Colour terms in three languages:their distribution and function

Colour terms in three languages: their distribution and function František Čermák (Prague) ABSTRACT Czech, English and Dutch colour adjectives are examined on the basis of InterCorp and other data to establish their equivalence where the very different numbers representing the three languages readily point to some major overlap in interlingual relations. The most conspicuous cases are discussed against the background of the typological features of these languages. There exist some familiar differences, such as the Czech červený-rudý corresponding to a single English equivalent, red, or the Dutch rood, although other, less familiar discrepancies may be found as well, such as the Dutch roos-roze for pink, růžový, occurring mostly in compounds. Out of the three basic sentence functions of adjectives which may be (1) attributiva tantum, (2) predicativa tantum, or have (3) non-specialized function, it is the last use (3) that is chosen for a further examination based on the available corpus data, as colour terms occur in all standard syntactic positions. However, only the predicative use of these colour adjectives is focused on as it appears to be special and not much research interest has been paid to it in general. Overall, the research did not produce a sufficient number of examples to allow detailed conclusions. KEYWORDS colour universals, parallel corpus, Czech, English, Dutch 1. COLOUR TERMS: INTRODUCTION If a comparison cannot be made directly between two objects, any other comparison is mostly viable only where a clear tertium comparationis (a third element) providing a platform of reference between the two (or more) objects is available. This tertium comparationis is external to the objects compared, in contrast to a direct comparison where one of the objects becomes such a (third) element of comparison. In the case of colour terms it is reasonable to leave aside the possibility of measuring the physical wavelength of a particular colour as it would not tell us much anyway, even though it is possible. Such information would be only general and impossible to relate to a specific human language perception of colour. In the case of colours or, rather, colour names, this tertium comparationis is then only a particular colour name in a particular language, nothing more as there is no external standard available. In languages, as we know, the situation is far from simple, offering few exact counterparts between them. We may recall the pioneering research by Berlin and Kay (1969) and the subsequent protracted discussion of the issue of possible colour name universals and their distribution over languages. The goal of the present study, a comparison of the basic colour terms in three languages, Czech, English and Dutch, based on the InterCorp, one of the largest non- FRANTIŠEK ČERMáK 137 Internet multilingual parallel corpora (Čermák, in print), may seem to be limited, as the number of language pairs (compared in both directions) is small and the number of texts is small too (three original texts with translations: Kundera for Czech, Adams and Rowling for English, and no original text for Dutch, only Dutch translations from Czech and English). Hence, the data used here is not quite balanced. Unfortunately, at the time of writing (spring 2012), there was no suitable Dutch original available in the InterCorp that would be translated into the other two languages. Still, if we take this study as a preliminary fact-finding probe, preceding larger-scale research, it might actually be seen as a good start, however modest. It is obvious that each of the three languages examined will, in turn, serve as a tertium comparationis. Typologically, it may be useful to note that each language displays different features, representing three dominant language types and their dominant traits, inflectional (Czech), isolating (English) and agglutinative (Dutch). The general functional framework for adjectives is simple, including those used either (a) attributively only, i.e. before nouns and related to them (attributiva tantum, e.g. very, upper, sheer), or (b) predicatively only, i.e. after a predicate verb (predicativa tantum, e.g. afraid of, fond of), or, finally, having (c) both functions and, therefore, not functionally specific (e.g. interesting, metallic); this group involves most adjectives. It will be interesting to find out where the colour terms stand in this framework. The main questions to be asked are: Is the number of basic colour terms the same in all three languages? What are the observable differences between the equivalents? What is the distribution of the colour terms in the three languages? The first and the second question will be answered in detail during the analysis of the results obtained, while a general outline of the distribution of the colour terms (the last question) will be offered in Section 2 below. However, a preliminary answer to the first two questions, based on common sense and experience, is obvious: the number of colour terms is not the same and there are, accordingly, many differences between the languages in colours and their equivalents. Any analysis of a language pair, based on translation, must allow for mismatches. They are either due to the translator’s mistake, oversight or unusual interpretation of the text when the translator chose to view it somewhat differently and diverge from, say, a standard dictionary equivalent and general expectation. The isolated status of such mismatches (some of which will be pointed out in the following) is indicated by their low frequency of use. Obviously, they can hardly be recognized as standard translation counterparts. By way of a warning it must be said that only those equivalents were inspected that were identified automatically by the PARK software used in the InterCorp, with 138 LINGUISTICA PRAGENSIA 2/2014 some manual input behind it, however. There is no way to know the margin of error caused by this procedure; in the following, it will be bracketed under the general label omission. 2. BASIC COLOUR TERMS AND THEIR COMPARISON It is an interesting task to compare a very small sample of a language in a rather welldefined area, such as the three texts and the colour terms used here and their translations or translation equivalents. It is even more interesting to observe their distribution following, basically, from the nature of the texts and the specific topics dealt with. In this respect the distribution is due to chance only. To see how different the distribution is from the perspective of an objective tertium comparationis of sorts, we will compare the text distribution of the colour terms using two basically objective and general standards. One is offered by the results of Berlin and Kay’s (1969) universal research study based on more than 100 languages. Their original finding was that there is a numeric scale of various colour terms in various languages starting with as few as two and ending with 11 and more, namely black, white, red, green, yellow, blue, brown, purple, pink, orange, grey This scale should, however, be read in an inclusive way. To demonstrate this, two signs will be used, namely a slash / (denoting possible alternatives, i.e. ‘or’, but also inclusion of both or all items) and an arrow ← (denoting inclusion on the scale, meaning that any term occurring on the scale to the left of the one chosen must obtain in the particular language as well). Thus black/white ← red ← green/yellow ← blue ← brown/purple/pink/orange/grey means that any language which has a name for blue must also have all of the colour terms to the left of blue on the scale. Despite lengthy discussions, this pattern of colour term occurrence in languages has been generally proved universal, while it is stressed that the occurrence of colour terms in individual languages is relative. This strongly goes against the primitive view that since people around the world have the same eyes they will (a) perceive the same colours, and (b) call them the same way. While (a) is indisputable, (b) is not true as the results of specific research studies covering languages ranging from Czech, Finnish to Australian languages demonstrate. The situation is highly variable and differs from language to language (see also Omniglot web-page www.omniglot.com/language/colours/index.php). The scale of English colour terms above is in fact general and representative of any language and so can serve as a standard. In Czech it corresponds to bílý, černý, červený, zelený, žlutý, modrý, hnědý, purpurový, růžový, oranžový and šedý, or FRANTIŠEK ČERMáK 139 bílý/ černý ← červený ← žlutý/ zelený ← modrý ← hnědý ← růžový/ purpurový/ oranžový / šedý. As we shall see later, there are many alternative names used for the same colour in one language, one example is the notorious problem of two names for red in Czech, červený and rudý. This kind of splitting the same small semantic space into two or more alternative (and basically synonymous) names is familiar from many languages (e.g. two words in Russian for the English blue, etc.). This is not to say that there are no other synonyms for red in Czech (in fact, there are at least ten to be found there). But the existence of the Czech červený and rudý is a different case. The second standard to be used for comparison is the frequency of these colour terms in a frequency dictionary. Comparing their frequencies with the data in the Czech frequency dictionary and recalculating them according to a balanced distribution by the ARF formula (average reduced frequency, see Čermák and Křen 2004; Čermák and Křen 2011), we get a very different picture, the frequencies being far from the same. The simple, traditional frequency is, basically, similar to that found for English (in Section 4 below) or, remotely, to the sequence found in the InterCorp data. Also their frequency scale has nothing in common with the universalist scale posited by Berlin and Kay, indicating, for example, that the terms black (černý) and white (bílý) are far from being top colour terms; in fact they are preceded by a number of other, more frequent terms. The frequency dictionary gives the following scale (ARF) růžový (2303), hnědý (2148), rudý (2243), žlutý (1532), modrý (927), zelený (798), červený (687), černý (398), bílý (383), šedý/šedivý (2534), fialový (00). It may be surprising that the name for pink (růžový) has the highest frequency of all; no data for violet (fialový) are due to the size of the dictionary used (the 2011 one used here has only 5000 lemmas, although it is based on a much larger dictionary from 2005, with 50 000 lemmas; both derive from a very large corpus of 100 million words). However, the obvious starting point for the general framework is English, because of its familiar status and offering at the same time a minimum of terms, with only ten basic language terms used. The adjectives inspected using a mutual interlingual comparison are (starting with English): black, white, green, red, yellow, blue, grey/gray, pink, brown, violet. Some considerations had to be applied to the choice of the basic colour terms and its limitations. Thus, the English term purple, one of the basic universals, is not included in the choice of colours used for the examination as in Czech it is split into several equivalents with a very low frequency (nachový, purpurový, brunátný, rudý; all of them were found in the InterCorp to translate the English purple, although none is dominant). Another decision to be made, to illustrate at least one more of the minor problems, concerned the word class membership problem. In English it is often a problem to distinguish a noun from an adjective: these cases of homonymy (homomorphy) had to have a limiting influence here. This is notoriously illustrated by orange which is both a noun (fruit) and adjective (colour), however they might be related. 140 LINGUISTICA PRAGENSIA 2/2014 After taking all these issues into consideration, the distribution of the selected universal colour terms in the Czech and English originals and their Czech, English and Dutch translation counterparts were investigated, correlated and quantified, and the results are presented in Table 1 below. English Czech Dutch Total black 97 černý 106 zwart 108 311 white 65 bílý 59 wit 86 210 green 57 zelený 58 groen 48 163 red 44 červený 28 rood 59 131 rudý 21 21 yellow 27 žlutý 30 geel 32 89 blue 25 modrý 26 blauw 28 79 grey/gray 23 šedý/šedivý 23 grijs 24 70 pink 21 růžový 17 roos 21 59 brown 15 hnědý 11 bruin 11 37 violet 4 fialový 5 paars 15 24 lila 3 3 378 384 435 1197 table 1: The distribution of colour terms (as the tertium comparationis) in the InterCorp texts In practical terms, although Czech is mainly a target language in two translations from English (Adams, Rowling), and in only one case a source language (Kundera), the Czech colour terms will be examined first with respect to the distribution and nature of their translation counterparts of colour terms, or rather their equivalents, in the other languages. 3. THE CZECH COLOUR TERMS AND THEIR COUNTERPARTS Owing to the equivalents occurring in the InterCorp, the original number of 10 colours in English increased to 11 in Czech. This is due to the overlapping use of two terms corresponding to red, namely červený and rudý in Czech (a distinction mentioned above that is lacking in English), which are not significantly different in frequency, however. On the other hand, two names for grey, namely šedý and šedivý, are just historical and stylistic variants of one term only. Although the frequency order of these adjectives is the sum total of the use by the three authors of the three texts only and does not tell us anything about the real general frequency, it may at least suggest certain preferences. Hence the final number of 11 terms, presented in decreasing frequency order that are used here, is the following (without any lemmatization): černý (106), bílý (59), zelený (58), žlutý (30), červený (28), modrý (26), šedý/ šedivý (23), rudý (21), růžový (17), hnědý (11), fialový (5). FRANTIŠEK ČERMáK 141 Let us first have an illustration of the first occurrences of bílý (white, w/Wit) where the correlation is so far rather regular (although the original PARK windows look different). Bílý-White-Wit (Czech-English-Dutch): first examples in context (the original search starting from Czech) 1-CZTo, co slyšel, se dalo přirovnat k pocitu člověka, jenž se dívá na černou siluetu dvou tváří, a najednou zjistí, že je to vlastně obrázek bílé svíčky. EN-He was experiencing the aural equivalent of looking at a picture of two black silhouetted faces and suddenly seeing it as a picture of a white candlestick. NL-Hij onderging het auditieve equivalent van het gevoel dat je krijgt als je naar een plaatje van twee zwarte silhouetten van gezichten kijkt , en je in de ruimte ertussen plotseling een witte kandelaar ziet. 2CZ- „To to ale bylo snadné,“ libuje si člověk, a protože ještě nemá dost, dokáže, že černé je bílé, a na nejbližším přechodu ho zajede auto. EN- “Oh, that was easy,” says Man, and for an encore goes on to prove that black is white and gets himself killed on the next zebra crossing. NL- “Dat was even simpel,” zegt de mens en als toegift bewijst hij dat zwart wit is en wordt overreden op het eerste het beste zebrapad. 3CZV kabině obdélníkového půdorysu převládala bílá. Velikostí se místnost blížila menší restauraci. EN-The cabin was mostly white, oblong, and about the size of a smallish restaurant. NL-Het vertrek was grotendeels in wit uitgevoerd , langwerpig en had de grootte van een klein restaurant. 4CZSeděla na gauči a upřeně se dívala na maličkou klícku, jejíž obsah byl jejím posledním a jediným pojítkem se Zemí— dvě bílé myšky. Prosadila tehdy, že jí Zafod dovolil vzít je s sebou. EN-She sat on a couch and stared at a small cage which contained her last and only links with Earth— two white mice that she had insisted Zaphod let her bring. NL-Ze zat op een bank en staarde naar een kleine kooi die het laatste en enige bevatte dat haar met de Aarde verbond— twee witte muizen. The standard view in the PARK software is vertical, allowing for as many columns as the screen can accommodate. Here, only three columns (Czech, English, and Dutch) for the same data are used to illustrate the output. intercorp_cs intercorp_en intercorp_nl To, co slyšel, se dalo přirovnat k pocitu člověka, jenž se dívá na černou siluetu dvou tváří, a najednou zjistí, že je to vlastně obrázek ==bílé== svíčky. He was experiencing the aural equivalent of looking at a picture of two black silhouetted faces and suddenly seeing it as a picture of a white candlestick. Hij onderging het auditieve equivalent van het gevoel dat je krijgt als je naar een plaatje van twee zwarte silhouetten van gezichten kijkt , en je in de ruimte ertussen plotseling een witte kandelaar ziet. 142 LINGUISTICA PRAGENSIA 2/2014 „To to ale bylo snadné,“ libuje si člověk, a protože ještě nemá dost, dokáže, že černé je ==bílé== , a na nejbližším přechodu ho zajede auto. “Oh, that was easy,” says Man, and for an encore goes on to prove that black is white and gets himself killed on the next zebra crossing. “Dat was even simpel,” zegt de mens en als toegift bewijst hij dat zwart wit is en wordt overreden op het eerste het beste zebrapad. V kabině obdélníkového půdorysu převládala ==bílá==. Velikostí se místnost blížila menší restauraci. The cabin was mostly white, oblong, and about the size of a smallish restaurant. Het vertrek was grotendeels in wit uitgevoerd , angwerpig en had de grootte van een klein restaurant. Seděla na gauči a upřeně se dívala na maličkou klícku, jejíž obsah byl jejím posledním a jediným pojítkem se Zemí— dvě ==bílé== myšky. Prosadila tehdy, že jí Zafod dovolil vzít je s sebou. She sat on a couch and stared at a small cage which contained her last and only links with Earth— two white mice that she had insisted Zaphod let her bring. Ze zat op een bank en staarde naar een kleine kooi die het laatste en enige bevatte dat haar met de Aarde verbond - twee witte muizen. table 2: The PARK software view of the data Studying adjectives does not require much context, as a rule, although it can also be easily enlarged in the software used. Here, a default automatic solution restricting it to one sentence is offered. The most frequently used Czech adjective černý (106, black) corresponds, surprisingly, to the English black in slightly over half of the uses (56×) only, some of which are compounds such as black-haired, the rest being different. However, dark as the closest equivalent next to black is used (31×); in addition, there have been six omissions. The expected intuitive correspondence one to one has thus not been found. The Czech-Dutch comparison seems to be only slightly better, yielding zwart/e (64×), the other equivalent was donker/e, although once duister was used, too; the rest are omissions. However, Dutch equivalents include, here and elsewhere, a number of compounds, such as zwartgallig (gepeins, i.e. thoughts), inktzwart (meer, i.e. lake), pikzwart (haar, i.e. hair, ballen, i.e. balls), gitzwarte (lichaam, i.e. dragon body). The 5 uses of compounds in Dutch are typical and do not generally correspond directly to the counterparts in the other languages. Rather, it is the translator’s clever interpretation of the original texts. Bílý (59, white), coming second in text frequency, has a rather safe equivalent in the English white although it was once left out and once the equivalent silver appeared. A notable exception, as expected, was found, here and elsewhere, in the translation of idioms and set phrases. Thus the Czech idiom za bílého dne is rendered as in (broad) daylight where no colour term is used. Almost the same holds for Dutch (wit/witte including the inflected form), where, however, 4 omissions were found and one unexpected equivalent bleek (i.e. pale), although that is also due to an idiom (collocation), i.e. Haar lippen waren bleek, translating Her lips were white. Similarly the Czech idiom za bílého dne mentioned above is translated differently, namely by op klaarlichte dag using no colour term. FRANTIŠEK ČERMáK 143 The third most frequent Czech adjective is zelený (58, green). It also has a rather safe equivalent in the English green which was used as a translation counterpart in almost all cases. The two infrequent exceptions are one omission and one alternative equivalent emerald. The Dutch translation, too, consistently used the expected equivalent pair, groen/ groene (the longer being the inflected form), except for two cases with no equivalent. Also, compounds such as donkergroen, lichtgroen, gifgroen, felgroen, smaragdgroen, etc., were frequently employed. Žlutý (30, yellow) has a significantly lower frequency of use, having, basically, always the same English equivalent yellow, with one exception where a compound is used, namely sandy-haired (2×). Dutch is similar in using geel/gele with only two omissions and, interestingly, one different option rossig/e appearing unexpectedly where the English has the compound sandy-haired. Červený (28, red) has its basic equivalent in the English red although it is left out three times or translated by scarlet (4 times). There is, however, one odd mismatch between the English black in the original and the Czech translation using červený, presumably due to an oversight on the part of the Czech translator. There is, however, a far greater problem in correspondence here, as has been signalled above, owing to the continuum of the English (and Dutch) red being split in Czech into two names, červený and rudý (see below). Hence, any comparison has to take into account both Czech terms corresponding, basically, to only one term in English and Dutch. The equivalence of the Dutch rood/rode is similar to that of the English red, showing the same červený-black mismatch and one omission. Modrý (26, blue) remarkably has the safest, one hundred percent correlation with the English blue, and the same holds for Dutch blauw/e. Šedý/šedivý (23, grey) has, likewise, almost a default equivalent in the English grey/gray (the former appears to be the more frequent in use in Britain, the latter is generally regarded as the US spelling variant). Interestingly, gray is more frequent than grey (14:9), which may be indicative of where the translation was published. There is also one omission. The Dutch counterparts are not so straightforward, however. The basic equivalent is grijs/grijze, but grauw is used twice alongside one typical compound staalachtig and one omission. Rudý (21, red), corresponding to the English red, is slightly less frequent than its competitor červený (see above); another equivalent, scarlet, is also found (2). The omissions of rudý are due to the non-occurrence of a colour term in the idioms used in the original English text which are translated by an equivalent using rudý in Czech, such as blush up to his ear (byl rudý až za ušima) or face like a gigantic beet, a special case being flushed faces. In Dutch, the basic equivalent is rood/rode although twice paars is used, too; there is one omission recorded. Again, due to the absence of a colour term in the original English, no equivalent colour term is used corresponding to the een reusachtige biet (gigantic beet), either. The remaining three colours are, due to their low frequency, rather marginal. 144 LINGUISTICA PRAGENSIA 2/2014 Růžový (17, pink) has always the English counterpart pink, although Dutch uses roos/roze and in one case blos. The variant rozen, found in compounds only, is homonymous and is related to roos (flower rose). Hnědý (11, brown) is brown in English, another found equivalent being maroon in two cases, while Dutch has bruin; the only occurrence of zwart is an odd mismatch. Finally, fialový (5, violet) corresponds to the English violet mostly, although also to mauve (4:2), while Dutch has lila and paars (3:2). 4. THE ENGLISH AND DUTCH COLOUR TERMS AND THEIR CZECH COUNTERPARTS Naturally, due to the uneven number of represented colour names and almost no one-to-one correspondence (i.e. strict colour-to-colour correlation), the Czech colour equivalents of the English and Dutch colour terms may not seem (a) to be the same numerically and (b) to be equally represented unlike in Section 3. While no systematic attempt will be made here to go into details as to (b), the answer for (a) is the different number of colour terms, i.e. 10 in English and 11 in Dutch, which has certain consequences. We will now reverse the starting point of view, which in the previous section was that of the Czech language. While there Czech was used as the (first) tertium comparationis, in the following the function of tertium comparationis will be performed by English and Dutch, respectively. Looking first at the same data starting from English, it may be surprising to find that the overall figures reflecting the number of occurrences are not the same, either. They are in descending order as follows black (97), white (65), green (57), red (44), yellow (27), blue (25), grey/gray (23), pink (21), brown (15), violet (4). Likewise, the figures for Dutch are different: zwart (108), wit (86), rood (59), groen (48), geel (32), blauw (28), grijs (24), roos (21), paars (15), bruin (11), lila (3). To offer at least some explanation for this, let us notice two factors. The first one, the existence of červený-rudý in Czech (both signifying red in English or rood in Dutch), has been mentioned above. On a closer look, the English and Dutch names should numerically be roughly the same but they are not. The discrepancy is to be sought in the number of other alternative colour names used in English, such as scarlet, purple and some omissions. The other factor is the special situation with the distribution of colour names that are covered in both Czech and English by a single term, namely fialový and violet. The situation is not unlike that of the two Czech equivalents of red. The Dutch language uses paars and lila here (found in a 15:3 ratio) and in some cases, also roos (which, however, basically corresponds to pink).