"Knowing" Words in Indo-European Languages (original) (raw)

The first systematic theory of the relationships between human languages began when Sir William Jones, "Oriental Jones," proposed in 1786 that Greek and Latin, the classical languages of Europe, and Sanskrit [_Sãskṛta_, ], the classical language of India, had all descended from a common source. The similarities between the languages had already been noted in 1768 by Gaston Cœurdoux, who informed the French Academy. The evidence for this came from (1) the structure of the languages -- Sanskrit grammar has detailed similarities to Greek (and, as would later be seen, Avestan), many similarities to Latin, and none to the Middle Eastern languages, like Hebrew, Arabic, or Turkish, interposed between Europe and India [note] -- and (2) the vocabulary of the languages.

Thus, "father" in English compares to Vater in German, pater in Latin, πατήρ, patêr in Greek, , pitṛ in Sanskrit, , pedar in Persian, etc. On the other hand, "father" in Hebrew is , ābh, and in Arabic is , ab, which hardly seem like any of the others -- they are Semitic languages. Similarly, "daughter" in English (with its mysterious "gh") compares to Tochter in German, θυγάτηρ, thugátêr in Greek, and , dokhtar in Persian. Latin uses a different root. Meanwhile, "daughter" in Arabic is , bint, which is also clearly a different root. As it happens, the "gh" in English, now silent, was once pronounced like the ch in German. Persian and German "daughter" differ only in accent, the voicing of the initial consonsant, and some vowel quality.

Jones's observation became the theory of "Indo-European" languages, and today the hypothetical language that would be the common source for all Indo-European languages is called "Proto-Indo-European" [note]. The following table shows a genealogy for two "knowing" roots, which in modern English turn up as "know" and "wit."

Words that are related to each other by descent from a common source are called "cognates." English "wise" and Sanskrit "veda" are thus cognates. Note that descent can become confused when words are subsequently borrowed. English has borrowed "idea" and "agnostic" from Greek, "video," "visa," and "cognition" from Latin, "vista" from Spanish, etc. Otherwise, we see variations on two roots, "*wid-" and "*gno-." "*wid-" has contributed mainly "seeing" words to the Romance languages, through Latin, but Greek and English have retained "knowing" meanings for both, even though the actual verb witan of Old English and witen of Middle English has become obsolete.

A curious instance in Greek of the use of both knowing roots is a phrase from the Bible. Genesis 2:9 speaks of the "tree of knowledge of good and evil" in the Garden of Eden. In the Greek text of the Septuagint, this is, τὸ ξύλον τοῦ εἰδέναι γνωστὸν καλοῦ καὶ πονηροῦ, tò ksýlon toû eidénai gnôstòn kaloû kaì poneroû. The literal translation here is "The tree [_tò ksýlon_] of the [_toû_] to-know [_eidénai_] knowledge [_gnôstòn_] of good/beautiful [_kaloû_] and evil [_kaì poneroû_]." So we have a perfect infinitive εἰδέναι, eidénai, "to know" (of οἶδα, oîda, "I know") and an object, an adjective in the neuter accusative, γνωστόν, gnôstón, "knowledge" (from γιγνώσκω, gignóskô, "I know") So we get a verbal knowing with "*wid-" but a substantive knowing with "*gno-." This is a very odd expression, especially when it translates only one word in Hebrew (, ha-da'at, "[the] knowledge") and is then translated into only one word in Latin (scientia, "knowledge," from a different root than the ones I have been considering). It is as though the translators (there are supposed to have been Seventy or Seventy-Two) wanted to cover all the bases.

A dualism of knowing words, however, is conspicuous in the languages of Western Europe. These languages share enough features that they might be called a Sprachbund, which means a group of languages, who may even be unrelated, that have nevertheless assimilated features from each other. In the table we have parallel knowing words from Greek, Latin, and Germanic and Romance languages. The citation forms in Greek and Latin traditionally are the first person singular of the verb. I have given that for Greek and Latin, but in Latin the infinitive is also shown, for comparison with the modern languages, where the citation form is generally the infinitive. Latin sapio/sapere is actually not really a knowing word. It means "to taste."

abstract	concrete
Greek
German	wissen	(er)kennen
Dutch	weten	kennen
English	to know
Latin	scio/scire(sapio/sapere)	nosco/noscerecognosco/cognoscere
Italian	sapere	conoscere
Spanish	saber	conocer
French	savoir	connaître

However, all the following words in Romance languages are derived from it; and sapiens, "wise," and sapientia, "wisdom," show that the Latin meaning is already drifting in that direction. The contrast with cognosco/cognoscere is otherwise attested with scio/scire (giving scientia, "knowledge").

English is the odd man out here, despite its being a Germanic language closely related to German, and despite its extensive borrowing and influence from French and Latin. Indeed, English still had the verb witen in Middle English, even as Dutch today retains weten. Why English has lost the verb and the distinction is curious, but I am unaware of any explanations for it. English must reach out to other expressions, "acquainted with," or "familiar with," for semantic equivalents to kennen or connaître [note].

Thus, native English speakers have difficulty learning to properly use the respective verbs wissen/kennen in German, savoir/connaître in French, or the corresponding words in the other Romance languages. It is therefore noteworthy how this is explained in English dictionaries for those languages. The Larousse Concise French English, English French Dictionary [Larousse, Paris, 2005] glosses the meaning of savoir for "know" as "gen," i.e. "general," and that of connaître as "person, place" [p.304]. This parallels The New Cassell's German Dictionary [Funk & Wagnalls, 1958, Cassell & Company, 1965], which glosses wissen for "know" as "a matter," and kennen as "a p. or th.," i.e. "a person or thing" [p.273]. The University of Chicago Spanish Dictionary [1948, Fourth Edition, 1987] glosses conocer for "know" as "to be acquainted with," and saber as "to have knowledge of, to know how to" [p.372].

In the table I have therefore characterized the application of one set of verbs as "abstract" and of the other as "concrete," since the reference of the latter is to objects that are persons, places, or things that are matters of (at least potential) direct acquaintance. The former would therefore tend to be the verb used for indirect speech, e.g. "I know that..."

I am not sure that this is actually the distinction that we see in Greek (which obviously was not part of a later Western European Sprachbund), where the Lexicon seems to distinguish the meaning of γιγνώσκω mainly with the addition of "learn" and "perceive" [Liddell and Scott, An Intermediate Greek-English Lexicon, Oxford, 1889, 1964, p.165]. That may parallel the distinction in German and the Romance languages, but is not stated in the same terms. On the other hand, we have the curiosity that in Greek there are two roots used that parallel the English words "holy" and "sacred" -- ἅγιος, hágios, and ἱερός, hierós, respectively -- where the former terms are used for concrete, individual application, and the later in more general and abstract senses. Thus, English makes a distinction that it fails to make in the knowing words, where with Greek I am not sure whether the sacred and knowing distinctions do match up. On the other hand, German, which makes the abstract/concrete distinction in its knowing words, only has one word for "holy" and "sacred": heilig. I have not seen an explanation for this, but I think that in fact it attests to the existence of a random semantic variation of a kind that some have found difficult to accept, for ideological reasons, in gender grammar and vocabulary.

The peculiar expression in the Septuagint that I discussed above, eidénai gnôstòn, "to know knowledge," which looks redundant when translated in the simplest way into English, might be more naturally translated into German or French as wissen Erkenntis or savoir connaissance, respectively. Whether this matches the intention in Greek, or exactly what it would even mean in German or French, is a matter calling for further consideration.

Another striking example of cognates across Indo-European languages are all the following words for "is" -- modern French and Persian pronunciation is given in brackets. By a series of simple steps, we see the relationship between "is" in English and ast in Persian.

English	German	French	Latin	Greek	Sanskrit	Persian
is	ist	est[ē]	est	esti	asti	æst[ē]
, hāl-ē šomā čētor æst[/ē],"How are you/what is your condition?"

A noteworthy variation in a comparison of these "is" words would be Spanish, where the third person singular takes two forms, es and esta. These are from different verbs, ser and estar, respectively, with the former expressing permanent or innate attributes and the latter temporary ones (or location, even if permanent). There is no precedent for this in Latin or parallel in other Romance languages like French or Italian; so one might wonder about its origin.

The answer may be that Spanish belongs to its own smaller, Iberian Sprachbund. Mediaeval Spanish grew up, in the north of Spain, in close proximity to Basque, an autochthonous language that is not of Indo-European origin. Basque, as it happens, also has two, or even three, verbs that mean "to be," izan, egon, and ibili. I don't think that the meaning of these matches very well with ser and estar. Their very existence, however, is intriguing and suggestive, even as other influences of Basque on Spanish are clear, such as the -ez/es patronymic ending. Izan has the principal meaning of "to be, exist" [Gorka Aulestia, Basque-English Dictionary, University of Nevada Press, 1089, p.324], while egon is "to be, to consist of, to stay, to remain," or "to reside, to dwell" and "to wait" [p.157]. The primary meaning of ibili is "to walk, to move, to pass," etc. or "to be, to stay" [p.291]. Izan can also mean "to have" when it is used as an auxiliary with intransitive verbs, corresponding to ukan, "to have" [p.515], which is used as an auxiliary with transitive verbs. I see no hint in these definitions of a contrast between permanent and temporary attributes, although the meaning of egon as "to reside, to dwell" might suggest the locative uses of estar. The abundance alone of this vocabulary in Basque may be all the influence Spanish needed, meanings may have subsequently changed, or ser and estar may be one of those things in language that just develops without explanation.

Traditionally, all Indo-European languages were divided into "centum" and "satǝm" languages, after the Latin and Avestan words for "100," respectively. This is an "isogloss" (like an "isotherm" or "isobar" in meteorology) that distinguishes languages where, in certain environments, an Indo-European k has remained a k and where it has turned into an s or ch (and g to j, etc.), that is, velars are palatalized into sibilants or affricatives (e.g. Latin rex/regis, "king," Sanskrit raja). Most importantly, the Indo-Iranian (Sanskrit, Persian, etc.) and Slavic languages are "satǝm" languages. However, this particular isogloss is now no longer taken to reflect a fundamental division in descent. In the chart above, Russian, the principal Slavic language, will be seen to be more closely related to German and to Latin than to Sanskrit; and Greek, a "centum" language, is more closely related to Sanskrit (perhaps) than to the others. What has happened is that more features have been taken into account and the overall greater similarities between Greek and Sanskrit outweigh a lesser point that Sanskrit seems to share with Slavic languages. On the other hand, the whole picture of branching descent, while perhaps appropriate for organic evolution, may not be as appropriate for languages, which can borrow features from even unrelated languages in geographical proximity. The Slavic and Indo-Iranian languages, because of their geographical proximity (in Southern Russia), thus may well have shared a certain sound change, even while retaining closer affinities to other groups.

The following chart demonstrates a way other than descent to look at the relationships of these languages. I originally saw a diagram like this when I took an Indo-European linguistics class with Raimo Anttila at UCLA in 1970. I recently found a similar diagram in The Oxford Introduction to Proto-Indo-European and the Proto-Indo-European World by J.P. Mallory and D.Q. Adams [2006, p.73]. Unfortunately, Mallory and Adams actually do not discuss the individual isoglosses. The present diagram is thus based on one by Thomas Pyles and John Algeo [1993], though I have added the tenth gloss for the reason given below.

What we see here looks very much like a dialect map of languages that occur near each other and so exchange influences with adjacent languages. The theory that goes with it is called the "wave model," that innovations spread out across the field like waves in a pond (as also in a Sprachbund). The line marked #1 in red surrounds the Satǝm languages. The line marked #2 in blue surrounds Greek and the Italic languages (like Latin), where we have voiceless sounds for Indo-European voiced aspirates, i.e. ph in Greek and f in Latin for Indo-European bh (Germanic languages have b). The line marked #3 in light green surrounds the Italic and Celtic languages, which have passive forms of the verb in -r, e.g. Latin laudor, "I am praised" (active laudū). The line marked #4 in light purple surrounds the "North-West" group of languages, which share some common vocabulary that does not occur elsewhere among Indo-European languages. The line marked #5 in dark green surrounds the south-eastern languages that have a prefixed vowel in the past tense or aorist, e.g. Greek ἔλιπον, élipon, "I left" (present λείπω, leípô). The line marked #6 in gray surrounds northern languages where (according to Pyles and Algeo) "medial schwa [an indefinite vowel, traditionally written "ǝ"] was lost." The line marked #7 in orange surrounds the western languages that share some common vocabulary not found elsewhere. The line marked #8 in light blue surrounds northern languages that have a dative plural in -m, e.g. Gothic dagam, "to/for days" (nominative singular dags, dative singular daga -- Modern German now has -n in the dative plural, den Tagen, but -m in the [masculine/neuter] singular, dem Tag), or Russian dnyam, "to/for days" (nominative singular dyenь [with the final "soft" sign], dative singular dnyu). The line marked #9 in dark purple surrounds the Indo-Iranian languages, i.e. the Indic and Iranian, where (according to Pyles and Algeo) "schwa became i" -- though there are many features that unite the Indo-Iranian group, including vocabulary items, e.g. the god Mitra in Sanskrit and Miθra in Iranian (Avestan, Persian). Finally, the line marked #10 in yellow surrounds Greek and Armenian, where Mallory and Adams say, "[T]here were close contact relations between Greek and Armenian" [p.79].

In a dialect map, we are usually looking at variations across a language that geographically stays in place. With the diagram for the Indo-European languages, we may be looking at fossil evidence of when the languages were dialects of a language in a particular geographical area, probably Eastern Europe, stretching down into the Balkans and out into the Ukraine. From the Ukraine, the Indo-Iranian group took off across the Steppe (following Tocharian). Once separated, the language groups can experience changes that will not be reflected in any other related languages, for instance that the Indic group acquires the retroflex consonants that figure in the unrelated Dravidian languages but not elsewhere in Indo-European, or that New Persian (like Urdu) borrows a large vocabulary from Arabic, a consequence of Iranians converting to Islam.

The absence of Tocharian and the Anatolian languages (Hittite, Luvian, etc.) from the diagram is significant. Tocharian, from people who advanced across the Steppe all the way to China and ultimately show up in India as the Kushans, could be expected to orginate from the east side of the language community and thus most likely be a Satǝm language. But it wasn't. It thus may well be that Tocharian speakers left the dialect area before palatalization occurred in the Satǝm languages. Hittite, the earliest attested Indo-European language, and its related Anatolian languages, seem to have left the dialect area even before Tocharian. Hittite retains very archaic features of Indo-European, like laryngeals (or pharyngeals, though exactly what these were is still unclear -- they would be like sounds that still exist in Arabic, and are to be found the earliest in Ancient Egyptian), but then it is missing many features that may have developed later in the dialect area.

The Caucasus-Indo-European Connection

The Alphabet and Phonology of Armenian

Greek, Sanskrit, and Closely Related Languages

Dialects of Greek

The Pronunciation of Greek

Tense and Aspect in Greek

The Celtic Languages

The Spread of Indo-European and Turkish Peoples off the Steppe

The Germanic Languages

The Slavic Languages

Philosophy of Science, Linguistics

Philosophy of Science

Philosophy of History

History of Philosophy, Indian Philosophy

History of Philosophy

Home Page

Copyright (c) 1998, 2000, 2008, 2010, 2011, 2012, 2014, 2015, 2016, 2019, 2020 Kelley L. Ross, Ph.D. All Rights Reserved

"Knowing" Words in Indo-European Languages, Note 1;

The Case System of Nouns

A conspicuous feature of Indo-European grammar is the original extensive inflection of nouns and verbs. In the table are the cases that occur in the inflection of nouns and adjectives in a selection of Indo-European languages.

The vocative (Voc) occurs when someone is being addressed -- which is why Shakespeare has Caesar say Brute rather than Brutus when addressing Brutus (although Suetonius actually has him speaking Greek -- καὶ σὺ τέκνον; kaì sù téknon? "And you child?" -- where "child" is neuter, with a vocative and nominative identical). The nominative (Nom) is the subject of a sentence. The genitive (Gen) can mean possession, "of" or "from." The accusative (Acc) is the direct object of a sentence or motion towards. The dative (Dat) is the indirect object or means "to" or "for." The ablative (Abl) means "from" or motion away from. The instrumental (Ins) is the agent for the passive voice or the means. And the locative (Loc) means "at" or the location of something.

All these languages actively inflect nouns and adjectives for case, gender, and number, except English, where there is only a remant of the system, mainly in the pronouns. Thus, he/his/him, she/her/her, and it/its/it, give us the most complete inflection that English still possesses. Sanskrit, on the other hand, retained nearly the full Proto-Indo-European system, including inflection for the dual number (like Greek) as well as the singular and plural. Except for the vocative, German still has the same cases as Greek, but there is a great deal of ambiguity in the case endings, whose identity must often be determined from context. See the discussion of Nietzsche's language.

Sumerian
Absolutive,Vocative	[none]
Ergative	-e
Genitive	-ak
Locative	-a
Dative	-ra
Comitative	-da
Ablative,Instrumental	-ta
Terminative	-è
Directive	-e
Equative	-gin7

As prepositions come to be used more extensively, they can have different meanings when used with different cases, or they can be fixed to take a particular case, which happens a lot in German. In English, all prepositions simply take the accusative, though in usage people are often confused and use the nominative "I" with prepositions after a conjunction (e.g. "between you and I").

It is always important to keep in mind, not only what something is, but what it isn't. Indo-European languages, with cases like nominative and accusative, are not "ergative" languages, like Basque, languages in the Caucasus, or Sumerian (which beats out Sanskrit with ten cases for its nouns, as seen at right).

In an ergative language, the subject of an intransitive verb and the object of a transitive verb take the same case, the "absolutive." The subject of a transitive verb then takes the ergative case. While this all seems strange, the division is natural enough. Only the subject of the transitive verb is actually doing something (Greek érgon is "work") to something else. The difference between nominative-accusative languages and ergative-absolutive serves to mark fundamental differences in language families.

Even Sumerian does not get the prize for number of cases. I don't know if the closely related Finnish and Estonian do either, but they beat out both Sanskrit and Sumerian with fourteen cases (only a dozen may be commonly used). Finnish and Estonian's relationships are better known than Sumerian's. They are Uralic languages, close to Hungarian. These are not ergative languages and have a nominative case. Suprisingly, we don't see an accusative, but there is functionally an accusative, which may be identical to the nominative or genitive in form. Some of the other cases match names with Sumerian (terminative, comitative), but we are also missing a dative and a locative, which we might otherwise expect. The functions of these are distributed among the cases that we see. The contrast is with languages like English, where the semantic distribution is found in the prepositions, not in the inflection of the noun.

Return to Text

"Knowing" Words in Indo-European Languages, Note 2

On the topic of "father," there is the curious circumstance that the word for "father" is atta in Gothic, attas in Hittite, ata (Ottoman ) in Turkish, and ada or adda in Sumerian. These are enough alike to suggest a relationship, perhaps with a common origin, as inferred from the examples above. These languages, however, are largely unrelated to each other. Hittite and Gothic are Indo-European, with Gothic in the Germanic family, like English. Turkish is in the unrelated Altaic family; and Sumerian belongs to the curious group of isolated and unrelated language families, part of which survives today in the Caucasus.

Since few in linguistics are willing to take the plunge and say that this suggests a common origin for Indo-European, Altaic, and perhaps some Caucasian languages (although Turkish nationalists like the idea that the Hittites had something to do with the Turks), some explanation is called for. But only speculation is possible. What is commonly asserted is that these similar words come from "nursery" or "infantile" language, which apparently is the claim that "ata," or the like, is a sound that all infants make and thus can easily and sponaneously enter any languages whatsoever. Much the same, however, might be said for "mama" and "papa"; and so, however appealing, it is not enough rule out more daring conjectures, for instance that the languages are in fact related, or at least have borrowed from each other because of proximity. For instance, Attila, apparently from a group, the Huns, speaking an Altaic language, rode into Europe bearing a name in Gothic, a diminutive, "Little Father." He did not get that from "infantile language," but from borrowing. Or perhaps the Hunnic word for "father" is like that in Turkish, and so only the diminutive ending, "-ila," is borrowed. We will probably never know.

Return to Text

"Knowing" Words in Indo-European Languages, Note 3;

The Conjugation of Witan and Witen in English

The paradigms of Old English witan and Middle English witen in the indicative mood we see at left and right, respectively. While it may be tempting to pronounce woot so that it rhymes with Modern English boot, the Great English Vowel Shift has not yet happened, and the vowel written "oo" is still a long "o," as in Modern English boat, or, for that matter, still Modern German boot, which is, and is pronounced (more or less) like English boat.

As these forms have been forgotten in Modern English and seem like a foreign language, it is notable that they were already very different from their cognates in German. Modern German present "know" is singular 1st & 3rd person weiß and 2nd person weißt. The past/imperfect "knew" is singular 1st & 3rd person wußte and 2nd person wußtest. The similarities are clear, but we also have considerable differences, especially in the vowels. Dutch, on the other hand, has simplified the paradigm of weten considerably. All the indicative present singular forms are weet, the plural weten. The indicative past singular forms are wist, the plural wisten -- much like Middle English (except in the second person).

A nice touch in Middle English is that, just as Latin had volo, "I will," and a corresponding negative, nūlo, "I won't," in Middle English woot, "I know," has a corresponding negative in noot, "I don't know." This economy of expression has been lost in the Modern language. Again, it is bewildering why this vocabulary stopped being used, especially with the loss of the wissen/(er)kennen distinction otherwise universal in the Western European Sprachbund.

Of course, it is unusual that vanishing words vanish completely. The Webster's New Collegiate Dictiontiary of 1980 still lists a verb, "wit, wist, witting," which is "archaic," meaning "know" or "learn." In the 1st & 3rd person singular this is still "wot," which also has a separate entry as an alternative verb with principle parts, "wot, wotted, wotting," noted as "chiefly Brit." Nevertheless, I have never seen these words used either in speech or writing, American or British. Of course, all sorts of strange things go on in English dialects, so there is no telling. And doubtless the Oxford English Dictionary chronicles various interesting uses. My wife remembers seeing the expression "I wot not of it" someplace, whose principal interest I see in the circumstance that Middle English noot could not pass into Modern English without being confused with the simple negative not -- we also see the archaic absence of an auxiliary verb here, which is necessary for Modern English. Meanwhile, woot itself experienced the Vowel Shift in an unpredictable way, as did book (Old English būc).

Return to Text

Greek, Sanskrit,

and Closely Related Languages

The Sanscrit language, whatever be its antiquity, is of a wonderful structure; more perfect than the Greek, more copious than the Latin, and more exquisitely refined than either, yet bearing to both of them a stronger affinity, both in the roots of verbs and in the forms of grammar, than could possibly have been produced by accident; so strong, indeed, that no philologer could examine them all three, without believing them to have sprung from some common source, which, perhaps, no longer exists; there is a similar reason, though not quite so forcible, for supposing that both the Gothick and Celtick, though blended with a very different idiom, had the same origin with the Sanscrit; and the old Persian might be added to the same family.

Sir William Jones (1746-1794), speaking to the Asiatick Society in Calcutta, February 2, 1786.

The following chart zeroes in on the relationship between Greek and Sanskrit [], with the closely related Iranian and other Indo-European steppe languages, and the modern descendants of them all. Greek can be seen to radiate into a number of dialects, later to be consolidated into the koinê or "common" dialect of the Hellenistic period.

The name Yuèzhī, "Moon Tribe," was given by the Chinese to an Indo-European group who came off the eastern end of the steppe. Latter, under pressure of Turkish or Mongol peoples -- especially a defeat by the Hsiung-nu in 170 BC -- they fell back into the Tarim Basin (the "Lesser" Yuèzhī, ) and Transoxania (the "Greater" Yuèzhī, ). The latter eventually descended into India, as the Kushans (1st century AD). The texts that survive in the Tarim Basin, in languages usually called "Tocharian," attest this obscure branch of Indo-European [note]. The Iranian group of languages also includes that of a people, the Saka, who had previously (1st century BC) also ended up in India, providing the benchmark historical era for India (79 AD).

Otherwise we see several modern descendants of Iranian languages, from Modern Persian and Kurdish all the way to the unique survivor of the North Eastern group, Ossetian, in the Caucasus (though this is now North West of the others). Iazyges were settled in Britain by Marcus Aurelius, and Alans spread across Gaul and Spain after crossing the Rhine in 407 AD. Although students of both Greek and Latin may be impressed with their similarities, Latin does not have a dual number, a middle voice, or an aorist tense, which both Greek and Sanskrit share. These features, and others, draw Greek away from Latin, to be more closely associated with the Indo-Iranian languages. In general, this is the most conservative branch of the Indo-European languages. My Indo-European linguistics professor at UCLA said once that you can get a sort of "instant Proto-Indo-European" by combining Greek vowels and Sanskrit consonants.

East of the Caspian Sea, the Indo-Iranian group of languages came down into the Middle East and India. The furthest penetration west into the Middle East was by the Mitanni, who provide the earliest texts using Vedic gods and other Indo-European words. The Mitanni, however, do not last all that long, and it is Persian and Avestan (the language of the Zoroastrian book, the Avesta) that produce most of the Indo-Iranian inscriptions and literature. A difference in pronunciation of the name of the Vedic god Mitra is indicated in the chart, between India, the Mitanni, and Persian. Meanwhile, the Ārya had descended into India, c.1500 BC, the first Indo-European group to do so (before the Sakas & Kushans). As discussed elsewhere, the Ārya plunged India into its Dark Ages, until around 800 BC, when an alphabet was borrowed from the Middle East.

The map shows the present distribution of the Indo-Iranian languages, from Kurdistan to Sri Lanka. Ossetic (Ossetian) is all the remains of the former Iranian presence on the Steppe, being derived from Scythian and Alan, which used to dominate the European Steppe in and around the Ukraine. The Sakas, who were on the Asiatic Steppe, are long gone, though their invasion of India is remembered there. The Dravidian languages, which are not Indo-European, are shown because their outliers bespeak their former presence in the North, as well as the South, of India, while features of Dravidian languages (like the retroflex sounds) influenced the Indic languages, starting with Sanskrit itself.

The word ārya, which later simply meant "noble" in Sanskrit, was of course used in European theories of the "master race," the "Aryans" -- as we even see in the writings of Friedrich Nietzsche. This had one curious consequence. Airya was the form of the same word in Avestan, and Irān is its modern Persian descendant. When Shāh Rezā Pahlavi heard that the "Aryans" were supposed to be the master race, he thought, "Hey! That's us!" The official name of his country was then changed from Persia to Irān. This ended up being an unfortunate move for him. In World War II, he was more than a little sympathetic for the "Aryans" of Nazi Germany, and the result was that he got overthrown and Irān was occupied by British and Russian forces.

In the Indian Dark Ages, a sacred oral literature developed, the Vedas. The language of the Vedas can then be called the Vedic language, and Indian history from c.1500 down to c.400 BC can be called the Vedic Period. Even though the Vedas could be written down after 800 BC, they have always been taught and remembered orally, and have always been thought of as essentially sound -- in contrast to Jewish beliefs about the Tūrah and Moslem beliefs about the Qurʾān, that they were essentially written. The Vedas are still taught orally.

Once the Vedas came to be written, a disturbing thing was soon noticed. The spoken language was diverging from the written language. Language, indeed, changes all the time, but this may not be noticed in an oral tradition. When it was noticed, the reaction was horror, for the belief was that the Vedas had to be remembered with absolute accuracy for them to be ritually effective. The result was an effort to describe and fix the language of the Vedas so that it would never change again. The process culminated about 400 BC with the grammar of Pāṇini.

The language that resulted was tidied up a bit and not precisely identical to the surviving language of the Vedas. It was called Sãskṛta, , Sanskrit, which means "prepared," "cultivated," "polished," "correct." The language based on Pāṇini can be called "Classical Sanskrit," and that of the Vedas "Vedic Sanskrit." Classical Sanskrit remained the language of religion, philosophy, and high literature in India for centuries, and survives today as the indispensible language of religion and serious scholarship.

Meanwhile, the spoken language had not only changed but split up into dialects that eventually grew into separate languages. These new spoken languages are called "Prakrits," from Prākṛta, "natural," "ordinary," "common," "vulgar." The first examples of written Prakrit words are in Sanskrit texts where someone is speaking, e.g. from a Once Born caste, who is not allowed to speak Sanskrit. Eventually, however, some Prakrits developed their own literature. When the canon of essential Buddhist texts was set down in Sri Lanka, the Prakrit Pāli was used -- hence the "Pāli Canon." That has suggested to some that the Buddha himself spoke Pāli, but this does not seem to have been the case. The Buddha probably spoke Māgadhī.

From the Prakrits, most of the modern languages of India are derived. The exceptions are the languages of the Dravidian group, largely spoken in the south. Some examples of Dravidian languages, and discussion of the relationship of Hindi to Urdu, can be found elsewhere.

The oldest alphabet used in India was the Brāhmī script. Later, other alphabets developed, like Kharoṣṭhi; but Sanskrit is written in an alphabet especially designed by the grammarians for it: Devanāgarī. This is also used with some modern languages, like Hindi, and is the source for many more, including the alphabets for Burmese, Thai, and Cambodian. Actually, Devanāgarī is not a true alphabet but a syllabary. It writes syllables, and it does so on the basis of a couple of odd conventions. For one thing, even though Sanskrit has many consonant clusters, every syllable is written ending with a vowel. This means that all the consonants, even ones from preceding words, are piled on to the beginning of the following syllable.

The word Sanskrit itself has three syllables. Most Devanāgarī letters have a horizontal line on top and a vertical line at the right. The plain form for each letter automatically is read with the vowel a. In the word at right, therefore, reading from left to right, we first have the letter s, which is read sa. Over it is a dot, transcribed as an "m" with an underdot, which stands for the nasal sound found as the "n" in the French word on [/õ/]. This is very common in Sanskrit. The second syllable in the word is skṛ, where the r is given an underdot to show that it is a vowel. Both "r" and "l" can be vowels in Sanskrit -- though no longer in Hindi (ṛ is prounced ri). The basic form of the syllable is the letter k. Attached to the front of it is the letter s, which we've already seen, without its vertical stroke, and under it is attached a hook that indicates the vowel ṛ. For the final syllable we write t, which is given the vowel a. A short final a, it should be noted, is not pronounced in Hindi: thus, Sanskrit words like yoga and names like Arjuna can now actually be found pronounced yog and Arjun.

Another Sanskrit word to consider might be that for the supreme Being of the Upanishads: Brahman. Here there are two syllables and a final consonant. In inflection, the final n is ordinarily going to be lost or written with the following syllable; but we can add a diacritic to show that it is without a vowel. In the first syllable, bra, there is a little complication. R, even when it is a consonant and not a vowel, is written more like a vowel, with a diacritic. The basic form of b is a loop with a line through it. The r is indicated with a diagonal stroke attached to the bottom of the loop. The vowel a is then understood. An r that precedes, rather than follows, another consonant, is written with a hook at the top of the letter. The second syllable, hma, poses another problem. H is one of the letters that does not have a vertical line at the right, as it is shown written independently below Brahman. Combining h with m requires running them together, as shown. The form of this combination is conventional and cannot always be predicted. It must simply be learned. The full form of m can be seen in the next example, below. Finally, the absence of a vowel on the final n is indicated with the diagonal stroke at the bottom of the vertical line.

Next, we can examine a whole sentence. This is the famous tat tvam asi, "Thou are that," one of the four Great Sentences of the Upanishads. This consists of three words, but four syllables, where the final consonant in the first two words is attached to the first syllable of the following word. Ta is familiar. The second syllable, ttva, involves a conventional combination. When two t's are stacked on each other, one straightens out into a horizonal line. This can be seen in the tta combination given below the sentence. Va itself is just a loop, like b without the line through it (the similarity is no accident; v and b were both recognized as "labials," i.e. letters that use the lips). The third syllable is ma, where we simply write the form for m, with the understood vowel. Finally, the form for s is familar, but this time we must indicate that it has the vowel i rather than the vowel a. This is done by adding another vertical line to the left of the letter and connecting it to the letter with the loop at the top.

Finally, we might consider the sacred syllable Om, as found in the Māṇḍūkya Upaniṣad. Here, at left, we have the independent form of the letter a with a diacritic (vertical line and stroke) indicating that it has the vowel o (originally au). M follows with the diacritic indicating no vowel. A more compact form of the word, however, can be written. If the m is considered to be the nasalized m., it can simply be written with a dot over the o. The m is a real m, but everybody knows that anyway, so the more compact form can be written for convenience.

Since the syllable Om is written down frequently, for good luck and as a blessing, it is not surprising that abbreviated forms have developed. In the one at right preserves recognizable parts of the fully written (though already reduced) form.

Some more examples of Devanāgarī writing can be seen in the essay on karma.

In many Sanskrit words, like the name of the Māṇḍūkya Upaniṣad, it will be noticed that the letters ṭ, ḍ, ṇ, and ṣ may have underdots. These are a separate order of letters from ordinary t, d, n, and s. The ordinary t, etc. are what in linguistics are called "dentals," because the tongue touches the teeth (#1 in the diagram). The underdot ṭ, etc., are called "retroflexes," because the tongue curls up towards the roof of the mouth (#3 in the diagram). This makes for very distinctive sounds, which Sanskrit and the descendants of the Vedic language share with Dravidian languages, but not with any other Indo-European languages. Curiously, t, d, and n in English are not true dentals. The tongue touches the gums above the teeth, the alveolus, rather than the teeth (#2 in the diagram). This makes them "alveolars" rather than dentals. In India, this sounded to people more like the retroflexes than like the dentals. English words borrowed into Hindi, like "doctor," are thus pronounced with the retroflexes -- ḍocṭor. At the same time, Hindi has lost separate, phonemic ṇ and ṣ sounds. Ṇ occurs as a dental n, and ṣ occurs as an ordinary palatal sh (often written for Sanskrit as an ś with an acute accent on it). The name of Krishna in Sanskrit is Kṛṣṇa, but this then is just pronounced in Hindi as, of all things, Krishna.

At right is the entire Devanāgarī syllabary. In an alphabet invented by grammarians, it is not surprising to see it laid out according to phonetic principles. Thus, the alphabetical order begins with the vowels, then runs through the diphthongs, the stops, the semi-vowels, the sibilants, and finally h. The vowels, when syllabic, have independent forms; when not, they are, as we have seen, indicated with diacritics.

The stops, which means sounds where the vocal tract closes, pose some pronunciation challenges. K is pronounced as in English skit, and kh as in English kit. This is the difference between an unaspirated and an aspirated stop -- one has no breath coming out, the other does. Similarly, t is pronounced as in English stop, and th as in English top. The "th" sounds in English "thin" or "that" do not occur in Sanskrit. P is pronounced as in English spot, and ph as in English pot. "Ph" is never pronounced f. Sanskrit c is like the ch in English, but is unaspirated, making it unfamiliar. The voiced stops (g, j, d, ḍ, & b), where vocal chords vibrate, all also have their corresponding aspirates. In sounds like gh, jh, etc., however, the breath coming out is also voiced. Consequently, the voiced "aspirates" are also called murmur stops, since the sound is more like murmuring than breathing. These are sounds rarely seen in other world languages.

Several of these phonetic characteristics of Sanskrit can also be found in the (unrelated) Mandarin Chinese. Notice that "swastika" is a word from Sanskrit (svastika). In the Nazi version, the top bar points to the right. In India, or in Buddhism, the top bar tends to point to the left, but traditionally this is not always the case and both right and left handed swastikas can be found. It was not just a coincidence that the Nazis liked this symbol. They saw themselves as the heirs of the Ārya.

A curious foodnote to the languages of India is one that made its way to Europe. This was the language of the Gypsies, whose origin was long a mystery to Europeans, and after a while perhaps even to themselves. We begin to hear about them in Roman, i.e. Byzantine, sources as early as the 11th century, and then they appear in direct reports as close as Crete, the Morea, or Corfu by the 14th. Confusion about who they are comes early, and the most familiar name for them is Αἴγυπτοι, i.e. "Egyptians," shortened to Γύπτοι, "Gypsies," in Modern Greek. Sometimes the Gypsies even thought of themselves as Egyptians; but Greek sources also confused them with other peoples, so various names were used.

This wandering people were characterized as magicians, snake charmers, fortunetellers, beggars, liars, and thieves. This identity stuck to them as they moved across Europe, creating a lot of local hostility, and giving us words in English like "gyp," i.e. "cheat." Although "gypsy" now mainly means someone who moves around a lot, and can be neutral, positive, or even charming in its connotation, politically-correct scholars now prefer to use the term they have recently been using for themselves, which is "Roma" for the people and "Romany" (or "Romani") for their language.

This adds to the confusion, since, as the people had picked up an Egyptian identity perhaps from the transit of some through Egypt, these names make it look like they eventually picked up a name from Ῥωμανία, Romania, the Mediaeval Roman Empire. Othewise, "Roma" would be the name of the actual City of Rome, while "Romani" would itself mean "Romans."

Until a few Europeans had been to India, and became a bit familiar with the languages there, there would be no way to otherwise correctly identify the language of the Gypsies as Indian, as eventually it was. Since "Roma" is a term from their own language, an origin can be sought for it in India also. Nevertheless, considerable uncertainty remains. It looks like the term may ultimately derive from Sanskrit , a caste of musicians. This comes down into Hindi as , which seems to be for more than one caste, including musicians, dancers, rope and basket makers, and "workers at cremation grounds." Several of these are suggestive, since it would be characteristic of performers, or even of the rope makers, that they would move around, to offer their services or wares -- perhaps including the instruments that musicians would necessarily make for themselves -- in various places.

When we realize that by a minor phonetic change becomes (and so ), the matter becomes suggestive. There is nothing to stop itinerant performers or peddlers from at some point moving themselves right out of India, as some of these obviously did. And after Mahmūd of Ghazna invaded Indian in 1001 AD, we can imagine more regular protected transit into and out of India, through the domain of the Ghaznavids. This would soon get the wandering into areas where they would encounter Christians and Europeans at the times when we hear about them.

Thus, "Roma" and "Romany" have nothing to do with Ῥωμανία, unless, of course, amid all the uncertainties and confusions of these names, they do. Meanwhile, I have heard of a tradition that the languages of North-Western India are the "Seven Sisters": Punjabi, Kashmiri, Gujarati, Marathi, Sindhi, Hindi, and, last but not least, Romany. I don't know the provenance of this idea.

Classical Languages

Katy Perry's Sanskrit Tattoo

Preston and Child quote Aeschylus

"Knowing" Words in Indo-European Languages

The Caucasus-Indo-European Connection

Dialects of Greek

The Pronunciation of Greek

Tense and Aspect in Greek

The Celtic Languages

The Spread of Indo-European and Turkish Peoples off the Steppe

The Germanic Languages

Philosophy of Science, Linguistics

Philosophy of Science

History of Philosophy, Indian Philosophy

History of Philosophy

Home Page

Copyright (c) 2000, 2003, 2005, 2008, 2009, 2010, 2012, 2017, 2019, 2021 Kelley L. Ross, Ph.D. All Rights Reserved

Greek, Sanskrit, and Closely Related Languages, Note 2

The pop singer Katy Perry ("I kissed a girl," among other songs) has been photographed sporting a tattoo on her arm: . This is supposed to translate, "Go with the flow." It transcribes as, anugacchatu pravāham. Literally, it is "Go," gaccha, "after," anu, the "stream," pravātam (in the accusative case). "Go after" we can render "follow." The -tu is the suffix for the third person singular imperative. So a better translation would be, "Let him/her follow the stream." I am told by my Sanskrit consultant, Herman Tull, that the anusvāra dot for the m is used but is not considered the best form at the end of sentence. Instead, the m should be written out as, .

Although nicely translated, this is not necessarily a sentiment from Indian philosophy or religion. It is, indeed, an old saying from the 60's and sounds, if anything, more like Taoism.

It is not clear if this tattoo commits the political crime of "cultural appropriation," since it only uses the Sanskrit language, and not really anything from Indian culture. Perry did need to get advice from someone who knew some kind of Sanskrit. Also, I wonder if this was only a temporary tattoo. I have not noticed other images of it on her.

See what kind of trouble people can get into with their tattoos elsewhere.

Return to Text

Greek, Sanskrit, and Closely Related Languages, Note 1

The word "Tocharian" is often said to be used "as the result of a mistaken identification" [Winfred P. Lehmann, Historical Linguistics, Third Edition, Routledge, 1992, 1997, p.81]. The word was taken from Greek historians who were talking about a people, the Tokharoi, of the Fergana Valley (in the headwaters of the Jaxartes [Syr Darya] River, between the Pamirs and the Tian Shan mountains) who converted to Buddhism and migrated to India. This does sound like the Kushans, but may have nothing to do either with them or the Lesser Yuèzhī of the Tarim Basin.

However, it turns out that among the Tocharian manuscripts is one written in Uighur, which is an Altaic language close to Turkish and represents the next wave of nomadic migrants into central Asia (c.600 AD). The Uighur text says that it was translated from a language called twghry -- the lack of vowels is an aritfact of Uighur using the alphabet from Syriac, which, like Arabic and Hebrew, typically doesn't write vowels. Twghry looks close enough to Tokharoi to now properly motivate the identification. So it must not have been mistaken after all.

I now see assertions that the Kushans spoke Bactrian, a S.E. Iranian language, as we see in the text above. I am unaware of the nature of the evidence for this. If it is well attested, it could mean a variety of things: (1) the Kushans were not Yuèzhī at all; (2) the Kushans were Yuèzhī who had begun to adopt, in whole or in part, the local language; or (3) the Yuèzhī had always been Iranians and the identification of even the Lesser Yuèzhī with the Tocharian speakers was a mistake. While it is quite possible that the Kushans were simply not Yuèzhī at all, I suspect that the second possibility is more likely. It also depends on the nature of the evidence. A Bactrian text among the Kushans does not even mean that the Kushans were speaking Bactrian among themselves. They may have, by necessity, begun using Bactrian to communicate with the locals, even as they certainly would have needed to use a local language once they moved on to dominion in India.

Return to Text