HdB: Etymology

Sanskrit and ultraconserved Eurasian words

Prof.dr. J.M.H. du Buf - 1/6/2026

Email: dubuf AT ualg DOT pt

This is a bit scientific, sorry!

Introduction

Linguists are trying to find the origin(s) of human language(s), plural or singular, that's the question. One way is to try to establish a genealogical tree. Sir William Jones (1746-1794) was a British Orientalist, he studied at Harrow and Oxford, and learned some 28 languages. While in Calcutta, he learned Sanskrit, and then postulated the common ancestry of Sanskrit, Latin and Greek. In 1786! His work was the start of comparative linguistics in early 19th century. See Garland Cannon's biography.

It is now generally acccepted that many languages have a common origin, see Colin Renfrew's paper on the spread of farming after the decline of the hunters-gatherers, and the proliferation of settlements and villages. He argued that all began with the spread of farming in Anatolia, the Hittites in Turkey, see this interview about his life and Archaeology and Language. See also the TED talks of Michael Corballis (The Origins and Evolution of Language) and Dan Everett (How Language Began). Mark Pagel, see also below, talked in his TED talk How language transformed humanity and Gifford Lecture The Evolution of Language: From Speech to Culture about the development of language, humanity and culture. He showed in his Stony Brook Provost Lecture (The evolution of human languages) the current PIE genealogical tree: Proto-Indo-European was the precurser to Hittite (extinct Anatolian) and then to Sanskrit. Another branch went to Greek. Yet another branch went to Latin (and finally to French plus Italian), and in parallel to English and German.

All about 6000 languages can be grouped into some 12 macrofamilies. Large databases are required for linguistic research, such as the LWED, Languages of the World Etymological Database, which is part of the Tower of Babel. LWED is continued below. First I cite from the Evolution of Human Language (EHL) project, from 2001 to 2014 I think, coordinated by the Santa Fe Institute:

There are currently about 6000 languages on our planet, some spoken by millions, some by only a few dozen people. The primary goal of the international program known as EHL (Evolution of Human Language) is to work out a detailed historical classification of these languages, organizing them into a genealogical tree similar to the accepted classification of biological species. Since all representatives of the species Homo sapiens presumably share a common origin, it would be natural to suppose - although extremely hard to prove - that all or most known human languages also go back to some common source. The only way to proceed here is “bottoms up”: classifying attested languages and dialects into groups, groups into families, families into “macro-” or “superfamilies” and so on, as far as one can penetrate using comparative-historical and cladistic methodology. Most existing classifications, however, do not look behind some 300-400 language families that are relatively easy to discern. This restriction has natural reasons: languages must have been spoken and constantly evolving for at least 40,000 years (and quite probably more), while any two languages separated from a common source usually lose almost all superficially common features after some 6,000-7,000 years.

Nevertheless, despite widespread skepticism and reluctance to tackle the problem, there are a number of scholars who believe that these obstacles are not insurmountable. Research has been going on over the past several decades that appears to indicate that larger genetic groupings are not only possible, but indeed quite plausible. It can be shown that most of the world's language families can be classified into roughly a dozen large groupings, or macrofamilies.

These groups are shown on the EHL map at the Santa Fe Institute. If you are interested in chronological relations between languages, you should also read about Morris Swadesh and glottochronology.

Ultraconserved words and deep language ancestry

Mark Pagel et al. (2013) published the paper “Ultraconserved words point to deep language ancestry across Eurasia” in the Proc Natl Acad Sci USA, 110(21), pp 8471-8476; see the copy at the PubMed Central. They analysed 200 words from the Swadesh lists of 7 of the 12 groups in the LEWD proto-words. These 7 groups are: Altaic, Chukchi-Kamchatkan, Dravidian, Eskimo, Indo-European, Kartvelian and Uralic. They looked for corresponding cognates. In historical linguistics, cognates or lexical cognates are sets of words that have been inherited in direct descent from an etymological ancestor in a common parent language. First they counted the number with corresponding cognates in the 7 groups. Words with less than three counted groups were excluded from further analysis: this were 12, so there were still 200 - 12 = 188 words.

The cognate class size is defined as the number of groups (language families) with corresponding cognates. The histogram of the number of lexical items against the cognate class size (CCS) shows (all approx):

45 with CCS=1
75 with CCS=2 (the peak)
46 with CCS=3
15 with CCS=4
5 with CCS=5

As expected, the higher CCS's 6 and 7 had fewer lexical items: exactly one!. CCS=1 means that no cognate in any of the other families was found, whereas CCS=7 implies a cognate in all 7 families. Below we will see that CCS=7 and 6 had only one cognate, i.e., only in their own family or in their own plus one other.

Pagel et al's list of 23 words

Pagel et al. also analyzed other word properties. Wiki: Linguistic half-life refers to the rate at which a language loses its speakers or proficiency over time, often due to factors like cultural assimilation or the dominance of a more prestigious language. This concept highlights how languages can gradually become less used until they eventually die out. This is correlated with the rate of lexical replacement and the frequency of use. Unfortunately, the list of 1000 most frequent Sanskrit words that I used myself does not mention frequencies or other data, so I jump this information and discuss Pagel et al's list of 23 words.

They listed all words with cognate class sizes (CCS) of 4, 5, 6 and 7. Remember that CCS is the number (of seven) of Eurasiatic language families that are reconstructed as cognate for the word used to convey the meaning as shown. Here lurks some assumption that I cannot yet fully understand. Anyhow, here is the top of the list with the CCSs and half-lifes in 1000s of years:

thou - 7 - 11
I - 6 - 77
not - 5 - 8.5
that - 5 - 3.7
we - 5 - 19
to give - 5 - 9
who - 5 - 77
this - 4 - 3.2
what - 4 - 10

I included the word what because of the half-time of 10,000 years, but the words I and who break all records: 77,000 years. All other 23 - 9 = 14 words with CCS=4 are:
man/male, ye, old, mother, to hear, hand, fire, to pull, black, to flow, bark, ashes, to spit, and worm. I suppose that thou or Thou means you in a non-religious context, although in the case of Ye I am not sure.

I was pleasantly surprised by this list, because I have been looking for similar lists for a long time. The reason is the correspondence with my own Sanskrit word list, see HdB: on Sanskrit. I should emphasise that my own list was obtained by a completely different method.

I myself analysed the list of 1000 most commonly used words on Travel with Languages. This list is based on texts from the Mahabharata (a long epic narrative from ancient India), several of the Upanishads (ancient Indian religious and philosophical texts), as well as several yogic texts (Patanjali and Pradipika). Apart from the list, the only information is that the most frequent words are listed first, the least last.

I took the 1000-word list, and selected only all words with 2, 3 and 4 characters. The reason was that I realised that these had very elementary meanings for early cave dwellers, the hunters-gatherers who cooked on fire, some 400,000 years ago. And I later realised that archeolinguists and glottochronologists etc., however intelligent and knowledgeable they are, will never be able to jump from 40,000 years ago to 400,000 years ago. Although: I think that, currently, linguists approached some 4000 years BCE. At the other end, I saw a figure about the development of world population and settlements (Renfrew?). Some 400 kya ago, hominins started to experiment with sounds and speech, and soon they realised that they needed a basic vocabulary based on their sound inventory. I guess that must have been a relatively rapid process. From 400 to 40 kya, the devlopment was rather slow. But after 40 kya, the development was exponential, not only of the world population and settlements, but also in the number of languages. Perhaps, from 40 kya to 4 kya, there are still vestiges of very early language(s). So that's why I speculated about Sanskrit as one (of how many?) remnants of Paleolithic language. Whether upper, mid or lower, paleo or neo, I had to give it a name...

Anyhow, let me now list the words of my own 160-word list that are represented in Pagel et al's list, from top (CCS=7) to bottom (CCS=4):
thou, I, not, that, we, to give, who, this, what, man/male, ..., mother, ..., fire, ..., worm.
The ... indicate jumps over some words. These are 13 of Pagel et al's list, half! The missing words are not in the 1000-word Sanskrit travel list, so naturally I missed these, but they are all in the Swadesh list: old, hand, black, bark and ashes, plus the verbs to hear, pull, spit and flow. Checksum: 13 plus 9 is 22 (not 23, the word Ye I simply ignore). In an additional webpage I further discuss the word lists and vocabulary, extend the list with 5-character words, and link to the Indo-European vocabulary table; see HdB on early vocabularies.

Discussion

This is not a scientific paper, because of my own unconventional approach, but Pagel et al's work is! Nevertheless, here follow a few general remarks.

First, some form of disclaimer: Sanskrit is included in the Indo-European family, and sometimes considered within the broader context of the Eurasiatic language hypothesis (a suggested connection between various language families, including Indo-European and thus also Sanskrit). I don't know how deeply Sanskrit is represented in the LEWD proto-words, only in Indo-European or perhaps also in Dravidean and Kartvelian?

Pagel et al. only published cognates with their class sizes, so we do not know in which languages the cognates have been found, in which combinations of languages, whether these are completely random or yet reveal some structure. Pagel et al. did not show or discuss CCS smaller than four, so we must assume that there is absolutely no important or consistent information?

Pagel et al's scientific work is very challenging, but first results have been obtained. However, Pagel et al's word list is rather limited and very general, one can expect these words to exist in almost any language. Hence, an approximate syntax must be used as a selection criterium. At the bottom of my Sanskrit page, HdB on Sanskrit, in the section about Proto-Indo-European or PIE, I put two small lists with common words in English, PIE, Ancient Greek and Iranian, and also emphasise similarities in Gothic, Slavic, Baltic, Celtic and Armenian. Except for PIE itself, all are descendants of PIE, so the problem remains to find precursors of PIE: using Pagel et al's method?

Finally, my own work is not scientific but rather intuitive. Based on a few simple observations and a few simple selections, I arrived at a list of 160 basic words that represent an almost complete, but not yet final, vocabulary which was suitable for early cave dwellers. The earliest cave dwellers? Or maybe some later generations with a more sophisticated language? In an additional webpage I further discuss the word lists and vocabulary, extend the list with 5-character words, and link to the Indo-European vocabulary table; see HdB on early vocabularies.

The question is how to proceed from here...