar charts used in this chapter to present the results of the research I used a value d that represents the deviation from average of a certain rating of a group, in effect normalizing the rank sum of a group. This value is calculated using the following formula (in which R is the sum of the rankings and n the number of members of the group):
|n * 203.5|
In addition to the Mann-Whitney Test to determine the differences between various groups of posters I have also looked at the occurrence of the grammatical structures. This procedure of analysis can only be used for the grammatical rules, not for complexity and modality. The occurrence of grammatical rules is represented using two numbers for all grammatical rules. The first number is the average rating given for that rule, the second number is the percentage of posters that is given a rating of 1 or higher for that rule.
Finally, I have calculated to correlation between the ratings. This was done using the rank-correlation method of Spearman. Again this method uses ranks instead of the original scores. These ranks were calculated in the same way as the Mann-Whitney test. The square of the difference in ranks of a data-pair (d2i) was than used in the following formula (in which n is the number of pairs, e.g. 406):
|rs = 1 -||6 ∑d2i|
|n3 - n|
The result rs is a number between 1 and 1. A result of 1 indicates a true correlation, 0 indicates no correlation and 1 indicates negative correlation. For the number of pairs used (n) a value of larger than 0.3 or smaller than 0.3 is said to be significant with a probability of 95% or more (Baarda & De Goede, 1995: 228-230).
In the text below I have used mainly visual representations of the statistical data for the Mann-Whitney test, and occasionally other data as well. The numeric data behind these graphs can be found in the appendixes with this thesis. In addition, the records of all posters can be found in the appendixes. These records are represented in the same order of occurrence as the poster in the book A Century of Posters.
What follows are two examples of the posters processed, chosen to show how the rating worked.
Figure 5.2 is the poster described by the ratings given in figure 5.3. The poster is classified as a commercial poster with 'transport' as its subgenre. In the studied posters there are several posters of airline companies and posters aimed to invite people to travel to certain countries that all fall within this genre and sub genre. Quality of articulation is the most important aspect of this poster. The fluidness of the air and the water in the poster gives the scene an almost magical character. This is effect is largely due to the sharp articulation of the rocks and the lack of articulation of the air and water. For this reason one can say that the Rule of Intuitive Correspondence applies, although it does not play a very important role. The grouping of the air and water due to this effect is more important. For these reasons R7, R8 and R9 are rated 1, 3 and 2 respectively.
Figure 5.2 - Swissair (H. Leupin 1966)
Figure 5.3 - Ratings of Swissair
The only vector in the poster is from the man in his boat towards the airplane. This vector is given substance by the waving arm. This vector is both an action vector (waving or pointing) and reaction vector (the representation of the man's head suggests it is looking upwards). The vector is weak and plays only a minor role in the picture. Therefore both R10 and R12 are rated 1. A compulsion (R11) is not found in this picture.
The man in his boat is embedded within the lake and the plane is embedded within the sky (R13). This strengthens the grammatical symmetry between the man and the plane (R15). Both these rules are not very strong in this poster. Although the vertical relation these rules help to set up are important (R18). The other syntactic rules (R14, R16, R17, R19, R20) are almost absent in the poster.
In this poster Lexical Iconography and Lexical Grouping are also absent. However, Lexical Connotation is an important factor in this poster as the magical quality set up by the articulation affects the lexical meaning of the poster. The poster is also not very complex. There are only a few participants which are clearly set apart. The articulation is the most important aspect of the poster. The syntactical and lexical structures of importance draw on this articulation.
Finally, the modality of the poster differs from the natural coding orientation. The blue hues of the sky and water are far to bright; the colours are 'more than real'. On the one hand, colour saturation (M1) is therefore rated 3. On the other hand, depth, colour modulation and contextualisation (M2-4) are 'less than real'.
Figure 5.4 - Potemkin (G.A.W. Rudeman 1926)
Figure 5.5 - Ratings of Potemkin
Potemkin (figure 5.4) is a Dutch film poster from 1926. Its ratings are represented in figure 5.5. Like all film posters in the collection I have classified it as a cultural poster. Articulation is again an important feature of this poster. The hard and angled line underline the violent scene that is represented. This makes Quality of Articulation (R8) an crucial feature of this poster as well. However, in contrast with the Swissair poster, Intuitive Correspondence (R7) plays a more important role: the articulation of the lying man are more round than the straight lines of the background and the soldier. This sets up an important contradiction between the soldier and the lying man, in which the different poles of meaning (violent aggressor versus humane victim) can be corresponded to the used means of articulation (angular versus round).
The vectors and thus the representation of action and reactions are also stronger in the this poster than in the Swissair poster. The vectors formed by the gun, the legs and the eye-lines are more prominent. But again this poster also does not show a compulsion structure.
Although this poster has a far greater narrative quality it does feature one important conceptual structure: Attribution by Proximity (R14). The gun and the red background, which might be interpreted as flames, are attributes that stress the violence of the soldier.
In the composition of this poster, only vertical relations (R18) can be said to play a role. Other compositional structures are absent.
The represented scene can be said to be prototypical. The soldier is not represented as a particular soldier but can be interpreted as icon for all aggressors. Likewise the man on the ground is not a particular victim and is also a universal victim. This is an important aspect of the poster and therefore it has a rating of 3 on Lexical Iconicity (R20). The poster also scores 2 on the rating for Lexical Connotation (R21) as the articulation strongly connotes the violence of the depicted scene.
The modality of this poster scores lower than the natural coding orientation on all scales. Most significant is the absence of depth in this poster: it is unclear if the soldier is standing on top of or behind the man on the ground.
In this section I will discuss the data on an overall level. It does include some statistics, but the real analysis will occur in the next sections (5.4 & 5.5). There are 406 posters in the data I collected. Of these 198 are classified as commercial posters, 135 cultural posters and 62 political posters. I failed to classify 11 posters, mostly because I could not understand the language of the text and their intention did not become clear in another way (see figure 5.6).
Figure 5.6 - Posters by genre
Amongst the political posters are posters that advertise a political party, posters used by peace protesters, posters used for fundraising for several welfare and environmental organisations, and posters aimed at addressing unwanted social behaviour, etcetera. The political posters are further divided in to three groups: political progressive posters, communist posters and posters made by the 'establishment'. Of these the progressive group is the largest with 32 members. Posters are considered progressive when they advocate progressive political ideals, such as socialism, environmentalism, peace and other issues that can be considered 'leftwing'. The communist posters, only 9 in total, are left out of this group as they are visually very different from the progressive group (see section 5.4, below). The 'establishment' posters are posters made by the government, rightwing or conservative parties, and welfare institutions that can not be considered progressive (such as the American National Foundation For Infantile Paralysis, Inc, a group that fight polio). Only 1 poster could not be grouped under these three classifications. This is a poster of Hans Richter made in 1919. I was unsure whether it should be considered a progressive or establishment poster as its visuals suggest the first group while its message suggests the second group.
The cultural posters fall apart into three distinct groups: film posters, theatre posters and posters for exhibitions. The first group is self-evident and contains 47 posters. The theatre posters advertise stage-performances, this can be theatre but also dance, ballet or musical performances. In total there are 49 theatre posters in this research. The last group contains posters for mostly art-exhibitions in museums but also other kinds of cultural exhibitions. There are 26 of such posters. This leaves 13 cultural posters that are not part of one of these other groups. This includes architecture posters and posters that advertise cultural discussions, amongst other things.
The large group of commercial posters is also the most diversified group. I grouped the posters further by type of product or service these posters advertise. In the analysis below I have only used the 5 largest groups amongst these. These groups are: Alcohol/Tobacco (27 posters); Travel (22 posters, mostly posters that advertise countries or places as nice destinations for tourists); Transport (18 posters, including airlines, public transport and taxi companies); Confection (16 posters, including clothes, shoes and other items of fashion); Food (16 posters). Other groups are Electronics; Household Products; Newspapers & Magazines; Office Equipment; Petrol; and Vehicles.
Most of the posters come from western countries, with over half of the posters coming from Germany, The Netherlands and France (see figure 5.7). Only 57 (or 14%) posters do not come from Europe, of this group the largest part comes from the United States. Posters coming from a certain country do not always form a heterogeneous group. The posters from France, for example, mainly come from the earlier periods, while the posters from Switzerland for a large part were made during the 1930s and the 1950s. This distribution does to some extend reflect the real situation. Posters from The Netherlands and Switzerland seem to be over-represented considering these countries' size and population. But these countries also have a well-established tradition in visual design with a high output of quality posters. Therefore the high number of posters from these counties is not a significant deviation from the actual situation. On the other hand, Posters from outside Europe, and most notably from the Far East, are under-represented. Therefore this collection bests reflects the European situation. Likewise, the large number of early posters of France reflects that the French design tradition peaked during this period.
Figure 5.7 - Country of origin
Figure 5.8 - Country of Year of creation
The year of creation of the posters is grouped into four periods (see figure 5.8). The earliest period stretches from 1805 to 1918, although there are only three posters dated before 1890. The second period stretches from 1919 to 1945 and contains the most posters (131) although it is only 1 poster more than the third period from 1946-1967. The last period from 1968-1982 was taken apart from the third period because the visual style of posters started to change tremendously from the late 1960s onwards.
Figure 5.9 - Average ratings
When we look at the average scores of the ratings of the whole set of data (se figure 5.9) the following observations can be made. On the one hand, quality of articulation (R8) and grouping by articulation (R9) are used pretty often in posters, also the action structure (R10) and the use of lexical iconicity (R20) are frequent.. On the other hand, conceptual structures (R13 R16), and most notably explicit subordination (R16) is used less frequently. The same goes for intuitive correspondence(R7) and reaction structures (R15).
All in all these result are easily explained by the nature of these posters. Most posters try to convey a relatively compact message; a poster is designed to communicate its message in a short time, as most people will look at it only briefly. The action structure is simple structure that is easily understood by many. The use of lexical iconicity (especially in political posters) is also a quick way to represent meaning. That is not to say that these posters are not rich in structure nor meaning. Quality of articulation, attribution by proximity (R14) and lexical connotation (R21) are all frequently encountered in these posters and can all be considered to set up connotations.
Conceptual structures such as explicit subordination (R16), but also lexical grouping (R22) and to some extend intuitive correspondence (R7) are encountered less frequent in this set of posters. That was also to be expected. These structures are more at home in diagrams and information graphics, although they are used more extensively in Communist posters (as we will see below).
Figure 5.10 - Occurence and average ratings
When we look at both the occurrence and average scores of the grammatical rules (figure 5.10) it becomes possible to say something about the importance of the different structures when used. If the bar representing average score is relatively long compared to the bar representing occurrence then that structure is used strongly when used at all. This effect is most noticeable for syntactic grouping (R15), explicit subordination (R16) and lexical iconicity (R20). The first two have to be explained in the light of the discussion above. Both syntactic grouping and explicit subordination are more at home in diagrams, but when they are used in posters they tend to play an important role in the design. As for the strength of lexical iconicity, the effect here can be attributed to the cultural strength of icons and symbols in general.
Figure 5.11 shows the correlation between the various ratings. The Spearman method I used dictates that for a set of 30 or more pairs a correlation of 0.306 or more is significant with probability of 95%, while a correlation of 0.432 is significant with a probability of 99%. However, a correlation of 0.45 is still weaker than a correlation of 0.60.
Figure 5.11 - Correlation diagram
R16, the rule of explicit subordination, shows the strongest correlation with other rules. This is probably largely due to the fact that explicit subordination is also relatively rare. The same goes for R22 (lexical grouping): it was encountered only infrequently but shows strong correlation with all other ratings.
The strongest correlation (0.71) is between explicit subordination (R16) and embedding (R13). This is explained by the fact that embedding is almost a prerequisite for explicit subordination. This also explains the strong correlation between explicit subordination and syntactic grouping (R15) and lexical grouping (R22). All of these rules are frequently encountered in unison as they all contribute to setting up hierarchical structures (such as Kress and Van Leeuwen's covert taxonomy, see section 4.4).
Likewise, the correlation between grouping by articulation (R9) and R15, R16 and R22 should also be high, as I consider R9 also to be part of this group. However, this is not true. A correlation between these was found but is not very high: 0.37, 0.44 and 0.43 respectively. The only explanation I have for this result is that grouping by articulation (R9) was found more frequently in the posters than the others of this group.
Another strange result is the fairly high correlation between compulsion (R11) and R15, R16 and R22: 0.52, 0.62 and 0.63. I have no good explanation for this result, although I can write it off as a result of the low frequency in which compulsion was encountered.
Otherwise I am quite happy with the correlations as they stand. A little correlation between most rules do occur, but there is not one rule that seems to overlap with another, and all rules can be and are shown to have been used independently of each other. Also, all correlations are positive: the rules do not seem to exclude each other. This is an important result in the light of the theoretical framework in which convergence plays an important role.
On a whole I think it is safe to conclude that the visual grammar passed this test.
The second objective of this research is to determine if there are groups of posters that make use of grammatical structures differently. To determine if there are different visual languages that are employed by different genres of posters, so to speak.
Figure 5.12 - Ratings by genre
Figure 5.12 shows the results of the Mann-Whitney test for the three main classifications of the posters: commercial, cultural and political. The bars represent the deviation from the average ratings. When a bar is grey, this indicates a significant deviation. As can be seen in figure 5.12 there are some significant differences between the three groups. Commercial posters tend to shun narrative structures other than the action structure (R10), while this is the opposite for cultural posters. Cultural posters also make much more use of embedding (R13) and a centre hierarchy (R19). The use of lexical iconicity (R20) is infrequent in commercial poster but frequent in political posters. Commercial posters tend to be rather simple while cultural posters are more complex (C). Finally commercial posters score higher on the modality ratings, indicating these posters more closely resembles the natural coding orientation, as overall modality of these posters was rather low.
Although these groups do differ, they do not differ by much. The question is, of course, how homogeneous are these groups really? In order to answer that question we will need to look at some subgroups. Figure 5.13 shows the deviation for the cultural subgroups. Of these groups, the exhibition posters differ the strongest, even though it is the smallest group and therefore needs the strongest deviation from the norm for that deviation to be significant. Most of these posters are advertisements for modern art exhibitions. The visual style these adopted is as a matter of fact the visual style of modern art itself. Even though these still retain certain aspects that are typical for posters. This leads me to expect that the scores for this group of posters resembles the scores for modern paintings themselves, had they been included in this research.
Figure 5.13 - Ratings by subgenre (cultural)
An interesting observation can be made from figure 5.13: film posters seem to deviate from the norm more strongly than theatre posters. For a reader of the 21st century that may seem a little bit odd. The explanation is that the film posters in this collection mostly come from a period (1920-1960) in which film was less established as popular art. Film posters of that time represented that, and are in many ways more progressive than theatre posters of that same period.
Figure 5.14 - Ratings by subgenre (political)
Figure 5.14 shows the deviation for the political subgroups. Of these groups the Communist posters stand out most significantly. Again this is due to the fact that this group is visually the most homogeneous. Also, the posters in this group are almost all from the USSR and most were made around 1932. Interestingly, what sets Communist posters apart is their high use of conceptual means. Often this was to show that all people with their cultural differences rallied behind the Communist cause. I rarely found this practice in socialist posters which ideals are similar.
Progressive political posters are usually thought to be more progressive visually as well. I expected that visually progressive posters would also score differently. Figure 5.14 shows that this is not really the case. Progressive political posters are simpler than others: low complexity (C), little use of colour or context (MS and MC) and strong use of lexical iconicity (R20). Otherwise they do not differ much from other posters in use of grammatical structures.
Figure 5.15 - Ratings by subgenre (commercial)
Unlike the political and cultural subgroups, the subgroups of the commercial posters are less distinct, with travel posters being the exception (see figure 5.15). Travel posters score very high on quality of articulation (R8), vertical relations (R18), lexical connotation (R21) and the last three modality ratings. Apart from the vertical relations these high scores can be related: travel posters aim to represent a location as wonderful and exotic. But on a whole, I conclude that commercial posters are more conventional, in the sense that they all use the same techniques to communicate their message.
Figure 5.16 - Ratings by period
The type of poster is not the only way of classifying the posters. Figure 5.16 shows the deviation of the posters grouped by period. In this figure we can clearly see that the average poster has changed over time (although in the last period there are many political posters and this influences the scores for this period). For example the quality of articulation (R8) was very important in the early period but over time lost significance. This can be explained by the more common use of photographs in the later periods. Photographs are usually not very strongly articulated. Posters have become less complex over time as is indicated by a general fall of usage of most grammatical structures.
Figure 5.17 - Ratings by country of origin
Another grouping is shown in figure 5.17. Here the posters are grouped by country of origin. Comparison between these countries is made a little difficult because for some countries there is a strong correspondence with the periods. Posters from France, for example come mostly from the first period, while the Swiss posters come mostly from the second and third period. Still a few things stand out from this picture. The little use of narrative structures in German posters, and the frequent use of these structures in American posters is one example. The lack of use of lexical iconicity in Dutch posters is another example.
Can we conclude on the basis of this research that we have found distinct differences between the various groups of posters, and by extension, that the grammatical rules are thus a good way of analysing visual representations? I feel the answer to that question is not a pronounced 'yes', but not a 'no' either. The question also deserves a longer answer.
From the results of the research I do feel that some groups are quite distinct, most notably Communist posters and posters for cultural exhibitions. The method seems to work quite well for these groups as the differences between these posters and the rest could for a large part be attributed to different use of grammatical structures. On the other hand, for quite a lot of other groups the same claim cannot be made. This might be explained by the fact that these groups were not very homogeneous. But in fact I do think the 'visual languages' of these groups is simply not as pronounced as the visual language of Communism or modern art. It would be interesting to continue this research and add to the collection more data of communist posters and modern art itself. More so, it would also be interesting to try and find posters of other distinct groups that have been left out such as oriental posters. Also, these days there are groups of posters that have grown more distinct than they were in the period from which the posters in this collection have been taken. I feel that contemporary film and theatre posters have changed tremendously. This might lead to more pronounced results than this research have given me so far.
On the other hand, modality seems to be a good indicator of differences between the groups discussed above. Better in fact than the grammatical rules. This might lead to the conclusion that the grammatical rules are not very useful tools for the analysis of visual languages, but I think that conclusion is unjustified. There are some differences between the grammatical structures of the groups of posters even though they are not very pronounced.
I do not wish to claim, however, that the grammatical structures as presented are here are not up to review. As a matter of fact, when conducting the research I noticed structures that are different from the once I used. For example, a structure which I dub 'superimposure' for now frequently appeared in the posters but I hade no rating to represent this structure. Figure 5.18 is an example of what I call superimposure. The female legs can be seen as self-evident, with their own connotations. At the same time they form a swastika. These two meanings are very different but are superimposed on the same participant.
Figure 5.18 - Kabaret (Wiktor Gürka)
Also, I frequently encountered posters that had different modalities. For example a poster that combined abstract figures configured in strong modulated colours with a unmodulated black and white photograph. The modality of the photograph is such case is very different from the modality of the rest of the poster, but I had no good means of recording these different modalities.
To cut a long story short, I do think this research has met with limited success. Some differences have been shown and a few group of posters stand out distinctly. These can be said to have their own grammatical 'fingerprint' and this is in itself an interesting result. Further research is required to bolster these claims. Ideally further research will include groups of posters that have left out this research. Contemporary posters, for example, seem to be much more diversified and form therefore an interesting addition to this research. As are posters from oriental cultures as these are often be said to different from their western counterparts as well. I will discuss what can be concluded for the grammatical structures themselves in conclusion of this thesis, which starts at the next page.
 Actually, almost all posters are part of the research. I left out around 3 posters. One poster was damaged, while I could not understand the Hungarian and Japanese texts of two others, I left these out because the intention of these poster remained unclear to me.
 This procedure can be found at: http://home.clara.net/sisa/ordinal.htm
 With the exception of course of R13 and R16 as was discussed above.
 It is worth to notice that the deviation from average needs to be stronger for smaller sets of posters before it is considered to be significant than with larger sets of posters.
 And one might claim that in a way modern art adopted some of the stylistics of posters.