Semantic variation and semantic change in the color lexicon

2.1 Lexical sources of the main terms for white

According to my data, basic color terms for white are originally mainly coded by expressions of brightness, more rarely by structures describing lack of saturation or paleness, as well as by non-chromatic denominations of simplicity or plainness – which may overlap and be difficult to disentangle.

Brightness, for example, is expressed by the lexical sources of English white and of the cognate basic color terms for white of the Germanic languages, as in (2a), all of which derive from Proto-Germanic *hwīta- and ultimately from the PIE root *ḱwit- ‘shine’, cf. Old Church Slavonic svĭtěti (sę) ‘to shine’, Lithuanian švitė́ti ‘id.’, etc. The same root, added with other suffixes, underlies not only numerous expressions of brightness and whiteness in ancient Indo-Iranian, but also the basic color terms for white of many of their daughter languages, as shown in (2b). In particular, Avestan spaēta- ‘bright, white’ has the same morphological structure as the New Persian basic color term sefīd ‘white’. From Iranian, these terms have been borrowed in most modern Indic languages,^[6] as in Hindi, Kashmiri, and Gujarati saphed / safed, which is also the basic color term for white in these languages. Instead, the Old Indic cognate śvitrá- ‘whitish, white’ is the native source of the Panjabi basic color term ciṭṭā ‘white’, and Old Indic śvaitra- n. ‘white leprosy, vitiligo’ (a rarely attested derivate of śvitrá-) is the source of the Nepali basic color term seto ‘white’. Basic color terms such as Sinhala sudu ‘white’ and Dhivehi hudhu ‘id.’ independently attest the same semantic pattern, as they derive from Old Indic śuddhá- ‘cleansed, purified, clear, bright, white’ from the root śudh ‘to purify; be or become pure, cleansed, clear’, from PIE *ḱud ^h -.

A similar semantic pattern, from brightness to whiteness, is attested by several other terms for white of the IE languages, such as Hittite ḫarki- ‘bright, white’, etymologically related to the PIE name of silver (cf. Latin argentum); Ancient Greek leukós ‘bright, white, fair’, related to English light; Latin candidus ‘shining white’, related to Latin candēla which has been borrowed into English candle; Irish bán ‘bright, white’; Albanian bardhë, the basic color term for white in this language, and so on.

A semantic change from paleness or lack of saturation to whiteness can be seen in the development of the Marathi basic color term pã̄ḍhrā ‘white’, derived from Old Indic pā́ṇḍara- ‘pale, whitish-yellow, white’. The latter adjective is connected to Old Indic pāṇḍú-, also meaning ‘pale, whitish-yellow, white’ (a word with unclear etymology, probably borrowed from a non-Indic substrate). In the Mahābhārata, Pāṇḍu was so called because his mother became pale at the frightening sight of her husband during conception. A lexical source describing plainness or simplicity underlies the Bengali basic color term śādā ‘white’, which has been borrowed from Persian sāde ‘simple’, a term widely adopted in many languages of Central Asia and of the Middle East with non-chromatic values. In Turkish, sade ‘plain’ can be used to describe a kind of coffee without sugar.

The derivation of terms of white from expressions of paleness or plainness – be they inherited from previous stages of the languages or borrowed from other languages – is much rarer in my database than the semantic change from brightness to whiteness illustrated in (2). The typical situation observed in my data is that the basic color term for white of an IE language is related to expressions of brightness – adjectives and verbs of shining, as well as nouns of brilliant objects, such as light, silver or the moon – attested in numerous other IE languages.

As we try to identify the most common lexical sources of color terms, it is also important to identify which sources are uncommon or lacking – with two provisos. Firstly, as anticipated in Section 1, we speak of basic color terms, and not of all possible chromatic denominations. Otherwise, one may have the wrong impression that everything is equally possible, which is not the case.^[7] Secondly, when dealing with ancient languages, the apparent rarity or lack of structure may be due to the deficient transmission of ancient texts. Even the presence of a structure must be handled with care, since the texts we have available often represent the literary language and can only give partial insights on ancient everyday speech. In this case as well, we maintain that formulating one hypothesis, as imperfect as it may be, is always better than no hypothesis at all. Moreover, my database is not limited to ancient languages – it contains many languages from different stages and from all IE branches, as illustrated in Section 1. This, in my view, may provide more reliable results.

According to my data, the basic color terms for white in the IE languages are not commonly drawn from names of objects, plants, or animals. This may happen for marginal denominations of white, e.g., the Latin poetical adjective niveus ‘snow-white’, derived from the name of snow, Latin nix, nivis f. But it does not often occur for basic color terms for white. In my database, only two out of 70 IE languages analyzed (=3 % of the cases) present their basic color term for white with a nominal origin – Modern Greek áspros and Konkani dhavo. The former, borrowed from Latin, derives from the (elliptic) denomination of a silvery coin, the nummus asper, which was widely used in late antiquity in the Eastern Roman Empire. This matches the above-mentioned tendency to use names of shining metal as a lexical source for denominations of white. Konkani dhavo ‘white’ derives from Old Indic dhavá- m., first attested in the Atharva Veda as the name of a plant, a kind of axlewood (Grislea tomentosa or Anogeissus latifolia) which is native to South and Southeast Asia, and which is used for many purposes, from firewood to tanning, the production of varieties of gum, wild silk, etc. Continuations of Old Indic dhavá- are widely attested in Middle and New Indic names of this tree or similar plants, cf. Pali dhava-, Oriya dhaa, ḍhaū, ḍhaa, Hindi dhau, dhawā, Sinhala davu, etc. This means that the derivational pattern from the name of a plant to a basic color term of white is well possible, though still turns out to be not preferred. This rarity may be therefore interesting from a theoretical perspective.

In 96 % of the cases, instead (cf. Table 1), the IE basic color terms for white has a non-denominal formation. Their underlying roots rather represent the source of names of referents characterized by a bright, clear or white aspect.^[8] In addition to silver, we often have names of the moon, a light, a cloud, the white of the eye, white spots (as in leprosy), egg white, flour, etc. which are derived from a root of brightness or a main term for white. The Romance names of the moon, such as Spanish, Occitan, and Italian luna (from Latin lūna f. ‘moon’ and ultimately from PIE *le/owk-s-neh ₂ -), come from the same PIE root *lewk- as Ancient Greek leukós ‘white, bright’. The name of the moon in Old Indic (candrá- m.) and in Albanian (hënë) is taken from the same PIE root *(s)kend- as Latin candidus ‘shining white’. The root of the PIE stem *alb ^h o- (cf. Pinault 2022) which underlies Latin albus ‘(mat) white’ and Umbrian alfu ‘id.’ (acc.n.pl), as well as Romansh alf / alv and Romanian alb, gives rise to Latin albūmen, -inis n. ‘egg white’, Ancient Greek álphi, álphiton n. ‘flour’, Hittite alpa- c. ‘cloud’, etc.

The same occurs for names of plants or animals characterized by a white or clear color. The PIE root *ḱwit- of English white and its cognates in (2) also underlies English wheat (in this case, the Proto-Germanic color term *hwīta-/*hwitta- ‘white’ is the source of the Proto-Germanic noun *hwaitja-, from which we also have German Weizen m. ‘wheat’, Icelandic n. hveiti ‘id.’, etc.). The PIE root *b ^h reh ₁ ǵ- of Albanian bardhë ‘white’ gives rise to English birch and its numerous IE cognates (German Birke f. ‘birch’, Lithuanian béržas m. ‘id.’, Russian berëza f. ‘id.’, etc.). For animals, we especially find names for the swan and for the pigeon / dove. Latin albus ‘(mat) white’ derives from the same root as Russian lébed’ m. ‘swan’, Serbo-Croatian lȁbūd m. ‘id.’, Czech labut’ f. ‘id.’, as well as Germanic nouns such as Old High German albiz, elbiz m. ‘swan’. Ancient Greek kúknos m. ‘swan’ is derived from the same PIE root *ḱuk- ‘shine’ as Old Indic śukrá- ‘shining, bright, white’. Lithuanian balañdis m. ‘pigeon’ derives from the PIE root *b ^h elH- which forms the basic color terms for white of Lithuanian (báltas), of Latvian (bal̃ts), and of all Slavic languages (cf. Old Church Slavonic bělŭ, Bulgarian bjal, Serbo-Croatian bȉjel, Polish biały, Ukrainian bilyy, etc.). There are multiple further examples of this in my data.

2.2 Lexical sources of the main terms for black

According to my data, the lexical sources of basic color terms for black indicate darkness or dirtiness or have meanings of smoking, burning, or shining.

Darkness is the most common source in this case, whereby expressions generically meaning ‘dark, obscure’ later establish a meaning of ‘black’ as well as of other dark colors. An example of this can be seen in New Persian siyāh ‘black’ and in its Indo-Iranian cognates, which express various dark colors (3). In New Persian, siyāh shares the status of a basic color term for black with the most recent term meškī. From Iranian, this term has also been borrowed in Armenian (Classical Armenian seaw ‘black’, Modern Eastern Armenian sev ‘id.’).

A different lexical source but the same semantic pattern emerges in Pashto, where the basic color term for black tūr is derived from the PIE root of darkness *temH-, cf. Vedic támas- n. ‘darkness’, támisrā- f. ‘darkness, dark night’, tamasá- ‘dark’, Avestan tamaŋha- ‘id.’; Latin tenebrae f.pl. ‘darkness’; German finster ‘dark’, etc. Another example may be provided by the Modern Greek basic color term mávros ‘black’, a variant of Post-Classical Greek maurós ‘black’ and of a much more ancient form amaurós ‘dark’, attested since Homer and probably of non-IE origin. Cf. also Ancient Greek mauróō ‘to darken, make obscure’. The basic color terms for black of the Celtic languages, such as Irish dubh, Welsh and Breton du, present a similar semantic pattern, as their underlying PIE root *d ^h ub ^h - is also the basis of expressions of darkness such as Old Irish dobur ‘dark, obscure, unclean’.

A semantic change from dirtiness to blackness, sometimes overlapping with that from darkness to blackness, is quite common, as stains or dirtiness make an object dark or black. This is the semantic pattern of the basic color terms for black in all Germanic languages (except English), such as German schwarz, which is related to English swarthy, cf. (4). In English, swarthy has been displaced by black in the function of basic color term and only remains as a marginal expression of this color, mainly denoting skin color.

Another example of this semantic change can be seen in the Albanian basic color term for black zi, which is probably drawn from the same PIE root *g ^w ewh ₁ -d ^h - denoting excrement, dirt, or disgusting objects in different IE languages, as in German Kot m. ‘feces’, cf. also Sanskrit guváti ‘cacat’, gūtha- n. ‘feces, ordure’, Avestan gūþa- n. ‘id.’; Middle Welsh budyr ‘filthy, mean’; Old English cwéad n. ‘dirt’; Serbo-Croatian gȁd m. ‘loathing, nausea’, Russian gad m. ‘reptile, vermin’, etc.^[9] The hypothesis that the meaning of dirty is primary with respect to that of the black color is suggested by the fact that the former is much more widely attested in the analyzed languages. We can see another example of this in the development of the cognate basic color terms for black of Ancient Greek mélas and of Latvian mȩl̃ns, which are also related to expressions of stains and spots (cf. Ancient Greek molúnō ‘to defile, pollute, stain’) or of other dark colors (cf. Lithuanian mė́lynas, the basic color term for blue in this language) – all from the PIE root *melh ₂ -. Outside these two branches, this root emerges in Vedic mála- m./n. ‘dirt, filth, bodily excretion or secretion’, maliná- ‘dirty, filthy, impure’, málavat- ‘id.’, etc.

Expressions of smoking, burning, or shining can also bring about basic color terms for black, because what is burnt and carbonized acquires the aspect of a black color. A typical example of this semantic change is attested in Latin āter, which is cognate with several forms related to burning and fire in IE, as illustrated in (5).

Since the roots of burning and especially of shining were also common sources of basic color terms for white, cf. (2), the same lexical source may underlie basic color terms for both white and black. English black derives in fact from the same Proto-Germanic stem *blanka- which is the base of German blank ‘shiny’, and which has been widely borrowed into Romance basic color terms for white such as French and Catalan blanc (Old French blanc has been borrowed, in its turn, into English blank), Italian bianco (borrowed into Sardinian biánku), Spanish blanco, Portuguese branco, Catalan, Occitan and French blanc, etc. (Cf. also Haitian Creole blan from French. Owing to its spread, this formation has been adopted in the main term for white in Esperanto: blanka.) The Proto-Germanic stem *blanka- is also related to Proto-Germanic *blinkan, cf. German blinken ‘to flash, blink’, Dutch blinken ‘to shine’, etc. All these forms ultimately derive from the same root as Ancient Greek phlégō ‘to burn’ (tr. and intr.), Latin fulgeō ‘to flash, lighten, shine’, Tokharian B pälk- ‘shine, be highlighted’, etc.

Instead, no basic color term for black among the IE languages in my database has a denominal origin, from the noun of a black referent.^[10] Consider that nature is rich in black or dark animals, as well as in plants with black fruits. In principle, it is therefore possible that black is lexicalized as the color of X animal or plant. This occurs in fact for minor denominations of black in IE, e.g., the rare and technical Ancient Greek adjective korákinos ‘raven-black’ is transparently derived from the noun kórax, -akos m. ‘raven’. For basic color terms of black, however, we have the opposite pattern, that is, an X animal or plant is lexicalized as ‘the black one’. We will later see that the situation is different for other color terms. Basic color terms for black or their underlying roots in fact represent common lexical sources for names of black or dark plants and animals, as well as of concrete and abstract objects. Cf. English blackberry, blackbird, blackboard, blacklist, blackmail, blackout, etc. In Sinhala, from kalhu ‘black’, we have kalhu dhiri ‘plant Nigella damascena’, kalhu faana ‘fish whitespotted grouper’, kalhu kan’dhili ‘plant trailing eclipta’, kalhu landaa ‘bicolor parrotfish’, kalhubilamas ‘skipjack tuna’, kalhuhutthumeyvaa ‘fruit of the pond apple’, kalhukalhlha ‘zig zag plant’, etc. In the ancient IE languages, where compounds represent a somewhat marked strategy of word formation, names of objects, plants, or animals rather derive from the roots for black with various suffixes, e.g., Ancient Greek melainás f. ‘a black fish’, melánion n. ‘ink’; Latin ātrāmentum n. ‘ink’, nigella f. ‘a plant’, etc. Latin niger ‘(mat) black’, the source of the basic color terms for black in most Romance languages (cf. Italian nero, Spanish negro, Catalan and Occitan negre, French noir, Romansh nair, Romanian negru, Sardinian ni(gh)éḍḍu, the latter from Latin nigellus ‘blackish’), has an unclear etymology, but is certainly not denominal. Its stem rather presents an adjectival word formation with the suffix -ro- which is quite common for color terms in IE, similarly to the adjectival suffix -no-. The latter suffix occurs in the basic color terms for black of all Slavic languages (cf. Serbo-Croatian cȓn, Bulgarian čéren, Ukrainian čórnyj, Polish czarny, etc.), which go back to PIE *kr̥s-no-, with the same word formation as Old Prussian kirsnan ‘black’ and Old Indic kr̥ṣṇá- ‘black, dark’, the main expression for blackness in this language. They derive from the PIE root *kers- ‘black, dark, dirty’.

2.3 Lexical sources of the main terms for red

Studies of history of art show that painting in red was practiced much earlier than painting in other colors in prehistory, since red pigments may be easily produced using certain kinds of earth (cf. Jones and MacGregor 2002). It is therefore understandable that a root for red – PIE *h ₁ rewd ^h -, attested by English red – is one of the most ancient and widespread expressions of this color in the whole IE domain.^[11] It is also understandable that basic color terms for red often go back to expressions meaning ‘paint, painted, colored’, since red was originally considered as the color par excellence. As can be seen in (6), Old Indic rakta-, a verbal adjective derived from the Old Indic root raj ‘to color’ (and ultimately from PIE *(s)reg-, cf. Ancient Greek hrézō ‘to color, immerse in the dye’), is the lexical source of the basic color terms for red of Nepali, Sinhala, and Dhivehi, in addition to minor denominations for color, red, blood, or fire in many other Middle and New Indic languages.

A similar semantic pattern underlies the basic color term for red in Sindhi, g̠āṛho, derived from Old Indic gāḍha- ‘dived into, bathed in’, since textiles were immersed into the dye to be colored. The numerous Indic cognates of Sindhi g̠āṛho ‘red’ therefore represent images of immersion, compression, or solidity, often referring to liquids or tissues, and related metaphors (cf. Pali gāḷha- ‘thick, strong’, Panjabi gāṛhā ‘thick, close’, m. ‘a thick kind of cotton cloth’, Hindi gāṛhā ‘thick, dense, viscous, muddy; deep, dark (of a color); coarse, heavy (of a cloth); close, intense (of a friendship or enmity); bitter (of an experience); m. a type of coarse, thick cloth’, etc.

On the one hand, what has been painted may therefore appear as being darkened or dark with respect to the background. From this point of view, a term for red can have similar sources as a term for black, which we have seen often go back to roots of darkness or dirtiness. That is probably the reason for the etymological connection between English dark and Proto-Celtic *dergo- ‘red’, the source of the Modern Irish basic color term for red dearg, cf. (7). The same PIE root is in fact a cognate with some Lithuanian expressions of bad weather, since something wet may also appear as soiled, dirty, or dark.

On the other hand, what has been colored may be seen as something beautiful, lovely, or pleasant with respect to a plain undecorated object – what seems contradictory at first sight may well be explained simply from another point of view. It is enough to think of krásnyj, the basic color term for red in Russian, literally meaning ‘beautiful’ (as can be seen in Slavic cognates such Old Church Slavonic krasĭnŭ ‘beautiful’, Serbo-Croatian krȃsan ‘id.’, Czech krásný ‘id.’, etc.). Consider also Serbo-Croatian ljubičast, used for a variety of red, such as purple, and clearly related to ljúbiti ‘to love, desire’. We have also seen in (6) that, in addition to its basic meaning ‘painted’, Old Indic rakta- not only means ‘red’, but also ‘dear, lovely’. Another example of this semantic pattern is illustrated in (8), where the basic color terms of most New Indic languages represent borrowings from a Persian expression, lāl, characterized by this range of meanings.

Although the etymology of the Persian source lāl is not watertight, a basic meaning ‘dear’ may explain not only its usages in the sense of red and various red objects, but also its semantic extensions as a denomination of a child as well as of a mute person – people with disabilities are often denoted by forms of endearment. All these meanings have cognates in Indic, where the root lal is attested with the meanings ‘to favor, desire; to play, sport, dally, frolic, behave loosely or freely’ (cf. Kashmiri lalawun ‘to fondle’; Gujarati laḷvũ ‘to be in an ecstasy of love’, Hindi lalaknā ‘to long for’), as well as to denote a child or a boy (cf. Hindi lalā, lallā m. ‘boy, darling’, etc.). Moreover, the Indic root lal may indicate the loll of the tongue by a sort of onomatopoeia and, by metonymy, saliva or mucus (cf. Hindi lāl, lār ‘saliva’). In highly polysemic lexical units as these, it would be wrong, in my opinion, to search for one meaning that may be compatible with all other usages. Rather, we have to reconstruct a meaning, in this case a meaning of endearment, fondness, and pleasure, which is more widely attested and may give rise to most other semantic extensions. The latter, in their turn, are the source of further partly overlapping (and sometimes contrasting) figurative usages, by a relation of family resemblance, in Wittgenstein’s sense. In general, suggesting a broader meaning may be the only way to explain apparently widely divergent meanings in related terms which can often cause scepticism even among linguists.

As a manifestation of beauty and pleasure, the color red is often associated with other bright and warm colors and draws on lexical sources of burning and shining. In this sense, basic color terms of red may be cognate with denominations of brightness and whiteness, as in (9), where the basic color term for red of Tajik and Pashto, surx and sūr, respectively, derive from an Old Iranian expression of redness and warmth and are related to Old Indic denominations of bright and (more marginally) white. In New Persian and in Dari, surx is also common as a term for red, but is higher in register and relatively less frequent in daily speech with respect to qermez, see below.

New Indic structures such as Kashmiri surakh/surkh are borrowed from Persian. A similar semantics underlies the Kashmiri basic color term wŏzul ^u ‘red’, derived from Old Indic ujjvala- ‘burning, blazing up, bright, luminous’, an assimilated form from the prefixed root ud-jval ‘burn brightly, blaze, glow, shine’. In IE, which originally presents a different morphology for the verb and for the noun, expressions of red are much more commonly derived from verbal roots of burning and shining than from the noun of fire. Ancient Greek pyrrós ‘flame-colored, red’, which is clearly derived from pûr, pyrós n. ‘fire’, is only a minor denomination of red, mainly used for the color of red hair (also used as an anthroponym as Pýrros m., Pýrra f.).

Other nouns denoting typical examples of red objects, e.g., the name of blood, are also attested as lexical sources of terms for red, although they are more common for minor expressions than for basic color terms of redness. An example of this can be seen in Hittite išḫarwīl- / ešḫarwīl- ‘red’, transparently derived from the noun ēšḫar, išḫan- n. ‘blood’, and less commonly used than mi(t)ta- / miti- (SA₅) (adj.) ‘red’; (c.) ‘red wool’. The latter is etymologically unclear. We are probably dealing with a pre-IE substrate to be compared with phonetically irregular correspondences such as Ancient Greek míltos f. ‘red earth, ochre’ and Latin minium n. ‘red-lead’ (cf. Cotticelli-Kurras forthcoming). Names of metals may be in fact quite relevant for expressing color categories. We have seen that terms of white are often cognate with names of silver. Terms for red are related to names of copper. Old Indic lohá-, derived from a late or dialectal variant of the PIE root *h ₁ rewd ^h -, means ‘reddish’ as an adjective and ‘copper’ as a substantive (m./n.). From the same root, the adjective lóhita- (a variant of róhita- ‘red, reddish’, see below) means both ‘red’ and ‘made of copper’ – both these values are attested for lóhita- since the Atharva Veda. The basic color term for red of Konkani, tāmbḍo, goes back to an Old Indic form tāmrá- ‘made of copper; copper-colored, reddish, red’; n. ‘copper’ with an extra-suffix -ḍa-, which brings about names of copper recipients in various Indic languages, e.g., Sindhi ṭrāmiṛī f. ‘copper pot’. The ultimate source of Old Indic form tāmrá-, however, is the same PIE root *temH- as in Old Indic támas- n. ‘darkness’ discussed in Section 2.2. This tallies with the semantic pattern whereby terms for red are often derived from expressions of dark colors, as we have seen in (7). Nominal sources of basic color terms for red are quite frequent, instead, when they indicate the material substance, such as grains or larvae, from which red pigments are produced. This semantic pattern is found, in an Eastern area of the IE domain, in the basic color terms for red of most Slavic and Iranian languages, as shown in (10).

A borrowing from Iranian (cf. Middle Persian klmyr) brings about the Armenian basic color term karmir ‘red’, as well as, outside IE, Turkish kırmızı (which shares the functional domain of red with the inherited form kızıl in this language) and the Arabic minor denomination qirmizī, expressing a deep kind of red. (From Arabic, the word passed to Old Italian carmesi / cremesi / cremisi, which was later borrowed in several other languages of Europe, such as English crimson, in the wake of the trade of textiles, especially of silk, from East to West.) A similar semantic change independently occurs in some Romance languages, where Portuguese vermelho and Catalan vermell, the basic color terms for red in these languages, go back to Latin vermiculus m. ‘little worm’ (a derivate of Latin vermis m. ‘worm, maggot’ and ultimately of PIE *wr̥mi-), which is also the source of minor denominations of red such as French vermeil. They are cognate with Old Prussian wormyan ‘red’.^[12] In the same vein, the Modern Greek basic color term kókinos ‘red’ (Ancient Greek kókkinos) is a derivate of Ancient Greek kókkos m. ‘grain, seed; gall of kermes oak used to dye scarlet’. This is the earliest attested basic color term for red with a denominal source in IE. Its Latin borrowing coccum n. ‘berry growing upon the scarlet oak’ has been further borrowed into Welsh coch ‘red’. Vulgar Latin *cocceus has been also borrowed into Albanian kuq ‘red’. The derivation coccinus ‘scarlet-colored’ is the source of the basic color term for red of Romansh cotschen, this time by inheritance.

As can be seen, names of objects represent a lexical source for basic color terms of red more often than for basic color terms of white or black. In the case of red, 22 out of 70 IE languages analyzed (=32 %) present a color term with a denominal formation – especially a derivation from the name of grains, worms, or larvae producing red pigments, such as Portuguese vermelho and Ukrainian červónyj. I also counted Konkani tāmbḍo in the group of denominal formations, although its Old Indic source tāmrá- also denotes the color red, in addition to copper (cf. also Old Marathi tāṁbaḍā ‘red’, besides tāṁb f. ‘rust’ and tāṁbeṁ n. ‘copper’ – Marathi is very close to Konkani).

In the majority of cases, however (47 out of 70 = 67 %), the main terms for red of the IE languages again have a non-denominal word formation, from roots meaning ‘to be red, painted, colored’, ‘to be dark’, ‘to be warm, bright’, or ‘to be beautiful, dear’.^[13] Similarly to what we have seen for white and black, for red as well the preferred lexical pattern is to name an X plant or animal as ‘the red one’, rather than to name red as ‘the color of X’, although nature abounds in red or reddish animals, flowers, or fruits. In Old Indic, for example, deer, gazelles, antelopes, horses, mares, and cows are often named as red animals. Old Indic rohít- ‘red’, drawn from the above-mentioned PIE root *h ₁ rewd ^h - ‘red’, for example, is a common name for a red deer or a red mare, as are its cognates róhita- ‘red, reddish’; m. ‘red or chestnut horse’ and rohiṇī- f. ‘red cow’. The same occurs in Marathi rohī f., roheṁ n. ‘antelope’. Cf. also Old Indic aruṇa-kamala- n. ‘red lotus’, aruṇa-cūḍa- m. ‘a cock’ (lit. ‘red-combed’), aruṇa-dūrvā- f. ‘red fennel’; French rouge-gorge m. ‘robin’, rougeot m. ‘common pochard’, rouge-queue m. ‘a kind of redstart’; German Rotdorn m. ‘type of rose’, Rotrübe f. ‘beet’, Rotalge f. ‘dulse’, Rotvieh n. ‘kind of cattle’, and so on.

2.4 Lexical sources of the main terms for green

Basic color terms for green turn out to mostly derive from roots denoting growing plants in my data. A typical example of this can be seen in the etymological connection between forms such as English green, grow, and grass, according to a semantic pattern which is widely attested in Germanic (11). The basic color terms for green of all Germanic languages derive in fact from this root. Meanings related to growing plants, however, are more common in IE for this root than color meanings of green, which are limited to Germanic. The latter may be therefore considered to be the target of the semantic change.

Owing to this, basic color terms for green also often express succulence or freshness (and metaphorically youthfulness). Moreover, since growing and ripening plants often turn from green to yellow, the same term may be used for both green and yellow or for kinds of greenish-yellow. An example of this polysemy is attested by Hittite ḫaḫli- (SIG₇) ‘green, yellow, greenish-yellow’, which also presents various suffixed variants such as ḫaḫḫaluwant-, ḫaḫlawant-, ḫaḫliwant- with the same meaning, and which is derived from the form ^(GIS̆)ḫāḫḫall- n. ‘plant, vegetable’. In Hittite, the same term is therefore used for both chromatic categories, which may be more precisely distinguished from the context.

Alternatively, green and yellow may be expressed by synchronically different forms which originally go back to the same lexical source. In Classical Armenian, for example, the form dalar ‘green, fresh’ (replaced in Modern Eastern Armenian by the etymologically obscure basic color term kanač’) is cognate with dełb, dełin ‘yellow, pale’ (cf. Modern Eastern Armenian deghin ‘yellow’), as well as with deɫ ‘grass, potion, poison’, deɫj ‘peach’, etc. They derive from the PIE root *d ^h elh ₁ - ‘to sprout, grow’, which is also at the basis of Ancient Greek thállō ‘to sprout, grow, thrive, bloom’. In Baltic and in most Slavic languages, the basic color terms for green and for yellow are also cognates, see Old Prussian saligan ‘green’ and gelatynan ‘yellow’, Czech zelený ‘green’ and žlutý ‘yellow’, which are also related to nouns of vegetation such as Czech zelí n. ‘cabbage’, pl. ziele ‘herb, weed’. Their source, the PIE root *ǵ ^h elh ₃ -/*ǵ ^h leh ₃ -, yields basic color terms for green in most New Indic languages (cf. Nepali hariyo, Panjabi and Hindi harā, Marathi hirvā, etc.), in Eastern Iranian languages such as Sogdian (zrγwny = zərγōnē) and Pashto (zarghun), and in Classical Greek (khlōrós). At the same time, PIE *ǵ ^h elh ₃ -/*ǵ ^h leh ₃ - also gives rise to basic color terms for yellow in Western Iranian, as in New Persian, Dari, and Tajik zard.

A similar phenomenon emerges in borrowing, whereby terms of green may be borrowed in the sense of yellow or vice versa. Latin viridis ‘green’, typically used to describe vegetation in this language, is the source of the basic color terms for green in all Romance languages (Italian, Spanish, Portuguese, and Romanian verde, Catalan, Occitan, and Romansh verd, French vert, Sardinian vírde / bírde, etc.) – Latin viridis represents the most stable color term from Latin to Romance. This term has also been borrowed in Brittonic Celtic, cf. Modern Welsh gwyrdd ‘green’ and Modern Breton gwer ‘id.’, as well as in Albanian verdhë, which, however, means ‘yellow’. On the other hand, the (rarely attested) Latin form galbinus ‘yellow-green’, inherited as the basic color terms for yellow of most Romance languages (French and Occitan jaune, Romanian galben, etc.), has been borrowed in Albanian gjelbër, meaning ‘green’. In Albanian, gjelbër can be used besides jeshil and (less frequently) blertë as a denomination of green. While Albanian jeshil is a borrowing from Turkish yeşil ‘green’, Albanian blertë is cognate with English bloom and German blühen ‘bloom, flourish’, Blume f. ‘flower’, Blatt n. ‘leaf’, etc.

Despite the different lexical sources and the different routes either by inheritance or by contact, the semantic change from expressions of growing plants or fruits to basic color terms for green is strikingly consistent across different areas and different periods of IE. A variant of Persian sabz ‘green, verdant, fresh, unripe, raw’ (cf. also Tajik and Dari sabz ‘green’),^[14] which has several cognates denoting vegetables, has been borrowed into Indic in the basic color terms for green of Kashmiri (sabạz) and Bengali (śobuj). Sinhala koḷa ‘green’, also meaning ‘leaf’, derives from Old Indic kuvala- n. ‘jujube fruit’. Similarly, Dhivehi fehikula ‘green’ literally means ‘leaf-colored’ (it is formed by kula, a borrowing from English color, and faiy ‘leaf’, the latter from Old Indic páttra- n. ‘wing, feather; leaf, petal’, cf. Sinhala pata ‘id.’).^[15] Modern Greek prásinos ‘green’ is derived from the name of leek (práson n. in Ancient Greek), and originally meant ‘leek-colored’.

Other sources are definitely less common. In South Asia, renowned in antiquity for its precious stones, some languages derive their basic color terms for green from the name of the emerald. An example of this can be observed in Konkani patsvo ‘green; emerald’ (while Konkani tarno ‘green’, derived from Old Indic táruṇa- ‘young, tender’, is especially used for unripe fruits, cf. also Marathi tarṇā ‘young’). Names of gems, however, more commonly give rise to minor denominations of green, rather than to basic color terms, as in English emerald-green, or in Sanskrit mārakata- ‘green’, a quite rare form derived from marakata- n. ‘emerald’.

In my data, basic color terms for green derive from the name of a green object in five out of 70 IE languages analyzed (=7 % of the cases). Apart from Konkani patsvo ‘green’ from the name of the emerald, the other cases are derived from the name of a plant, as can be seen in Modern Greek prásinos, Sinhala koḷa, Dhivehi fehikula, and Sogdian zrγwn’k / zrγwnyy (=zərγōnē), the latter derived from Sogdian zrγwn (=zərγōn) ‘plant, vegetable’. In 58 out of 70 cases (=83 %), basic color terms for green derive from roots of growing or ripening vegetation, which may have cognates, but not lexical sources, in specific names of plants.^[16] Plants, or places rich in plants, are often named on the basis of the basic color terms for green, cf. English greengage for a sort of green plum, greenfield, greenfly for an insect which is harmful to plants, green onion, green pepper, green salad, green tea, etc. Persian sabz-bāl ‘a sort of grape’, sabz-dāya ‘a violet’, sabz-roshan ‘a kind of pigeon’, sabz-qabā ‘a kind of yellowish-green bird’, sabz-girā ‘another bird’, etc.

2.5 Lexical sources of the main terms for yellow

We have seen in Section 2.4 that the roots of basic color terms for green often bring about basic color terms for yellow, either in the same language or in related languages. The above-mentioned PIE root *ǵ ^h elh ₃ -/*ǵ ^h leh ₃ -, the most polysemic root of the PIE color lexicon, is the source of basic color terms for both yellow and green in Baltic and in most Slavic languages, e.g., Polish żółty ‘yellow’ and zielony ‘green’ (the exception, in Slavic, is represented by Slovenian, where the basic color term for yellow, rumen, is drawn from the PIE root *h ₁ rewd ^h - expressing red, on the basis of a polysemy between yellow and red). The same PIE root *ǵ ^h elh ₃ -/*ǵ ^h leh ₃ -, which underlies Hindi harā ‘green’, for example, is the source of the Germanic and Iranian basic color terms for yellow as well, cf. (12). Moreover, the PIE root *ǵ ^h elh ₃ -/*ǵ ^h leh ₃ - is the basis of denominations of gold (Vedic híraṇya- n., Gothic gulþ, etc.) and of bile (Avestan zāra- m., Ancient Greek khólos m. / kholḗ f., Latin fel, fellis n., etc.) in many IE languages. This polysemy between yellow and green, based on the growing and cultivation of plants and fruits, may motivate the development of the basic color terms for yellow of Spanish (amarillo) and of Portuguese (amarelo), derived from diminutive forms of Latin amārus ‘bitter, sour’, since green fruits often have this flavor. Here we can observe a synesthesia from the area of taste to that of color.^[17]

Polysemies other than yellow-red and yellow-green are much rarer but still attested for the lexical sources of the main color terms for yellow. Tocharian tute ‘yellow’, for example, may be ultimately derived from the same PIE root *d ^h uH- underlying denominations of dark or black colors, as well as of smoke or dust, in other IE languages, cf. Old Indic dhūmá- m. ‘smoke, vapor, mist’, dhūmrá- ‘smoke-colored, dark, gray, purple’. The latter is the source of Sinhala dumburu ‘brown’, as we will see in Section 2.7. The motivation for these usages may be that objects covered with dust may appear as yellowish, or that a pale aspect may be described as yellow or as dark, as deprived of luminosity.

Moreover, basic color terms for yellow are often derived from denominations of yellow referents, either typical examples of this color category or items which are used to produce yellow pigments. Lexicalizations based on concrete practices of making colorants prevail over more abstract comparisons with objects characterized by a certain color. In both cases, as in the domain of green, the lexical sources of basic color terms for yellow are mainly represented by nouns of plants or plant products. Greek kítrinos ‘yellow’, for example, a minor denomination of this color in Classical Greek but a basic color term in Modern Greek, is a derivate of Ancient Greek kítron n. ‘lemon’, and originally meant ‘lemon-colored’. Catalan groc ‘yellow’ derives from Latin crocum n. / crocus m. ‘saffron (Crocus sativus)’ (borrowed in its turn from Ancient Greek krókos m. / f. ‘saffron’). The Catalan term has been adopted in the form grògu in Algherese, a dialect of Catalan spoken in Alghero, in the northwest of Sardinia, and has later become the basic color term for yellow in the whole island. (Grògu is used besides borrowings from Italian such as giallu / zallu, whose usage varies according to different regions of Sardinia.) The use of dried stigmas and styles of saffron to produce yellow colorants was introduced to the Roman Empire from the East, where the cultivation of this plant was much more widespread since antiquity. Even today, saffron is mainly produced in Iran. Accordingly, denominations of the color yellow based on the name of saffron, as well as, more frequently, of turmeric, are attested in several Asian languages.

Turmeric (Curcuma longa) is actually a different plant, belonging to the ginger family, whose roots are ground to obtain yellow pigments. Its production requires a simpler process than in the case of saffron, and consequently saffron is a much more expensive spice than turmeric. However, since their pigments have a similar color – a vivid yellow or orange – the two plants are often confounded, to the point that turmeric is also called ‘Indian saffron’, as it is native to South and Southeast Asia. In Old Indic, turmeric has about fifty different names, owing to its widespread use in Ayurvedic medicine, as a healing spice, in addition to culinary practices (e.g., to prepare curry), dyeing textiles, cosmetics, etc. Old Indic haridrā- f. ‘Curcuma longa’, built on the basis of the Vedic form hári- ‘yellow-green, yellow-red, tawny, golden, brown’ and ultimately derived from the PIE root *ǵ ^h elh ₃ - in (12), is the most common among these denominations. Old Indic already attests a vṛddhi derivate hāridrá- with a chromatic meaning ‘colored with turmeric, yellow’; m. ‘a yellow color’ (cf. also hāridraka- ‘yellow’), although they are just marginal color denominations at this stage. In contrast, the name of the curcuma later gives rise to authentic basic color terms for yellow in Kashmiri, Bengali, and Konkani, among others, as illustrated in (13). In Dhivehi reen’dhookula ‘yellow’, which is also a basic color term in this language, the cognate name of turmeric (reen’dhoo) has been remotivated with the noun kula ‘color’. Similarly, the Sinhala form kaha, meaning ‘yellow’ and ‘turmeric, saffron’, derives from Old Indic source kaṣāya- m. / n., which indicates a yellowish plant and its astringent juice.

In most Indic languages, however, basic color terms for yellow go back to a rare Old Indic form pītala- ‘yellow’ (> Nepali pahẽlo, Hindi and Panjabi pīlā, Gujarati pīḷo, Sindhi pīlu, etc.) or from a cognate Middle Indic form pīvala- ‘yellow’ (the source of Marathi pivḷā). They are possibly derived from a root pi or pyai (intr.) ‘swell, abound’; (tr.) ‘fatten, cause to swell’, as fat and oily substances often have a yellowish aspect (cf. the English expression yellow grease to denote a kind of cooking oil). Names of animals or animal products turn out to be anyway much less common than names of plants or vegetal substances as lexical sources of basic color terms for yellow. Examples of this can be seen in Celtic. In Brittonic, in particular, Modern Welsh melyn ‘yellow’ and Modern Breton melen ‘id.’ can be traced back to the name of honey (cf. Welsh and Breton mel m. ‘honey’) and therefore originally meant ‘honey-colored’. Similarly, Romansh mellen, melen ‘yellow’ derives from Latin *mellinus ‘honey-colored’ (cf. REW 5483). Instead, Latin melinus ‘related to honey’ and melleus ‘id.’ were rare (and not always meant in the sense of color) in this language, which did not have any established basic color term for the yellow color category. Alternatively, a basic color term for yellow may derive from another color expression which was especially used to describe the coat of an animal. In Goidelic, Old Irish buide ‘yellow’ and Modern Irish buí ‘id.’ derive from a Proto-Celtic stem *bodyo- ‘yellow-brown’. The Celtic form has been probably borrowed into Latin badius ‘brown, chestnut-color’, a rare form specialized for the color of some horses (> French bai, Italian and Portuguese baio, Spanish bayo, etc.). In the language of farming, the color of animal hair is important to distinguish and classify animals.

This may find a parallel in the use of expressions of yellow to describe the color of blond hair. In Homeric Greek, the etymologically obscure adjective xanthós ‘blond, yellow, yellow-red, reddish-brown’ mainly refers to human hair or animal fur. The proper name Xánthos denotes one of Achilles’ horses, as well as a river of the Troad. The latter usage may be motivated by the fact that some kinds of earth or mud of the riverbed make its water appear yellowish (cf. the Chinese Huáng Hé, the Yellow River) or to the fact that hair may be associated with the image of waves (cf. the expressions in English wavy hair, French cheveux ondulés, etc.). In Classical Greek, xanthós extends its usage and is applied to all kinds of yellow or yellowish objects, from honey to wine, fried food, etc., but the color of blond hair continues to be its main functional domain (this has also remained as the main meaning of the minor color denomination xanthós in Modern Greek). Latin flāvus ‘golden yellow, reddish yellow, flaxen-colored’ (cognate with Old Irish blá ‘yellow’) and Latin fulvus ‘deep yellow, reddish yellow, gold-colored, tawny’ are also mainly used to describe blond hair, as well as the waves of a river. They are, however, poetic expressions (often translating Homeric xanthós), which predictably decay in the Romance languages.

In my database, I found 11 out of 70 (=16 %) cases of basic color terms for yellow which are derived from names of yellow referents. As we have seen for the color green, nominal sources of basic color terms for yellow also mainly denote plants or plant products. On the other hand, terms for yellow may represent the lexical source for naming the most various yellow or yellowish referents. Cf. English yellowhammer ‘a bird (Emberiza citrinella)’, German Gelbklee m. ‘a plant of the clover family (Medicago lupulina)’, Gelbholz n. ‘a tree, also called American yellowwood (Cladrastis kentukea)’, Gelbwurst f. ‘a kind of sausage from Bavaria’, Dhivehi reen’dhoo iruvaahudhu ‘a kind of yellow-white heron’ (hudhu means ‘white’ in Dhivehi), etc. The usual pattern is that of naming an X animal, plant, or object as ‘the yellow one’, and not that of naming yellow as ‘the color of the X animal/plant/object’ – apart from the case of some plants producing yellow pigments, such as saffron and turmeric. Moreover, as roots of yellow terms often denote bile, basic color terms for yellow often represent the source for the name of jaundice, which is determined by bile pigments in the blood, and which causes a yellowish, pale, or livid color of the skin or of the eyes. English jaundice, borrowed from Old French jaunice (French jaunisse f.), also shows this pattern, as the French forms are derived from Old French jaune, jalne (French jaune) ‘yellow’, from the rare adjective galbinus ‘yellow-green’ mentioned in Section 2.4. Cf. also the English expression yellow fever.

2.6 Lexical sources of the main terms for blue

Expressions of blue originally had a marginal status in IE. Some ancient IE languages, such as Early Latin, Homeric Greek, and Vedic, did not have a proper basic color term for blue, which developed only later in their color lexicon from denominations of blue-green, blue-black, or other mixed color categories. PIE did not have any root specifically used for blue either. The PIE root *ǵ ^h elh ₃ -/*ǵ ^h leh ₃ -, probably meaning ‘yellow-green’ in origin, and widely used for basic color terms of green or yellow in IE, cf. (12), rather describes varieties of blue-green in Celtic, cf. Breton glas ‘blue-green’ (from Proto-Celtic *glasto- ‘green, blue’, cf. Matasović [2009: 160]). Proto-Celtic was therefore a grue-language. Old Irish glas was even more polysemic and could express green, blue as well as gray. Later, phenomena of semantic specialization are observable, as Modern Welsh glas tends to be used as the main expression of the color blue, while Modern Irish glas is used more in the sense of prototypical green (contrasting with the Modern Irish basic color term for blue gorm, see below). In Baltic, the root *ǵ ^h elh ₃ -/*ǵ ^h leh ₃ - also gives rise to the Latvian basic color term for blue: zils. The change of PIE *ǵ ^h elh ₃ -/*ǵ ^h leh ₃ - therefore follows a semantic path from yellow-green to green in some languages, to blue-green, and then to blue in others, owing to the continuous distribution of colors on the visible spectrum. As anticipated in Section 1, macro-categories such as grue or yeen are not in contradiction with Berlin and Kay’s (1969) sequence of basic color categories in (1) (cf. Biggam 2012: 76: “successive sub-dividings of these large categories result in increasing numbers of single-hue categories”). The development from the meaning of blue-green to that of blue is anyway much less frequent in IE than the semantic change from black or dark to blue. In most modern Indic languages, for example, basic color terms for blue, as in Hindi nīlā ‘blue’, are coded by forms derived from Old Indic nī́la-, which at the stage of Vedic, however, did not mean ‘blue’, but rather ‘black’ or generically ‘dark’ (14), although it was minoritarian for black as well with respect to Old Indic kr̥ṣṇá-.

Expressions of dark colors in fact represent a common lexical source for terms of blue. Serbo-Croatian plâv, the basic color term for blue in this language, derives from the PIE root *pel(H)-, which is one of the most typical IE denominations of gray, especially of gray or hoary hair, e.g., Vedic palitá- ‘gray, hoary, old’ (cf. Section 2.8). In Section 2.2, we have seen that Lithuanian mė́lynas ‘blue’ has cognates in basic color terms for black, such as Ancient Greek mélas. In this case, the semantic change from black or dark to blue may have started, in my opinion, in contexts referring to bruises, whose livid aspect may be seen as an expression of these colors. It is enough to think at expressions such as English to beat someone black and blue, or to the fact that similar expressions using terms for blue in a language may be translated with terms for black in another language, e.g., German blaues Auge vs. English black eye. The Germanic basic color terms for blue are consistently drawn from Proto-Germanic *blēwa- (> German blau ‘blue’, Yiddish bloy ‘id.’, Dutch blauw ‘id.’, Afrikaans blou ‘id.’, Frisian blau ‘id.’, Modern Icelandic blár ‘id.’, Faroese blátt ‘id.’, Danish, Norwegian and Swedish blå ‘id.’, etc.) and borrowed into the Romance languages (cf. Romansh blau / blo ‘blue’, French bleu ‘id.’ (> English blue, cf. Biggam [2006]), Catalan and Occitan blau ‘id.’, etc.). Originally, however, these terms were extended to other dark colors, cf. Old Norse blár ‘blue, black, dark’, Old High German blāo ‘blue, dark, gray’, etc. We can observe the relics of this polysemy in certain idiomatic expressions of North Germanic, where a black person is literally called ‘blue person’ (Modern Icelandic blámaður, Swedish blåman, etc.). The same usage appears in Celtic, as in Modern Irish duine gorm ‘black person’ (lit. ‘blue person’). This is because Old Irish gorm meant not only ‘blue’, but also ‘dark, black’, as can also be seen in some results of Proto-Celtic *gurmo-, such as Middle Welsh gwrm ‘dark, dark blue’ and Old Breton wurm ‘dark’. This usage may have been further reinforced by the Old Norse expression, as a semantic loanword, at the time of the Viking occupation of Britain and Ireland, in the mid-9th century.

On the other hand, terms of light blue, or blue in general, may be also derived from expressions of brightness, although in this case they tend to remain marginal denominations of this color, rather than basic color terms. Latvian dziedrs ‘azure’, for example, is etymologically connected with several Lithuanian expressions of a clear sky or serene weather, e.g., Lithuanian giẽdras ‘clear’ and giedrà f. ‘fine weather’, as well as with Ancient Greek phaidrós ‘bright, clear’ – ultimately from a PIE root *g ^wh h ₂ eyd- indicating luminosity.

The relatively late development of denominations of blue in IE, as compared to the terms of other colors discussed above, may be explained by the fact that blue coloring agents are difficult to find. They can be obtained by extraction from certain plants, such as the Indigofera tinctoria, and derived from precious stones, such as the lapis lazuli or the azurite, which were originally imported into Europe from various regions of Asia. The Indigofera tinctoria, whose fermented leaves produce the indigo dye, was imported from India – the name indigo comes from Latin indicum n. ‘Indian (substance), indigo’, borrowed in its turn from Ancient Greek indikòn phármakon ‘Indian dye’. The lapis lazuli mainly came from the mountains of Afghanistan. In the Roman Empire, blue pigments were therefore expensive and considered as luxury products (cf. Pastoureau 2006: 17ff). In my data, I could observe that names of mineral and vegetal coloring agents are in fact the most common lexical sources for basic color terms of blue.

The case of terms for blue going back to names of gems and precious stones appears in (not always synonymous) Romance forms such as Italian azzurro ‘light blue, azure’, French azur ‘azure’, Spanish and Portuguese azul ‘blue’, etc. (In Spanish and in Portuguese, azul is the basic color term for blue.) These forms have been borrowed from Arabic lāzuward / lāzaward ‘lapis lazuli; azure’ (the latter also produced derivates such as Arabic lāzuwardī ‘azure, sky-blue, cerulean’). Arabic lāzuward, in its turn, is borrowed from Persian lājward (probably derived from the toponym Lāžvard, in the Badakhshan region of Central Asia, which is particularly rich in these gems). Persian lājward is also the source of the Turkish borrowing lacivert ‘dark blue’, as well as of the Sanskrit borrowing rājāvarta- m. ‘lapis lazuli’ (lit. ‘royal jewel’, with a folk etymology on the basis of rājan- m. ‘king’). Another example from Iranian is presented in (15).

In the same vein, Modern Greek γalázjos and γalanós, expressing light blue, derive from Ancient Greek kalaïs, the name of a precious stone of greenish blue, probably a kind of turquoise or chrysolite. γalázjos is relatively more common than γalanós. It is often used in opposition to Modern Greek ble ‘blue, dark blue’, (the latter borrowed from French bleu). Romanian albastru ‘blue’ goes back to a Vulgar Latin form albaster ‘whitish’, a derivate of Latin albus ‘white, mat white’. The semantic change from white to blue is not usual (while that from black to blue, we have seen, is frequent). This apparently idiosyncratic pattern originated from the description of a prestigious kind of white marble characterized by blueish shades. Names of blueish kinds of metal are less frequent than names of gems but are also attested. Classical Greek kuáneos ‘dark blue’ (often opposed to glaukós ‘light blue, blue-green, bluish gray’ at this stage of the language), derives from kúanos m. ‘dark-blue enamel’, used to decorate armor. In Homeric Greek, where kuáneos rather indicates varieties of black or dark colors, the reference to metal is still evident.

Denominations of blue derived from names of vegetal coloring agents can be seen in the basic color terms for blue of some Slavic languages, such as Czech and Slovak modrý (16), which go back to PIE *mod ^h -ro-, and which have Germanic cognates denoting coloring plants, such as English madder. These forms especially indicate the Rubia tinctorum, which is usually employed to produce red or purple pigments. They therefore meant a generic colorant rather than specifically a blue dye in origin. Other forms derived from the same PIE root *mod ^h -/*m̥d ^h - may also be found in the old IE languages, e.g., in Hittite antara- /andara- ‘blue’ (ZA.GÌN) and in Tocharian B motartstse, which, however, meant ‘green’. The polysemy between blue and green is especially common in the languages of Central and Eastern Asia – in fact, the Tocharian denomination of blue, that is, tseṃ, is probably a borrowing from Middle Chinese chieŋ (Mandarin Chinese qīng), meaning ‘blue-green’. Another example of this semantic pattern (involving a borrowing) appears in Albanian kaltër ‘blue’, taken from a Vulgar Latin derivate *calthinus, from Latin caltha, the name of a kind of Calendula officinalis, which could be used for dyeing.

Alternatively, terms for blue may be drawn from names of animals, such as the dove or the pigeon, characterized by certain shades of this color. A typical example of this is provided by Russian golubój ‘light blue’, which is derived from golub’ ‘dove, pigeon’, and which in Russian is opposed to sínij ‘dark blue’. This dichotomy has been amply discussed in the literature, according to which Russian possesses 12 basic color terms and therefore contradicts the maximum number of basic color terms predicted by Berlin and Kay (1969: 35–36), cf. Paramei (2007). I have identified a similar formation in Old Prussian golimban ‘blue’ (a cognate with Russian golubój), which can be further approached to the semantic pattern of Tajik kabud ‘blue’ and Armenian kapowyt ‘blue, dark blue’. The latter is a borrowing from an Iranian denomination related to Old Persian kapautaka- ‘blue’ (cf. also Sogdian kp’wt (=kapōt) ‘blue, gray-blue’, kp’wtk (=kapōtē) ‘id.’) and further to Vedic kapóta- m. ‘dove, pigeon’. In New Persian, the minor denomination kabûd means ‘dark blue’ and is translated as Russian sínij.

Also of note are terms of blue derived from names of objects that can be seen as typical examples of the color blue, as the names of the sky or of water (17). Spanish celeste ‘light blue’, from cielo ‘sky’, has acquired the status of a basic color term in some varieties of South America, as in Uruguay (Lillo et al. 2018), although in Standard Spanish it is minoritarian with respect to azul. In Italian, celeste ‘light blue’ is also marginal with respect to blu ‘blue’.^[18]

A basic color term for blue based on the name of water is attested in New Persian ābī ‘blue’, also existing in Dari, a transparent derivation from āb ‘water’, therefore literally meaning ‘watery, water-colored’. The same semantic pattern has been borrowed into Turkish mavi ‘blue’, although in this case the specific lexical source comes from Arabic mā’ ‘water’, māwiyy ‘watery’ (which is not the basic term for blue in Arabic – the Arabic basic color term for blue is rather ’azraq).

As can be seen, basic color terms for blue derive from the name of an object more frequently than the basic color terms previously analyzed. According to my data, this occurs in 23 out of 70 languages, that is, in 33 % of the cases. Their lexical sources denoting gems, colorant plants, etc. are also quite consistent. In this case as well, however, most basic color terms have a non-denominal formation, especially from roots denoting dark colors.

2.7 Lexical sources of the main terms for brown

To denote the brown color category, PIE had one main root available, that is, *b ^h er(H)-, the same as in English brown. Germanic is in fact one of the two branches where this root has been better maintained until the present day (18a). The other branch is Indic, where the root *b ^h er(H)- is used to express brown since the earliest Vedic period. According to my data, Vedic is the first IE language where we have a basic color term for brown, such as babhrú-, which could also describe varieties of reddish-brown. Modern Indic basic color terms for brown mainly derive from a cognate non-attested form *bhrūra- (18b).

Other languages attest the PIE root *b ^h er(H)- in some marginal denominations of brown, as in Lithuanian bė́ras ‘tawny, brown’. Not all ancient IE languages, however, attest a specific term for brown. In my database, brown is not lexicalized in ten out of 70 IE languages (=14 %), that is, in Hittite, in Avestan, in Old Persian, in Ancient Greek, in Latin, in the Sabellic languages, in Gothic, in Old Church Slavonic, in Classical Armenian, and in Tocharian. In Gothic, this may be due again to a lacuna in our source material, as a Proto-Germanic form *brūnaz may be reconstructed by comparing the Germanic terms of brown in (18a). In the other languages, this lack is rather due to the usage of polysemic forms expressing a dark color in general, which can be interpreted as brown from the object at issue (e.g., if a piece of wood is described) and from the context. The Classical Greek form órphninos, for example, which according to Plato (Timaeus 68c) expresses a combination of white, red, and especially black, generically indicates a dark color. Latin also has various terms expressing darkness, such as the cognate forms furvus ‘dark, dusky, gloomy, swarthy, black’ and fuscus ‘dark, swarthy, dusky, tawny’, which may be occasionally used to describe brown referents. They are derived from the same root as English dusk, dusky as well as Irish donn, cf. (19). In Old Irish, in particular, donn competes with the forms odar and cíar (now obsolete in Modern Irish) to express the brown color category. In Modern Irish, donn is the basic color term for brown. According to my data, Old Irish is the ancient IE language that is richest in color terms, presenting even denominations for color categories that are not expressed, or marginally expressed, in other ancient IE languages.

According to the same semantic pattern anticipated in Section 2.5, the Sinhala basic color term dumburu ‘brown’ goes back to Old Indic dhūmrá- ‘smoke-colored, smoky, dark-colored, gray, dark-red, purple’, a derivate of the name for smoke (Old Indic dhūmá- m.) which could express the most varied dark colors.

Basic color terms for brown can also derive from denominations of (varieties of) red or yellow, although these patterns are less common than that derived from expressions of dark or black. The Lithuanian basic color term rùdas ‘brown’, for example, is cognate with Lithuanian raudónas ‘red’ and is drawn from the PIE root *h ₁ rewd ^h - which gives rise to denominations of red in most IE languages (20). Modern Breton ruzdu ‘brown’ is a transparent compound formed by the main Breton terms for red (ruz, again from PIE *h ₁ rewd ^h -) and for black (du), the constituents of the mixed brown chromatic category. The same pattern appears in Welsh cochdu ‘brown’, which, however, is minoritarian with respect to Welsh brown, borrowed from English.

In Modern Breton, ruzdu is used besides gell to express brown, the latter properly denoting a light kind of this color. Modern Breton gell is cognate with Celtic denominations of white (cf. Old Irish gel) or of yellow (cf. Middle Welsh gell), derived from the PIE root *ǵ ^h elh ₃ - which we have seen in (12) to underlie most IE terms for yellow or green. The highly polysemic Vedic form hári-, derived from the same root, can also be used for varieties of brown, reddish-brown, tawny, pale yellow, or golden.

Alternatively, basic color terms for brown can be derived from the names of brown referents, which are very common in nature – earth, wood, many fruits, the coat of many animals, or the color of human hair and eyes, etc. In Western IE, notably in most Romance languages, the dominant pattern is that of a basic color term for brown derived from the name of a chestnut. Italian marrone ‘brown’, Spanish marrón ‘id.’, Catalan marró ‘id.’, Brazilian Portuguese marrom ‘id.’, French marron ‘id.’, Romanian maro ‘id.’, etc. go back to the homonymous names of the chestnut (cf. French marrons glacés) and ultimately to a Vulgar Latin form *marrō, -ōnis ‘chestnut’. In the variety of Portuguese spoken in Portugal, the basic color term for brown is rather castanho, which is also derived from a name of the chestnut, that is, Portuguese castanha, from Latin castanea.^[19] A similar pattern appears independently in Modern Armenian, where the basic color term shaganakagowyn ‘brown’ literally means ‘of the color (gowyn) of the chestnut (šaganak), chestnut-colored’.

In Eastern IE, particularly in South Asia, the color brown is often expressed by adjectives derived from the name of the almond, as illustrated in (21). In this case, an originally Persian source bādām ‘almond’ has been borrowed into most modern Indic languages as the name of the fruit and, with an adjectival derivation, of the color brown. In Bengali, in particular, bādāmī is the main term for brown, while in other Indic languages the correspondent forms may have a more marginal status (usually denoting a kind of light brown, as in Hindi) as compared to the results of PIE *bhrūra- in (18b).

In the Middle East and in Southeastern Europe, by contrast, the source of basic color terms for brown is especially the name of coffee, according to my data. Modern Greek kafé ‘brown’, Albanian kafe ‘id.’, New Persian and Dari qahve’ī ‘id.’ ultimately go back to Arabic qahwa ‘coffee’ through an intricate pattern of borrowings. In particular, Modern Greek kafé is borrowed from French café m. ‘coffee’ (and not ‘brown’), which is borrowed in its turn from Italian caffé m. ‘id.’, borrowed from Ottoman Turkish kahve ‘id.’, borrowed from Arabic qahwa ‘id.’. The latter is also the source of the Persian borrowing qahve’ ‘coffee’ and qahve’ī ‘brown’. In Tajik qahvarang ‘brown’, the name of coffee is remotivated with rang ‘color’ (cf. Modern Turkish kahverengi ‘brown’). All these borrowings reflect the trade in this product and its cultural significance especially in the Middle East, as well as in the Balkan extensions of the Ottoman Empire, where coffee shops (although occasionally banned) always retained an important role for meeting and socializing. Cf. also Bulgarian kafjáv ‘brown’ and Macedonian kafeav ‘id.’. In Arabic, the lexical source of the basic color term bunnī ‘brown’ is different but the semantic pattern is the same, as this term is transparently derived from bunn ‘coffee bean’ and therefore literally means ‘of the color of coffee beans’.

Alternatively, terms to describe brown are often derived from names of kinds of tobacco. This especially occurs in some modern Indo-Iranian languages. Pashto naswārī ‘brown’, for example, a basic color term in this language, is transparently formed on the base of the noun naswār, indicating a sort of powdered, dipping tobacco commonly consumed in Afghanistan. Dari also uses naswārī besides qahve’ī. Luri triaki ‘brown’ literally means ‘opium-colored’. The Marathi basic color term tapkirī ‘brown’ comes from tapkīra f. ‘chewing tobacco’. Konkani puditso comes from puḍī f., which denotes a kind of powder used as tobacco. In Sindhi, the main term for brown is nāsī, derived from nās, the name of a sort of tobacco taken by the nose and also used as medicine (this name of tobacco derives in fact from Old Indic nāsya- n. ‘nose-cord; errhine’, itself a derivate of Old Indic nāsā- f. ‘nose’). The practice of chewing tobacco, especially the consumption of betel leaves and areca nuts (the seeds of the Areca catechu palm), is widespread in the whole of Asia. In India, they usually call it paan (from Old Indic parṇá- n. ‘leaf’). In the same vein, the name of the Acacia catechu, which can also produce the catechu extract as the areca palm, is the lexical source of the Nepali basic color term for brown khairo ‘brown’ (from Old Indic khadira-ja- ‘made from the Khadira wood’, that is, a kind of Acacia catechu). In Hindi, the Acacia catechu is named katthā, and its derivate katthā’ī (lit. ‘of the color of the Acacia catechu) means ‘brown’, especially a dark kind of this color. The same denominations recur in virtually all modern Indic languages, although they may have a more or less central function in the color lexicon of different languages. Outside of Indo-Iranian, the names of tobacco are not common sources of terms of brown in IE. An example emerges in Sardinian tabakkínu ‘brown’, lit. ‘tobacco-colored’, from Italian tabacco m. (Sardinian tabakkínu is, however, minoritarian with respect to castanzu from the name of the chestnut).

Other phytonyms also bring about terms of brown. In South Asia, besides the names of almonds and tobacco and the results of the PIE root *b ^h er(H)-, we can also identify the name of turmeric as lexical source, which is also the basis of many common expressions for yellow in this area (cf. Section 2.5). The Kashmiri basic color term kāʦur ^u ‘brown’ goes back to the (rarely attested) Old Indic form karcūra- m. ‘turmeric’; n. ‘orpiment’. Unlike Old Indic, the ancient Iranian languages, Avestan and Old Persian, do not attest any specific term for brown (which does not imply that they did not have any). By contrast, in Middle Iranian, I did find expressions for this color. Sogdian attests the form cnt’n β’m’k ‘brown’, literally meaning ‘with the color of the sandalwood’, from cnt’n=čandan ‘sandalwood’ (cf. Classical Sanskrit candana- m. / n. ‘sandal, Sirium myrtifolium’, the name of the tree, of the wood, and of its derived substance used to prepare highly esteemed ointments and perfumes) and β’m’k = βāmē, fāmē ‘color’ (cf. Avestan bāma- m. ‘light, splendor’, connected with the Vedic root bhā ‘to shine’ < PIE *b ^h eh ₂ -). A similar pattern emerges in Eastern Slavic, where Russian and Ukrainian koríčnevyj ‘brown’, for example, goes back to the name of ‘cinnamon’ (koríca, a typical Slavic diminutive form of the name of the ‘bark’, kora f. in Russian). Welsh gwinau, a minor denomination of brown with respect to the recent English borrowing brown, derives from the name of wine (Welsh gwin, borrowed from Latin vīnum n. ‘wine’).

According to my data, names of animals are remarkably less common than names of plants as lexical sources of basic color terms for brown. A rare case can be identified in Dhivehi mushi kula ‘brown’, literally meaning ‘mushi-colored’, from the name of a fish, the horse mackerel, called mushi or mushimas in this language (the noun mas means ‘fish’ and may be used as a suffix to form names of specific fishes). The rarity of lexical sources denoting animals may seem surprising for brown, since this color is very common in the animal domain. On the contrary, it turns out that names of animals often derive from expressions of brown in IE, as we have seen above for English beaver, derived from the PIE root of brown *b ^h er(H)-. Another example is provided by the cognate form bear (cf. also German Bär m., Dutch beer m., etc. from Proto-Germanic *beran- m. ‘bear’), a taboo denomination of bear as ‘the brown (animal)’ – an animal originally called *h ₂ r̥tḱo- in IE (cf. Hittite hartagga- c.; Vedic ŕ̥kṣa- m.; Ancient Greek árktos f., etc.).

Apart from plant products, names of other concrete objects are also not so common lexical sources for basic color terms for brown. In my database, I identified examples of this pattern in some Slavic basic color terms for brown derived from names of metals. Polish brązowy means ‘brown’ and ‘made of bronze’ and is clearly derived from the name of this metal (Polish brąz m. ‘bronze’, borrowed from French bronze, which in its turn is borrowed from Italian bronzo). Slovenian rjav ‘brown’ originally meant ‘rusty’ and derives from a Proto-Slavic name of rust (*rъd’a), ultimately derived from the PIE root *h ₁ rewd ^h - ‘red’ illustrated in (20). This is consistent with the above-mentioned semantic pattern of basic color terms of brown which go back to denominations of the color red.

I counted 29 out of 70 cases of terms for brown derived from the name of a brown referent (=42 %), again more than we have seen for the previous color category. That is, languages show a lexical pattern of naming brown as ‘the color of the X referent’ much more commonly than a lexical pattern of naming white, black, red, yellow, green or blue as ‘the color of the X referent’. This is not completely due to the fact that brown animals and plants are easy to find in nature, since, as mentioned above, names of animals are very rare lexical sources for basic color terms of brown in the analyzed languages. As in the case of yellow and green color categories, names of plants or plant products remain the preferred source for expressions of brown as well.

2.8 Lexical sources of the main terms for gray

Denominations of the color gray are ancient in IE, where at least two widely attested roots may be reconstructed for PIE. The first one, PIE *ǵ ^h er(h ₁ )-, illustrated in (22), is the source of English gray and of the basic color terms for gray of all Germanic languages. Armenian gorsh ‘gray’ also comes from the PIE root *ǵ ^h er(h ₁ )-, but it is marginal with respect to moxragowyn (see below) for the color gray. A Proto-Germanic form *grīsa- ‘gray’, probably etymologically unrelated to *grēwa- ‘id.’ in (22), has been borrowed (through Frankish *grīs) into Romansh grischun ‘gray’, French, Occitan, and Catalan gris ‘id.’ (borrowed in its turn into Spanish gris ‘id.’ and Italian grigio ‘id.’), etc. French gris has been also borrowed into Modern Greek and in Albanian gri ‘gray’ (as well as in Turkish gri ‘id.’).

The second root, PIE *pel(H)-, underlying English fallow, is even more widespread and certainly more polysemic. Nowadays, *pel(H)- is the source of basic color terms for gray in Celtic (cf. Modern Irish liath ‘gray’, Modern Welsh llwyd ‘id.’, Modern Breton louet ‘id.’) and in part of Baltic (Lithuanian pìlkas, cf. Hock et al. [2015: 772]). The PIE *pel(H)- underlies minor denominations of gray and related colors, such as black or white, as well as varieties of yellow, in other IE languages (cf. Latin pullus ‘dark, blackish, blackish-gray, dusky’; Old Church Slavonic plavŭ ‘white’, pelesŭ ‘gray, grayish black’; Lithuanian pal̃vas ‘yellowish, grayish’, etc.). We have seen above in Section 2.6 that this root has also given rise to the Serbo-Croatian basic color term for blue plȃv. This does not imply that the root pel(H)- could express the most varied range of colors – it did not express red or other vivid colors, for example. Instead of reconstructing a vague chromatic meaning, I see the motivation for this polysemy in the connotative potentials of the color gray, and particularly on the one hand in the grayish or whitish hair color of old people, and on the other in the grayish or blueish livid color and lack of saturation of a bruise or of a sick body – old age, weakness, and sickness being often naturally associated. These are in fact the connotations expressed by most derivates of the root *pel(H)-, as can be seen in Vedic palitá- ‘gray, hoary, old, aged’, Avestan pouruša- ‘gray, gray-haired’; Ancient Greek poliós ‘gray, grizzled, grisly’, peliós ‘discolored by extravasated blood, livid’; Latin pallidus ‘pale, pallid’, etc. Even Old Irish líath ‘gray, gray-haired, aged’ privileged the description of hair color of old people, while its Modern Irish descendant has widened its functional domain.

Denominations of gray derived from terms for other dark colors are especially common. An example of this emerges in Tocharian B kwele ‘dark gray, black’, derived from the PIE root *ḱi(H)- and cognate with New Persian siyāh ‘black’, cf. (3), as well as with Old Prussian sywan ‘gray’ and Lithuanian šývas ‘light gray’. Further cognates may be identified in the basic color terms of several Slavic languages, such as Serbo-Croatian sȉv ‘gray’, Bulgarian, Macedonian, and Slovenian siv ‘id.’, and Slovak sivý ‘id.’.^[20]

The most common lexical sources for terms for gray, however, are represented by denominations of ashes, dust, or powder in my data. In Western IE, this pattern is rare – it emerges in Portuguese, whose basic color term for gray is cinzento (in Portugal) and cinza (in Brazil). Cinzento is a derivate of cinza, which also means ‘ashes’ and goes back to a Vulgar Latin form *cinīsia, from Latin cinis, -eris m. ‘ashes’. In Eastern IE, instead, the semantic change from names of ashes to terms of gray is consistently found with different lexemes, and can be especially observed in Indo-Iranian, as well as in Armenian, which as usual presents borrowings or semantic calques from Iranian, notably from Middle Persian, in its color lexicon. The New Persian and Dari basic color term xākestarī ‘gray’ is a transparent derivate from xākestar ‘ashes’ and therefore literally means ‘ashy, ash-colored’. The same holds true for Tajik xokistarang, where the name of ashes is reinforced by means of the form rang ‘color’. In Armenian, the basic color term for gray is moxragowyn, literally meaning ‘ash-colored’, cf. Armenian mokhir ‘ashes’ and gowyn ‘color’. We have here a semantic calque. In Indic, I identified this pattern in Nepali, for example, whose basic color term kharānī ‘gray’ also means ‘ashes’, its original meaning. This term goes back to Old Indic sources expressing the process of being destroyed and reduced to ashes by caustic substances, as can be seen in (23).

The basic color term for gray of Kashmiri sūr ^ü derives from sūr m. ‘ashes’ (cf. also Kashmiri sūr rang ‘ash-colored’). Basic color terms such as Gujarati rākhōḍī ‘gray’ and Marathi and Konkani rākhāḍī ‘id.’, derived from the name of an ash pit (Old Indic *rakṣā-kuṇḍaka-), also mean ‘ashes’. Gujarati also presents bhūkhro ‘gray’, derived from the homonymous name of a powder. Similarly, the Bengali basic color term dhūśor ‘gray’ goes back to Old Indic dhūsara- ‘dusty, dust-colored, gray’, from an Old Indic root dhvaṁs ‘fall to pieces or to dust, decay, be ruined, perish’. Its PIE source *d ^h wens-/*d ^h uns- may be connected to the PIE root *d ^h us-, which we have seen in Section 2.7 to underlie English dusky and other expressions of dark colors. Cognates of Bengali dhūśor emerge as minor denominations of gray across other modern Indic languages, as in Hindi dhūsar, which, however, is nowadays less common than saleTī as an expression of the color gray (the latter, also prevailing in Panjabi, is a borrowing from English slate, the name of a gray stone). The Sinhala basic color term aḷu ‘gray’ also means ‘ashes’. In Dhivehi, we have alhikula, that is, ‘alhi-colored’, where alhi is again ‘ashes’, related to Sinhala aḷu.

Apart from the recurrent derivation from names of ashes (and to a lesser extent of slate stones), other nominal sources denoting gray referents turn out to be not so frequent for basic color terms of gray, or at least much less frequent than roots expressing darkness or lack of saturation, gray hair, livid complexion, etc. My data present 16 out of 70 denominal basic color terms for gray (=23 % of the cases), which is much less than what we have observed for basic color terms for brown. This is probably because names of plants, the most frequent lexical source for denominal basic color terms of brown, among other colors, are not commonly found for terms of gray, as gray is not as common as brown in vegetation. Similarly to basic color terms for brown, however, basic color terms for gray also disfavor nominal sources denoting gray animals, although animals of this color are easy to find in nature. A rare example of this semantic pattern can be identified in Latvian pelēks, the basic color term for gray in this language, derived from Latvian pelė̃ f. ‘mouse’ and therefore properly meaning ‘mouse-colored’. It ultimately goes back to the above-mentioned PIE root *pel(H)-, which underlies so many denominations of gray in ancient and modern IE languages. Otherwise, names of animals are attested for marginal denominations of gray, rather than for basic color terms (cf. the late and rare Ancient Greek form killós ‘donkey-colored, gray’ from kíllos m. ‘donkey’). I see a further piece of evidence indicating the markedness of color terms derived from names of animals in the fact that, according to my data, these denominations present phenomena of analogy to color terms derived from names of plants more commonly than the other way round. Sardinian múrru ‘gray’, for example, is derived from Latin mūrīnus ‘mouse-colored’ (cf. Latin mūs, mūris m. ‘mouse’), but its syncope presupposes an analogy to Sardinian múrino ‘brown, dark’, derived from Vulgar Latin *mōrĭnus ‘having the color of the blackberry’ (cf. Sardinian múra f. ‘blackberry’). The source of an analogy is typically unmarked with respect to the target, which undergoes its influence.

2.9 Lexical sources of the main terms for orange

Similarly to what we have observed in the case of brown and other color categories, basic color terms for orange are also mainly drawn from names of plant products characterized by this color, although different plants may be taken as reference in different languages. In most IE languages, these terms derive from the name of the orange fruit, as in the case of English orange (24a). This fruit, and often also its correspondent color, presents different denominations, which are related either to Portugal (24b) or, less frequently, to the orange fruit as the ‘Chinese apple’ (24c).

The first pattern in (24a), which is more familiar in Central Europe, uses a name of the orange fruit that is originally borrowed from Dravidian, as this plant is native of South Asia (cf. Malayalam nāraŋŋa ‘citrus’). From Dravidian, it passed to Indic (cf. Sanskrit nāraṅga- m. ‘orange tree’), from Indic to Persian nārang, from Persian to Arabic nāranj, and from Arabic to the various European languages. In English, this term arrived from Old French, since the orange fruit, as well as the tinctures of this color, were a privilege of the Norman aristocracy. The second pattern, related to the name of Portugal, as in Modern Greek portokalís ‘orange’ (24b), is due to the fact that the Portuguese were prominent in the trade of oranges at the beginning of the modern era. (When the Turks were blocking the connection to the East through the Mediterranean, the Portuguese could avoid the impasse by circumnavigating South Africa.) According to my data, the lexicalization of the color orange based on the name of Portugal is especially common in the area of the Eastern Mediterranean and of Southeastern Europe. A variant of it appears in some modern Indic languages, as in Hindi santarī, where the color orange is named after the Portuguese city of Sintra, close to Lisbon. The third pattern, where orange is described as yellow as the Chinese apple, is attested in a certain number of North Germanic languages, as in Faroese appelsingult (24c). The Chinese apple refers to China as the origin of a sort of orange (the sweet orange or Citrus sinensis, cf. Bogushevskaya [2018]) or as a general expression for the East. More frequently, the Chinese apple is the expression used only for the fruit, while the color orange is named according to the dominant European pattern in (24a). In German, for example, the color orange is consistently orange or orangefarben, but the orange fruit and the orange tree may also be called Apfelsine f., in addition to Orange f. and Orangenbaum m. I identified this distribution in Germanic and in Baltic, so we are therefore dealing here with a Northern IE isogloss. Cf. also Dutch oranje ‘orange (color)’ vs. sinaasappel ‘orange (fruit)’, Danish orange ‘orange (color)’ vs. appelsin ‘orange (fruit)’, Swedish orange ‘orange (color)’ vs. Apelsin ‘orange (fruit)’; Lithuanian oránžinis ‘orange (color)’ vs. apelsinas ‘orange (fruit)’, Latvian oranžs ‘orange (color)’ vs. apelsīns ‘orange (fruit)’, etc. We may hypothesize that the speakers of these languages tried to find synchronically transparent compounded expressions, which manifest a relatively recent development, to motivate the expression of a fruit which was not widespread in their northern regions.^[21]

These lexicalization patterns often overlap in function or in form. Firstly, a language may present expressions drawn from more than one pattern, especially in Indo-Iranian. In Hindi, for example, santarī is used besides nāraṅgī for the color orange, and native speakers often hesitate as to which is more important or which are their semantic or pragmatic differences. Secondly, expressions of a certain pattern may represent formal combinations with another pattern. For example, a lexicalization such as Polish pomarańczowy ‘orange (color)’ (derived from Polish pomarańcza f. ‘orange fruit; orange tree’) presents a first member pom- which originally means ‘apple’ (Polish pomarańcza is borrowed from Italian pomarancia, an earlier variant of melarancia, where both pomo and mela mean ‘apple’, cf. French pomme). In this, color expressions such as Polish pomarańczowy combine the orange pattern in (24a) with the pattern of the Chinese apple in (24c). Similarly, Yiddish marants ‘orange (color)’ is a short form of pomerants.

In Indic, the Bengali basic color term komolā ‘orange’ is also based on the name of the orange fruit, which in this case, however, is so named for its pale-red color. Its lexical source is Old Indic kámala- ‘pale-red, rose colored’, also used as the name of the lotus flower (Nelumbium). I could also identify Indic denominations of orange based on the name of other plant products, especially saffron, which in Section 2.5 we have seen to be a common source for terms of yellow. Saffron pigments have a yellow-red color and therefore may provide a convenient example for various warm chromatic categories. As illustrated in (25), basic color terms for orange such as Gujarati kesarī and Marathi keśrī derive from their noun for saffron (Gujarati kesar, Marathi kēśar), a pan-Indic expression (cf. also Hindi and Nepali kesar ‘saffron’, etc.), which in other modern Indic languages brings about minor denominations of the color orange. It derives in its turn from the Old Indic form késara- n., denoting the filament of a plant (in addition to hair, on the basis of a metaphor between animal or human hair and vegetal fibers), since pigments of saffron are precisely produced with the stigmas and styles of the Crocus sativus, and not with its petals or leaves.^[22]

My data show that basic color terms for orange are among the most recent acquisitions of the color lexicon of the analyzed languages. This is in agreement with Berlin and Kay’s (1969) low ranking of orange on their universal evolutionary sequence reported in (1). The ancient IE languages usually do not have a basic color term for orange – either they do not lexicalize it at all, at least as far as we can observe from their documentation, or they only have marginal color expressions for it. This does not depend on the fact that the orange fruit was not popular in antiquity in the West,^[23] since many other referents characterized by this color exist in nature and therefore offer potential lexicalization patterns – a flame, the colors of a sunset, as well as many flowers, fruits, and vegetables, such as pumpkins, melons, etc. These vegetables were cultivated in Eurasia since antiquity. The Romans had specific names for various sorts of gourd (cucurbita f.) and melon (mēlō, -ōnis m.), for example.^[24] Still, they never used these names as sources for basic color terms. The Latin adjective flammeus could mean ‘flame-colored’ (from flamma f. ‘flame’). The expressions velum flammeum ‘flame-colored bridal veil’ and flammārius m. ‘artisan specialized in dyeing with orange color’ show that techniques of orange dyes existed. Still, flammeus remains a marginal color expression in Latin (and therefore decays in Romance). It mainly has a concrete meaning ‘flaming, fiery’ related to a fire. As we have seen in Section 2.5, terms such as Latin flāvus and fulvus, meaning ‘golden yellow, reddish yellow, flaxen-colored’ – again mainly poetic expressions, which are not continued in Romance – were chiefly used to describe hair color in the sense of ‘blond’ and therefore did not mean ‘orange’. Similarly, Ancient Greek forms such as krókeos ‘saffron-colored’ and krokóeis ‘id.’, derived from the name of saffron (Ancient Greek krókos m.), are rare and poetic. We may assume that the Ancients, at least in the IE world, were not particularly interested in the color orange and did not assign to it any particular symbolic meaning.

Another finding of my data is that all languages analyzed that do have basic color terms for orange – the modern IE languages – consistently use nominal lexical sources denoting plants or plant products, rather than the name of the flame, of the sun or of fire, for example. Their preference for names of plants matches with the denominal lexicalization patterns that we have observed above for other colors. However, their constant association with denominal word formation contrasts strikingly with that. We have seen that only a minority of the IE basic color terms for white, black, red, yellow, green, blue, brown, and gray are derived from the name of an object (in different proportions according to different colors). By contrast, when they are available, IE basic color terms for orange all turn out to be derived from the name of an object in my data. In principle, they could lexicalize orange by means of derivations from other color categories such as red or yellow – these patterns are cross-linguistically attested for basic color terms of orange. Tibetan mar ser ‘orange’, for example, literally means ‘red-yellow’. In IE, however, this lexicalization is only found for marginal denominations of orange, such as Modern Irish flannbui (lit. ‘bloodred-yellow’), which is less frequent and more contextually marked than Modern Irish oráiste in (24a) for this color category.

Among the denominal patterns attested, the one based on the name of the orange fruit largely prevails, within and outside of IE, clearly because of the influence of English, French, and other Western languages that have adopted this lexicalization strategy. Other semantic patterns, however, emerge, often with an areal distribution, as the derivation from the name of saffron which is so common in Indic. Similarly, the Sinhala basic color term tembili ‘orange’ derives from the name of the king coconut, Cocos nucifera, which is native to Sri Lanka and which is therefore particularly relevant to the experiential domain of its inhabitants. In this case as well, what is lexicalized is primarily what is of most interest to the speakers.

2.10 Lexical sources of the main terms for pink

The case of the color pink is similar to that of orange, in that the dominant semantic pattern derives from the name of a plant-based product originally coming from the East and therefore lexicalized at a relatively recent stage in most IE languages. Both orange and pink may be seen as kinds of light red, and the above-mentioned example of Bengali komolā ‘orange’ (Section 2.9) ultimately derived from Old Indic kámala- ‘pale-red, rose-colored’ indicates that these colors may be lexicalized by similar sources. In the case of pink, the vast majority of the IE languages have basic color terms originally derived from the name of the rose. On the one hand, in most modern IE languages of Europe, the forms at issue are inherited or borrowed from Latin rosa f. ‘rose’.^[25] In this group, illustrated in (26a), the denomination of the rose and of the color pink may be identical, as we can see in French rose, or may be slightly different but anyway transparently related, as in Dutch roos f. ‘rose’ and roze ‘pink’.

Armenian vardagowyn ‘pink’ goes back to a non-Latin source but presents the same semantic pattern, as it literally means ‘of the color (gowyn) of the rose (vard)’. In this case, the name of the rose is borrowed from an Old Iranian form *wr̥da- (cf. New Persian gol ‘flower’).^[27] On the other hand, in most IE languages of Asia, especially in Eastern Iranian and in Indic, basic color terms for pink go back to a Persian compounded structure which originally means ‘rose water’, as in (26b). In Hindi, for example, gulābī ‘rosy, pink’ is derived from gulāb ‘rose’, which is originally a compound from gul ‘flower’ and āb ‘water’ – the rose is seen as the flower par excellence. Consider that rose water, a liquid made of roses’ petals, especially developed during the Middle Ages in Persia. It is therefore understandable that a Persian lexical source is also recruited for that.

In New Persian, instead, the basic color term for pink is rather Suratī (also meaning ‘facial’), a derivate from Persian Surat ‘face; form, shape’ (ultimately from Arabic Sura ‘form, shape; picture’), which therefore literally means ‘face-colored’. This is the most idiosyncratic pattern for pink in my data, where most denominations of the color pink are drawn from names of plants or plant products. Even English pink, which is not related to English rose, and which has been also widely borrowed (cf. German and Danish pink, Afrikaans pienk; Welsh pinc, etc.), was originally the denomination of a flower, a sort of Dianthus. The situation of English is therefore not so different from the semantic pattern illustrated in (26a).^[28] Names of other plants are in fact attested as lexical source for pink, albeit minoritarian with respect to the rose. In Dhivehi, for example, pink is fiyāthoshi kula, that is ‘color of the onion peel’ (fiyā means ‘onion’; thoshi means ‘bark, peel’ and is often used as a suffix to denote plants or parts of plants, e.g., fanthoshi ‘woven coconut palm fronds’).

According to my data, basic color terms for pink may be also derived from structures which mean ‘light red’, ‘pale red’, ‘white red’, et similia, and which express the low chromatic saturation of the color pink. In IE, this pattern only emerges in scattered linguistic spots, especially in Northern Germanic and in Celtic, as in (27), where Icelandic bleikur ‘pink’ originally means ‘pale’, for example; it is etymologically related to German bleich ‘pale’. Danish lyserød and Irish bándearg are transparent compounds. The former means ‘light red’ (Danish lyse is a cognate with English light). The same holds true for Faroese ljósareyður. Irish bándearg, instead, means ‘white-red’ and consists of the Irish basic color terms for white (bán) and for red (dearg). We may compare the German compounded form rosarot ‘pink’.

The case of basic color terms for pink derived from the name of the rose (or rose water) and of similar flowers is, however, much more common than the case of basic color terms for pink represented by expressions such as light red, white-red, or pale. The former case has 51 out of 70 occurrences (=73 %) in my IE database, the latter has four occurrences (6 %). All these terms are morphologically derived or compounded expressions. This suggests that pink is not a primary color in the IE color lexicon. This result, which is consistent with the low position of pink in Berlin and Kay’s (1969) sequence, is also supported by diachronic evidence, as the ancient IE languages usually did not have a basic color term for pink (this occurs in 15 out of 70 languages, that is, in 21 % of the cases). No main expression of pink is attested in Hittite, Avestan, Old Persian, and Vedic, for example. Consequently, no specific root for pink can be reconstructed for PIE. The ancient IE languages can well describe referents characterized by pink or similar rosy shades, e.g., some flowers, the skin or the cheeks of a young woman, the colors of the dawn, etc. The dawn, in particular, is represented as a beautiful woman in IE mythology and is often associated to the color pink. In Homeric Greek, it is typically described with the compound hrodo-dáktulos ‘rosy-fingered’, which is, however, a complex word formation. Other forms such as Ancient Greek hrodóeis, hródeos, or hródinos (derived from Ancient Greek hródon n. ‘rose’) are also poetic expressions, and are usually meant in a concrete sense of material as ‘consisting of roses, decorated with roses’, rather than as expressions of the color pink. The same holds for Classical Sanskrit pāṭala- ‘pale red, pink, pallid’; m. ‘name of a flower (Bignonia Suaveolens, etc.); rose color’ and pāṭala-varṇa- ‘rose-colored’. They are not basic color terms.

2.11 Lexical sources of the main terms for purple or violet^[29]

In the West, purple has been a highly appreciated color since antiquity, as the production of a variety of purple (the Tyrian purple dye) from the mucus secreted by a kind of murex was particularly difficult and time-consuming to extract and was therefore very expensive and typically used in garments worn by high magistrates, kings, or bishops (cf. Dedekind 1898–1908). In fact, in many IE languages, the basic color term for purple derives from the name of the purple substance. In Ancient Greek, we have the noun porphúra f. ‘purple fish and the dye obtained from it’ and the adjective porphúreos ‘purple-colored’. Ancient Greek porphúra has been borrowed into Latin purpura f. ‘purple fish; purple color; purple cloth’ (from which the poetic adjective purpureus ‘purple-colored’ is derived). On the one hand, Latin purpura has been inherited by Sardinian púrpura and by other minor denominations of this color, such as Italian porpora, French pourpre, Occitan polpra, Catalan porpra, etc. On the other, Latin purpura has been early borrowed into various Germanic and Celtic languages. Nowadays, not only English purple, but also Modern Irish corcra and Welsh porffor represent basic color terms for purple (while in other IE languages, as in Baltic, borrowings such as Lithuanian purpurìnis and Latvian purpurs represent minor denomination of the color purple with respect to Lithuanian violètinis and Latvian violets, see below).

More commonly, basic color terms for purple or violet are drawn from names of flowers, fruits, or vegetables characterized by this color, according to my data. The most frequent pattern, at least in the IE languages of Europe, is a word formation on the basis of the name of the violet flower, as reported in (28). Most of the time we deal with a synchronically transparent structure, as in Romance, where the names of the violet flower and of the purple or violet color are identical or similar – but this is not always the case. Lithuanian uses violètinis for purple or violet and žibuoklė f. for the violet flower. Most of the time, as we have seen for the pink color category, the source is found in Latin, in this case in the Latin name viola f. ‘violet flower; violet color’, which has been more or less directly borrowed, in its bare form with suffixes, in other languages outside Romance. But this is not always the case either. Armenian manowšakagowyn ‘violet, purple’, clearly meaning ‘of the color (gowyn) of the violet (manowšak)’, is borrowed from Persian, as usual. New Persian banafš presents the same pattern, since it means both ‘violet flower’ and ‘violet-colored’.

Some languages present more than one expression for violet and purple. Albanian, for example, has both vjöllce and lejla for this color, showing a derivation from the name of lilac which is also frequent as a source of terms for purple in my data. Consider German, Yiddish and Swedish lila, Danish, Norwegian and Faroese lilla; Spanish and Catalan lila; Bulgarian liláv, etc. The name of the lilac flower and of its color reached the European languages through Arabic, where we have the basic color term lailkī ‘purple’, formed by means of a nisba derivation from the noun lailak ‘lilac’. Arabic, in its turn, has borrowed lailak from Persian. The name of the mallow flower, characterized by violet petals, is also documented as a source of denominations of purple in my data. Basic color terms such as Occitan malve ‘purple’, Romanian mov ‘id.’, Modern Greek mōv, mov ‘id.’ ultimately go back to Latin malva f. ‘mallows’. While Occitan is directly inherited from Latin, Modern Greek mōv, mov is a borrowing from (the spoken form of) French mauve; Romanian mov is also a borrowing from French (possibly through Greek. In French, however, mauve is only a minor color denomination, in addition to being the name of a plant – the French basic color term for this color category is rather violet). A similar pattern emerges in some Iranian languages, such as Pashto and Dari, whose main color term for purple is arghawānī, clearly derived from arghawān ‘mauve; flower of the Judas tree’. The same holds true for Tajik arġuvon. Sogdian already attests the form ’rγw’n (=argγwān) ‘purple’, ultimately borrowed from Akkadian argamanu through Aramaic argwānā, as purple dyes and purple garments were precious objects of trade in the Middle East and in Central Asia along the Silk Road.

The most common names of fruits appearing as lexical sources of terms for purple are those denoting kinds of berries and plums in my data. Spanish morado and Catalan morat derive from Spanish mora and Catalan móra, the name of the blackberry, inherited from Latin mōrum n. ‘mulberry, blackberry’ (in its plural form mōra). We are probably dealing with an originally substrate Mediterranean word denoting a dark berry, cf. also Ancient Greek móron n. ‘black mulberry, blackberry’ (borrowed into Turkish mor ‘purple, violet’), Armenian mor ‘blackberry’, etc. The latter is also a basic color term for purple. Classical Armenian cirani ‘purple, purple-colored’ derives from ciran ‘apricot’ (while the Modern Armenian compound tsiranagowyn ‘apricot-colored’ is an expression of orange). Another example of this pattern, connecting the purple color to the name of a fruit, is extremely frequent in the modern Indic languages (29). In this case, Old Indic jambu- f. ‘rose apple tree’ (denoting the Eugenia jambolana and similar plants growing in tropical landscapes) is continued in numerous Middle and Modern Indic forms denoting this tree or its characteristic purple fruit, often called jamun, jambul, or jambolan (cf. Pali jambu-, Prakrit jaṁbū-, Sindhi j̄amū̃, Panjabi jammū, Nepali jāmu, Bengali and Hindi jām, Gujarati jām / jã̄bu, Marathi j̈ã̄b(h), Konkani jã̄mba, Sinhala dam̆ba, etc.). From these and similar forms we have derived adjectives denoting varieties of purple or violet.

The same semantic pattern appears in Dhivehi dhan’bu kula ‘purple’, lit. ‘dhan’bu-colored’, where dhan’bu or dhan’bu gas is the Java plum (gas means ‘tree’ in Dhivehi and is often used as a suffix to denote specific trees). Similarly, Sinhala dam indicates both the color purple and a kind of berry or small plum, called dan (Syzygium caryophyllatum), which is typical of Sri Lanka.

Consider that the jamun plays a very important role in Indian cultural and religious traditions. Its tree is a massive plant, which can grow up to 30 m and live more than a century. Its purplish fruits could be eaten raw by hermits wandering in the forest. Crucial events of many stories, in both Indic and Dravidian literature, happen beside the rose apple tree. Buddha entered into the first stage of his Jhāna meditation while sitting under a jambu tree. In the Tamil tradition, the poetess Avvaiyar was also sitting under a jambu tree when she decided to retire from her literary work, and only Murugan made her change her mind, and so on. India itself is often called jambu-dvīpa- m., that is, ‘island of the jambu tree’. It is understandable that the Indic languages take their main terms of purple, as well as of other colors, from the names of those objects that are more relevant for their experiential field.

Similar principles underlie basic color terms for purple derived from names of other vegetables. Nepali pyājī ‘purple’, a basic color term in this language, expressing various shades from dark pink to blueish red, is clearly derived from the Nepali noun pyāja ‘onion’ (the latter being borrowed from Persian). We have seen in Section 2.10 that the name of the onion provides a source for the basic color term for pink in Dhivehi. Much more commonly in Indic, the color purple is coded by expressions of the aubergine, as in (30). In this case, an originally Dravidian denomination of the aubergine (cf. Malayalam vaṟutina) had been borrowed into Indic in antiquity (cf. the rare form Old Indic vātiṅgaṇa- m. ‘aubergine’). Its reflexes can be seen e.g., in Hindi baiṅgan ‘aubergine, eggplant’, Nepali baigun ‘id.’, Bengali begun ‘id.’, Kashmiri wã̄gun ‘id.’ etc., which have produced derived adjectives such as Hindi baiṅganī lit. ‘eggplant-colored’ for purple. As in many other Indic languages, Hindi presents both baiṅganī and jāmunī for purple. Kashmiri reinforces the name of the eggplant with rang ‘color’, which indicates the transparent and recent word formation.

Although purple or violet hues are also found in precious stones (e.g., the amethyst), names of gems only rarely represent the lexical sources of basic color terms for this color in my data. An example of this may be seen in Kashmiri lājward ‘purple’, which goes back to the name of the lapis lazuli. The latter, however, is more commonly used as a source of terms for blue, as we have seen in Section 2.6.

Alternatively, basic color terms for purple may be derived or compounded from terms used for other colors, such as red or blue, the components of this color category. From terms of red, in particular, we have Portuguese roxo ‘purple’, for example, derived from Latin russeus ‘reddish’. Similarly, Latvian purpursarkana ‘purple’ remotivates the name of purple with that of red (sar̂kans), although purpursarkana is not as frequent as violets in this language. From terms of blue, we can mention Sardinian biaittu (meaning ‘violet’ or ‘blue’ in different varieties of this language), which is borrowed from Old Italian biadetto ‘blueish’, itself borrowed from Germanic (cf. Section 2.6). Dutch paars ‘purple’, Afrikaans pers ‘id.’, and Frisian pears ‘id.’ are borrowed from Middle French minor denominations of blue such as pers / perse (ultimately going back to a Vulgar Latin form persus ‘related to Persia’, since purple as well as blue pigments came from the East). Sometimes both denominations of red and blue, or similar colors, co-occur in a compound to denote the color purple. Modern Icelandic fjólublár ‘purple, violet’ is a transparent compound from fjóla ‘violet’ and blár ‘blue’. Modern Breton glasruz ‘purple’ is a compound from glas ‘blue-green, blue’ and ruz ‘red’. Such compounds recall the pattern attested for pink in forms such as Modern Irish bándearg, as we have seen in (27). As in that case, these compounds as well especially emerge in the Northern IE languages of Europe.

As we have seen in Section 2.10 in the case of pink, for purple and violet as well basic color terms are mainly derived from the name of an object characterized by this color, as in (28)–(30). In my data, this occurs in 55 out of 70 cases (=79 %). In eight out of 70 cases (=11 %) we have derivations from other color denominations, such as red or blue, as in (31). In the remaining seven cases (=10 %), the languages analyzed do not present any color term for purple or violet. This occurs in most ancient IE languages (in Hittite, in Avestan, in Old Persian, in Vedic, in the Sabellic languages, in Old Norse, in Tocharian in my database) – but not in all of them. In Ancient Greek, for example, porphúreos can certainly be considered to be a basic color term, as it is very frequent since the stage of Homeric Greek, where it is even more common than erythrós ‘red’.^[30]

Moreover, among the basic color terms for purple with denominal formations, those derived from the name of a plant or plant product are definitely more frequent than those derived from the name of an animal or product such as the purple fish.^[31] The former pattern has 42 out of 55 occurrences (=76 %), the latter pattern has 11 out of 55 occurrences (=20 %). In the remaining two out of 55 occurrences (=4 %), the lexical source is the name of an inanimate object such as the lapis lazuli. I consider it significant that the name of the purple fish is a minor lexical source for the purple color in IE despite the influence of English, which has such a semantic pattern. This result can be understood if compared to the lexical sources of other color terms, such as pink, orange, etc. in which names of plants and plant products also predominate.