II. RESULTS
2.3 Introductory comments on Russian colour terms
2.3.2 The Russian sorting task
The sorting task has several more aspects which may have influenced the results.
This makes it important when analysing the results to pinpoint which aspect affected the results the most. In some cases, an educated guess is needed as to which aspects of the sorting task matter most, whether stimuli selection, non-restricting instructions to the participants, non-verbal grouping before naming the groups, or something else.
The instructions for the sorting task make no mention of colour, and the participants are simply instructed to group the stimuli by similarity. The implicit expectation is that the participant will sort the stimuli by colour, but it is not explicitly stated that the sorting criterion for the stimuli is colour. After the list task, which primes the participants colour vocabulary, it is quite unlikely that the participant would not comply with the implicit instruction to sort by colour, but the decision on the criteria for sorting is left to the participant, whether to use the darkness or lightness of the shade for example, the feeling evoked by the stimuli, the name of the colour category that takes in the most stimuli, or another angle.
To orient the reader in the sorting task, a simple plot of the stimuli labelled with their most frequent group names was constructed. Since the Estonian and Russian data were collected using the same methods and stimuli, it is suggested that Figure 6 and Figure 13 be viewed side by side, as they depict the stimuli in the CIELAB colour space labelled with their most frequent names. All the figures for the Estonian and Russian plots are comparable in this way.
Figure 13 depicts the non-yellow stimuli in CIELAB colour space labelled with the most frequent group names. The most frequent stimuli names are sinij
‘blue’, žëltyj ‘yellow’ (not depicted), fioletovyj ‘purple’ and goluboj ‘light blue’.
Although this is a simple figure of the most frequent group names, comparison to the Estonian (see Figure 6), where there was only one stimulus where the group name helesinine ‘light blue’ was dominant, shows that the corresponding group name in Russian goluboj ‘light blue’ is in a category of its own.
Figure 13. Russian sorting task by most frequent stimulus name.
Most frequent stimuli names labelled: F – fioletovyj ‘purple’, G – goluboj ‘light blue’, S – sinij ‘blue’, Z – zelënyj ‘green’.
In Figure 13 the size of the dots, which represent the stimuli, is correlated with the frequency of naming, so the larger the dot, the higher the frequency. In Figure 13 ‘G’ marks the stimuli that were most frequently named goluboj ‘light blue’. Here is where a comparison with the corresponding Estonian plot, Figure 6, is invaluable. In the Estonian data (see Figure 6), helesinine ‘light blue’ is a not a frequent stimulus name or named most frequent, whereas in the Russian data (see Figure 13), the corresponding goluboj ‘light blue’ is much more visible.
Table 11 presents the data from the Russian sorting task in table form. The most prominent aspect in Table 11 is the most frequent names given to groups by participants. Sinij ‘blue’ was given as a name to a group of stimuli by the highest number of participants (10) and with the highest frequency (F=124), and it covered the largest range of stimuli (39) and had the largest average group size (12.4). In second place, nine participants named a group žëltyj ‘yellow’, but the range of stimuli and the average group size were very low (4), which
●
demonstrate the unintended distinction of the distractor stimuli in the sorting task.
Relatively large numbers of participants also named groups fioletovyj ‘purple’
(N=8) and goluboj ‘light blue’ (N=6). While their range and average group size are similar, fioletovyj ‘violet’ has a higher frequency (F=80) than goluboj ‘light blue’ (F=68). However, in comparison to the frequencies of the following group names birûzovyj ‘turquoise’ (F=45), zelënyj ‘green’ (F=24), and temno-sinij
‘dark blue’ (F=18), which were each given by four participants, the frequency of goluboj ‘light blue’ is high.
Table 11. Russian sorting task by no of participants (N > 1) Term
(in Singular) Gloss No. of
participants Frequency No. of tiles (range)
Average group
size
sinij blue 10 124 39 12.4
žëltyj yellow 9 36 4 4
fioletovyj purple 8 80 22 10
goluboj ‘light blue’ 6 68 20 11.3
birûzovyj turquoise 4 45 32 11.3
zelënyj green 4 24 13 6
temno-sinij dark blue 4 18 16 4.5
temnyj dark 2 24 22 12
sirenevyj tuman lilac smoke 2 20 16 10
lilovyj lilac 2 18 17 9
grâzno-sinij dirty blue 2 8 6 4 svetlo-sinij light blue 2 12 10 6
Due to the low consensus, all the stimuli names given by more than one participant are represented in Table 11. This low consensus in the sorting task might be better called ‘high variability’ or ‘broad inter-participant differences’
in naming the sorted stimuli groups, and it is also reflected in the subsequent multidimensional scaling analysis plot. Ideally the multidimensional scaling plot should reveal separate groups. The following multidimensional scaling plot of the Russian sorting task shows the grouping of the stimuli in Figure 14.
It could be speculated that the four yellowish distractor stimuli exhibit an extremely strong isolating effect. It is possible that the ratio of experimental stimuli, with 51 non-yellow tiles, to distractors, with 4 yellow stimuli, is too small. They are so far separated from the rest of the stimuli that all the non-yellow stimuli form an elongated horseshoe-shaped even spread of stimuli, rather than distinct groups. Most importantly for the comparison, the Russian goluboj ‘light blue’ has a more defined partition in the MDS spread in Figure 14 with a slight separation visible from sinij ‘blue’ and from zelënyj ‘green’.
Figure 14. Russian sorting task multidimensional analysis.
Most frequent stimuli names labelled: F – fioletovyj ‘purple’, G – goluboj ‘light blue’, S – sinij ‘blue’, Z – zelënyj ‘green’.
The output of the principal components analysis suggests that the first component counts for 44.3% of the variance, the first two cumulatively for 70.1%, the first three for 79.5%, the first four for 84.3, the first five for 87.8%, and the first six for 90.4% of the variance. The scree plot has a very steep decline after the first two components, and starts to level out at the third component.
In the hierarchical cluster analysis of the Russian sorting task (Figure 15) there are three big clusters, one with mostly fioletovyj ‘purple’ most frequent names, one with almost all sinij ‘blue’, and one containing many stimuli that were most frequently sorted into the goluboj ‘light blue’ group. The hierarchical cluster analysis plot (see Figure 15) reveals that the four yellowish distractor stimuli may not be the only ones to exhibit isolating tendencies.
● ●
Figure 15. Russian sorting task hierarchical cluster analysis.
That goluboj ‘light blue’ forms a sub-cluster in Figure 15, in which all the stimuli are sorted and named under the term named most frequent (nmf), is significant in comparison to the Estonian sorting task, where only one stimulus had a helesinine most frequent name, rendering the possibility of a helesinine cluster or category rather unlikely.