Eurasiatic: A Wild Pursuit (1)

The aim of Pagel et al. (2013) is to lend support to the Eurasiatic superfamily proposal by showing that the suggested cognate classes size of four of more (i.e. Proto-Eurasiatic words reflected in at least four out of the seven families making up Eurasiatic) correlate with the frequency of use of the corresponding meanings to a statistically significant degree. In other words, the frequency of use turns out to be a good predictor of cognate class sizes because words used more frequently are less likely to undergo lexical replacement and in general retain the form-meaning association more faithfully (see Pagel et al. (2007), where the mean frequencies in question are established for the Indo-European family and found to be correlated with the linguistic half-lives of words). High-frequency words are therefore likelier to survive in a larger number of lineages descended from a common ancestor.

Since the size of the cognate classes discussed in the article can only equal 4, 5, 6 or 7, and the total number of meanings is only 23, it is clear that any inaccuracy in determining the number of cognates can have a significant effect on the statistics. It really matters if a given word is judged to have a cognate class of 7 or of 5, for example. How are those numbers determined? The authors accept the cognacy judgements from the Nostratic/Eurasiatic list of etymologies in the LWED database. The actual “cognates” are not listed, but the reader can track them down in the database using the information from table 1 in the article. Cognacy is accepted, according to the authors, if the putative cognates have precisely the same meaning in different families (or rather the same reconstructed meaning in their respective protolanguages). It’s a conservative requirement, but does it really reduce the danger of spurious cognacy? Let us have a look at some of the 23 cognate classes, beginning with the one with the most impressive score – 7 cognates in 7 families. The most successful  Eurasiatic replicator is the 2sg. pronoun ‘thou’ (see this link).

The Proto-Eurasiatic pronoun is reconstructed as *ṭ[u] with an emphatic (ejective?) coronal stop in the Moscow dialect of Eurasiatic. It is clearly designed to yield the familiar IE stem *tu-, and so it does. Thanks to the *u being optional, it can be easily matched with the Uralic 2sg. pronoun, whose forms begin with *t-, though it has to be noted that a match based on a single segment (and a dental stop at that) is not particularly impressive. But Uralic, like IE, also has a 1sg. pronoun with initial *m-, which reinforces the impression that the pronominal systems of the two families are somehow related. And here easy comparison ends.

In view of 1sg. *kǝm in Chukotko-Kamchatkan (and the parallel 1pl. *muri, 2pl. *turi) the final dental in 2sg. *kǝð may plausibly reflect an original second-person marker, which would make Chukotko-Kamchatkan another “M/T family”. The Eskimo(-Aleut) 2sg. forms, cited but not discussed in the database, are much harder to analyse and their final *-n (*-nt?) cannot be unambiguously connected with the M/T system.

Left: Mars in 1894. Right: Mars now.
[source]

Dravidian has no matching form (the 2sg. pronoun there begins with *n-, uncontroversially attested throughout the family, including the outlying Brahui language). The editors of the database, however, are evidently prepared to bend over backwards to find a trace of the “real” Proto-Dravidian pronoun, so they list the verb ending  *-ti as a match, and Pagel et al. accept that despite their sworn insistence on strict semantic identity! Perhaps they didn’t know (because the database didn’t mention it) that the conjugational ending in question wasn’t even reconstructible to Proto-Dravidian; but at any rate, a verb ending instead of a pronoun does not constitute a match. There goes one member of the cognate class.

Altaic (assuming for the sake of the argument that it counts as a bona fide family) has a 2sg. pronoun in *s-, which can’t be derived from an earlier stop, emphatic or otherwise, by the Moscow School Nostraticists’ own rules. Mongolic, however, has an aberrant form which could be a possible reflex of *ṭ- (no *u again, but remember that the vowel is conveniently optional), so Mongolic evidence overrules the testimony of the remaining four subfamilies of Altaic and another match is arbitrarily declared.

Kartvelian 2sg. *sen isn’t a promising cognate, so a plural form is cited instead (as if the Proto-Kartvelian 2pl. *(ś)tkwen really looked like a plausible relative of IE *tu-, but then at least it contains a *t beside four other consonants). This alleged match flies in the face of the evidence and should have been disqualified by Pagel et al.

To sum up, we have possible cognates in three families (Indo-European, Uralic, Chukotko-Kamchatkan), one extremely doubtful case (Eskimo-Aleut), and three rather clear negatives (Dravidian, Altaic, Kartvelian). An optimistic but conservative determination of the cognate class size should be 3, certainly not 7, for the supposedly most successful Eurasiatic replicator. And even so, the “matches” involve a single consonant with the most common place of articulation. Such evidence would be easy to dismiss if it were not for the accompanying 1sg. m-pronouns. Nevertheless, ‘thou’ should really have been removed from the list rather than promoted to its top.

[to be continued]

[► Back to the beginning of the Proto-World thread]

Mind the Asterisk!

In the previous post I pointed out that biologists do not use hypothetical taxa and their inferred features as data. But wait, linguistics is a different discipline. Perhaps in linguistics it’s perfectly legal to use asterisked reconstructions (like, say, Proto-Germanic *wulfaz ‘wolf’) as data on which higher-order reconstructions can be based? The structure of language families is hierarchical. We traditionally group uncontroversially related languages into “branches”, and for the members of each branch a respective ancestral protolanguage is reconstructed by applying the comparative method, right? Then we compare the “proto-branch” languages to reconstruct the most recent common ancestor of the whole family, don’t we?

No, we don’t. Proto-Indo-European was not reconstructed by comparing Proto-Indo-Iranian, Proto-Slavic, Proto-Italic, Proto-Celtic, Proto-Germanic, Proto-Anatolian, etc., with one another. It has always been reconstructed by comparing data extracted from a multitude of documented languages such as Vedic, Avestan, Old Church Slavonic, Serbo-Croatian, Latin, Old Irish, Middle Welsh, Albanian, Classical Greek, Biblical Gothic, Old Norse, Old High German, Hittite, Luwian, and so forth. Proto-branch languages are reconstructed first and foremost for the sake of quality control. The nodes of the family tree are where the most conservative features of the whole branch roughly coalesce, and where it is convenient to check the consistency of the reconstruction. Proto-Germanic is not reconstructed just by comparing English with German, Dutch, Icelandic, Gothic, etc. The reconstruction is informed by the rest of the family tree as well. There is considerable feedback from reconstructed PIE to reconstructed PGmc. To give a historically important example (one among many), for more than fifty years in the 19th century a large set of exceptions to Grimm’s Law remained unexplained, until it occurred to Karl Verner to look for evidence in outgroup languages such as Classical Greek and Vedic. The conditioning environment of the Proto-Germanic process now known as Verner’s Law was obliterated in Germanic itself, but preserved elsewhere.

Linguistic reconstruction is not conducted consistently in a bottom-up fashion, by piecing together smaller units before handling larger ones. A phylogeny is assembled as a whole, and if it becomes part of a grander phylogeny, incorporating outgroup evidence, it may have to be reorganised, and hypothetical  protolanguages may have to be redefined in order to optimise the enlarged model. This is what happened to Proto-Indo-European after the Tocharian and Anatolian languages had been added to the family tree. The common ancestor of the remaining IE languages is not the familiar Brugmannian reconstruction from the last decades of the 19th century. It has been deeply affected by new discoveries despite the fact that most Indo-Europeanists place both Tocharian and Anatolian outside the “crown group” containing all the modern IE languages.

WARNING: RECONSTRUCTION!

Thus, protolanguage reconstructions are not “data”. They are forever provisional and hypothetical. Using them as data is a category error. There are some thoroughly studied families like Indo-European whose protolanguages are reconstructible in considerable detail. In such cases even a historical linguist may be tempted to believe that the great success of the model makes it “real”, and so a reconstructed PIE word is as valid a piece of data as a documented word from a documented language. Such a belief is perhaps partly justified when a reconstructed form is used as a piece of shorthand intelligible to experts (who do not have to bother to list the obvious cognates) – and only if the reconstruction is straighforward and uncontroversial. Even in IE, however, reconstructions may contain questionable elements or require special explanations. Other “Eurasiatic” families are even worse off. Reconstructible Proto-Uralic lexemes would barely fill a Swadesh list. The very validity of Altaic as well-defined clade is disputed, and even assuming optimistically that Proto-Altaic is a valid concept, little of its structure can be reconstructed with any precision.

The use of *protoforms in datasets is not justifiable in any way if the reconstructions are highly conjectural, if they might be biassed (“improved” to make a point without sufficient evidence), or if they represent preliminary, speculative research whose quality remains controversial. The Languages of the World Etymological Database (LWED), produced by the Tower of Babel project, is precisely such a pioneering enterprise. Little wonder that the “Eurasiatic” reconstructions therein make liberal use of wildcard symbols, optional segments, variants and reconstructed features poorly supported by the comparative data – hallmarks of questionable comparison. They are also based on material drawn from a haphazard collection of sources, including some hopelessly outdated etymological dictionaries. And yet the compilers of the database claim that those reconstructions “may be used for regular comparative purposes – establishing phonetic correspondences and reconstruction – by future researchers”.

It’s a dangerous declaration: researchers, especially scientists with no linguistic training, may take it literally and believe that the etymologies in principle encapsulate reliable data, so that all the dirty work of actual linguistic analysis can be outsourced to the Tower of Babel team; the scientists can then use the condensed final product and need not worry about the rest. In the PNAS study of the Eurasiatic protolexicon the “final product” is then used as the basis for the determination of the size of cognate sets with a given meaning . So if one “proto-word” generates a cognate set spanning all the seven putative members of the Eurasiatic superfamily, and another one gives rise to a cognate set of just three, this difference can be expected to correlate with something real – even if the basis for the reconstruction is extremely tenuous, e.g. if on closer inspection it turns out that one of the alleged cognates is misreconstructed, another has been assigned the wrong meaning, still another is a loanword, and most present formal problems obvious to linguists but not necessarily to non-specialists. I will show in the next post that the “cognate set sizes” based on the LWED cannot be realistically determined with any reasonable accuracy, given the information provided in the database. The error may be so gross that the sizes determined in the study are simply fictitious.

[► Back to the beginning of the Proto-World thread]

Hypotheses Aren’t Facts

When biologists prepare the data to be analysed by a phylogenetic inference program, they may use morphological characteristics of organisms (or of their fossils), appropriately measured and encoded as a data matrix. The choice of the characters to be analysed is important and difficult, and often controversial. When the family tree of a DNA or RNA  region is to be assembled, sequences of nucleotides are treated as data. The phylogeny of a family of proteins is inferred from their amino acid sequences. In every case the data come from empirical observations and measurements. One can include living and fossil taxa in the same phylogenetic analysis, provided that the respective datasets are of the same kind. One can even integrate morphological and molecular data for living taxa with only morphological data for extinct taxa (for which molecular data are not available), and still get valid results.

The Homo/Mus MRCA, an early restoration
(Disney 1928)

One can’t, however, replace real datasets with reconstructed ones derived from hypothetical taxa. Why? Precisely because hypotheses are hypotheses, not observations. You cannot measure the bones or count the teeth of a hypothetical animal, even if you have a fairly accurate idea of what it must have looked like. A scientific hypothesis is always tentative and open to revision or rejection. A family tree, no matter how well-supported, is also a hypothesis. If your analysis suggests that Homo sapiens and the house mouse (Mus musculus) are more closely related to each other than either is to the cat or to the horse, you can predict (or rather retrodict) what the most recent common ancestor of mice and men should have looked like, and in what way it must have been different from the common ancestor of cats and horses. Evolutionary trees feature numerous hypothetical taxonomic units with hypothetical morphological or molecular traits. They are very useful in making testable predictions, but they do not count as proven facts. New empirical evidence may make us revise the whole family tree, and many of our hypothetical taxa will be gone instantly. Whatever we can infer about them, anyway, always comes from the interpretation of relevant data – observable, empirical data.

The same data may be compatible with an astronomical number of alternative models. Competing analyses yield different topologies of the family tree and different inferences about the hypothetical ancestral nodes. Those inferences implicitly encode the structure of a given tree, and so are not new independent “facts”. There would be no ancestral nodes if we had no phylogenetic hypothesis in the first place. We choose a family tree whose branches appear to be robustly supported and tentatively regard it as “the best”, but there is no universal metric to guarantee that our currently preferred hypothesis is optimal in some absolute sense. We can only say that we have arrived at a reasonably accurate reconstruction of past reality given our present knowledge (until falsified by new evidence).

What does it all have to do with linguistics (and with the PNAS article)? I will get to that in the next post, soon.

[► Back to the beginning of the Proto-World thread]

Eurasiatic? It Ain’t No Rocket Science

Just as I was approaching the subject of long-range comparison and language superfamilies at a leisurely pace, newspaper headlines and the blogosphere went abuzz with big news about the very same subject. How timely from my point of view!

'Ershver tooni monhrr!' In other words, hey, can you give me a hand! The ice age 'superlanguage' Europeans spoke 15,000 years ago - and we can still understand today

• Researchers uncovered language Ice Age people used to communicate - with many of the words still in use today

• Believe 'superlanguage' may have existed so different groups could communicate

• Experts say we could 'hold a simple conversation' using the language

Estonia, Moldova, and Macedonia: Here be dragons

It looks ridiculous and it is ridiculous. But then, all right: journalists will be journalists. In particular, the undereducated and uncritical science reporters of today inevitably misunderstand and misreport at least 60% of what they find in a scientific journal. But what did they report in this case? An article in a very serious and highly prestigious peer-reviewed scientific journal, the Proceedings of the National Academy of Sciences (PNAS). I live in a backward Central European country in which the Sciences and the Humanities are not clearly distinguished. Like our German-language neighbours with their Wissenschaft, we use the word nauka for both. It could be roughly translated as learning into English (or as scientia into Latin). I am a linguist, so, in English terms, I may be a scholar, I may have a PhD and a university post, but I am not a scientist. The articles I write are therefore not scientific and journals like PNAS would not be interested in publishing them. But it is a totally different matter if a team of scientists attack a linguistic problem and apply the Scientific Method to it. Then, it seems, the results become scientific. It may even be cutting-edge science, wow!-science. Who knows, it might even shift a Kuhnian paradigm!

The article in question is this:

Mark Pagel, Quentin D. Atkinson, Andreea S. Calude, and Andrew Meade. Ultraconserved words point to deep language ancestry across Eurasia
Published online before print, 6 May 2013
doi:10.1073/pnas.1218726110

It has already been criticised here (by Sally Thomason) and here (by Asya Pereltsvaig), but there are several aspects of the study that have not been sufficiently covered so far, so I am going to devote a few posts to them. It is nothing personal, and please do not think that I am defending my linguistic turf against a rival gang. I hope it is clear that I am no Luddite afraid of computers and algorithms. I am a big fan of the natural sciences, especially biology. Some of my visitors may recall that in one of my previous posts I referred to another paper by almost the same team (published in 2007, also in a leading scientific journal, Nature in that case). In my opinion, that article was a valid and stimulating contribution to linguistics. I am sorry I cannot say the same of the present PNAS article. It is flawed in a very fundamental way; it abuses the scientific method. Its only use is that we can learn from it how interdisciplinary research should not be conducted. I know it sounds harsh, but I really mean what I am saying. Stay tuned in, and you will soon learn why.

[► Back to the beginning of the Proto-World thread]

Related or Similar by Chance?

There are many analogies between the history of languages and biological evolution, but the differences between the two domains are also important and should be highlighted. One of them is the degree to which evolving units can be expected to remain recognisable over long periods of time (let’s assume that “long” means millions of years in biology and thousands of years in linguistics). The contrast is particularly striking if we consider biological evolution at the molecular level. How do we decide whether two fragments of DNA in the genomes of two different species have shared ancestry? The problem is by no means trivial, but it helps if the fragments in question are sufficiently long. If we align them and find a high percentage of identical base-pair sequences, it becomes likely that the partial identity is due to common origin rather than convergent evolution or pure chance.  Of course sufficient care must be taken to rule out other possible explanations. For example, a lengthy periodic sequence in which a simple motif is repeated numerous times (like CACACACACACA...) is something that may have been generated many times independently by some kind of replication slippage. Whether the similarity of two such sequences means anything depends on the context in which they occur (e.g. whether they occupy the same locus in their respective genomes).

Note that the “alphabet” of the genetic code is very simple and (almost) the same in all domains of life, and the same is true of the interpretation of the codons (the translation of nucleotide triplets into amino acid sequences). Guanine is guanine, and cytosine is cytosine, no matter if they form a GC pair in the DNA of a blue whale, an amoeba, or a forget-me-not. The unity of the genetic code leaves no interpretative leeway: two nucleotides are either identical or different, not “similar” or “almost the same”.

By contrast, morphemes and words are encoded as strings of phonological segments whose inventories differ widely from language to language. There may be universal constraints on what is permitted to function as a speech segment, but individual languages enjoy a lot of freedom in this repect. Even varieties of one and the same language (say, Scottish English, Received Pronunciation, and General American) may have quite different phonological systems. The same speech sound may play a different role in different inventories. For example, the contrast between plain [t] and aspirated [tʰ] is not employed to differentiate meaning in English (the occurrence of the two sounds is predictable from the phonetic context), but in many languages (Mandarin, Armenian, Ancient Greek, Hindi, etc.) /t/ and /tʰ/ are distinct phonemes – contrastive units perceived as different “letters” of the mental code we use to represent words in the lexicon. Because of the operation of sound change, the pronunciation of a word (and its phonological encoding) can morph into something pretty well unrecognisable in the course of several generations.

Comparing genes
[source]

A typical gene may contain about 30,000 base pairs. A typical word contains just a few phonemes. This difference in internal complexity is really significant. If there is a 50% match between the base-pair sequences of two aligned genes, the probability of this agreement being accidental is practically zero (even if they encode for rather different proteins and so have different “meanings”). If two words with “the same” meaning have “the same” form in two languages (remember that “sameness” is a tricky notion in linguistics, hence the cautionary quotation-marks), their identity means nothing by itself: it may well be accidental.

The most powerful device in the toolkit of historical linguists, the comparative method, is not interested in lookalikes but in words displaying recurring systematic correspondences – patterns resulting from the operation of historical sequences of regular sound changes. In order to undergo such changes (and to acquire their characteristic imprint, like a certificate of origin) a word must spend enough time in the company of other words in the same linguistic lineage. It may be difficult to believe that e.g. the English word daughter (RP /dɔːtə/, Gen.Am. /dɔtɚ ~ dɑtɚ/) is related to Polish córa /ʦura/ (with the same meaning), but the relationship is more or less obvious to anyone familiar with the regular sound changes in Germanic and Slavic, and with the correpondences they have left behind. Remember: not because the words are similar (they aren’t) but because the relationship between the most conservative lexical strata of Polish and English conforms to a well-defined pattern of formal correspondences, described in detail in the linguistic literature.

Greek theós and Latin deus not only have the same meaning (‘god’) but also look very similar: /tʰ/ and /d/ are both dental stops, /o/ and /u/ are both back rounded vowels (and we know that Greek -os  often corresponds to Latin -us); the remaining segments are simply identical, and there is a similar hiatus (absence of a consonant) between the vowels in each case. There is, on the other hand, no plausible common source for the initial sounds, despite their similarity. No PIE word-initial consonant (or consonant cluster) can yield /d/ in Latin and / tʰ/ in Ancient Greek. The pair violates all known patterns of correspondence between the two languages. Moreover, a careful analysis shows that there are practically certain relatives of deus in Greek (e.g. the adjective dĩos ‘divine’ and the name of Zeús, the chief Olympian god), and plausible relatives of theós in Latin (e.g. fās ‘religious law’, fēstus ‘festive’, and fānum ‘temple’). They are hardly similar to theós, but they fit nicely into the known pattern of correspondences. One could entertain other possibilities, e.g. that theós might be a loanword from an unknown IE language in which the PIE *d > *tʰ; such an assumption would account for the deviation from the expected pattern. But – apart from the fact that there would be other features of theós left unexplained – ad hoc recourse to otherwise unheard-of languages with arbitrary charcteristics falls foul of Ockham’s Razor. The hypothesis that deus and theós are unrelated, and that theós is instead related to fās etc., is more parsimonious, and on the whole more compelling. The respective prototypes of deus and theós are therefore reconstructed as *deiwós versus *dʰh₁sós, derived from different PIE roots. It follows that their similarity is deceptive and there is no genetic link between them.

The likelihood that two words from different languages will display both a “matching” meaning and a “matching” form quite by chance is much higher than most people would imagine, and increases dramatically every time we relax the criteria of what constitutes a match. Without any formal controls, such as those that allow us to recognise spurious cognates in lexicons with reconstructable histories, matches are a dime a dozen and have no intrinsic value as evidence. If two languages are distantly related, real cognates are often as dissimilar as daughter and córa, and we can’t identify them just by eyeballing the material.

Pieces should form a pattern

To be fair, the oldest words in the IE languages have often retained enough similarity to be “visibly related” even to a layperson. After all, it was the bare resemblance of some Sanskrit words to those in Greek, Latin, etc. (famously observed by Sir William Jones in 1786) that led to the discovery of the IE family. But it was the careful application of the comparative method and the reconstruction of the common “protolanguage”, as well as the unravelling of the changes transforming it into the historically known languages, that allowed linguists to progress from impressionistic speculation to something resembling a scientific model. Many of the correspondences that seemed evident to linguists in the early 19th century  have turned out to be misleading, and lots of non-evident ones have been discovered. Despite being regular, they couldn’t be spotted by a naive observer.

[To be continued in the next post.]
[► Back to the beginning of the Proto-World thread]

Morphemes Are Forever

Where do words come from? We may form them by combining preexisting words or roots with derivational morphemes, but where do those preexisting components come from?

The fundamental grammatical units (morphemes and words) are sometimes created from scratch. Imitative and sound-symbolic words (meow, cuckoo, swoosh, bang), are obvious examples. Of course as soon as they become fully lexicalised, they may develop in parallel with other words and undergo the same sound changes, which may result in the loss of their iconic character. For example, Latin had the imitative verb pīpiō ‘pip, cheep (like a chick)’. The fact that we also have Ancient Greek pippízō or, for that matter, Finnish piipittää, all with the same meaning, tells us more about young birds (they made noises like [piː piː] in the past just as they do today) than about the languages that have “borrowed” the bird call in question. Such imitative word-coining must have happened independently many times. But Latin also had several words derived from the verb, among them post-Classical pīpiō (acc. pīpiōnem) ‘nestling, young bird’. That derived noun gradually developed a more specialised meaning, ‘squab, young pigeon (or its meat)’, and its form evolved according to regular sound changes. In the passage from colloquial Latin to French, the medial -pi- developed into *-βj- > *-βʤ- > -ʤ-, eventually yielding Old French pijon (~ pigon, pichon, pigeon) /piʤõn/, which was borrowed into fourteenth-century English (and spelt variously peion, pyion, pichon, pygeon, etc.). Modern French /piʒɔ̃/ and Modern English /'pɪʤən/ are not onomatopoeic any more. They have become “normal” words, for which the sound-meaning pairing looks entirely arbitrary. After all, they normally refer to adult pigeons today, and everybody knows that adult pigeons go “coo, coo” rather than “pee, pee”.

It is imaginable that many lexical roots began their existence in a similar manner, as iconic sound combinations mangled beyond recognition by the operation of historical sound change; note, however, that “imaginable” does not mean “demonstrable”. Most of the morphemes whose genealogy we can follow back to a remote past do not seem to have been onomatopoeic during their recoverable history. What is their ultimate source? We simply don’t know. If we say that an English word or lexical root is “of Proto-Indo-European origin”, we mean that at the present state of our knowledge we can trace it back to PIE and no further. If we say that it is “of Latin origin”, we point to Latin as the donor of a loanword and usually stop there; but of course the Latin source itself may have a deeper history in Italic, or it may be a Classical Latin derivative of something older, or it may be a loanword from a still different source (Greek, Etruscan, Gaulish, etc.), in which case the whole cycle repeats itself. Etymological dictionaries declare some words to be “of unknown origin” or “without etymology”, which means that at some point we run out of information as to their deeper origin. But to be honest, we always have to stop somewhere, even if “somewhere” refers to the oldest available reconstructed stage. The genealogies of words (or their parts) lead us deeper and deeper into the past, often forcing us to follow a complex trajectory of borrowing from language to language, or even between language families. In the end we either reach a point beyond which reconstruction is impossible, or lose track of the word in the uncharted wilderness of poorly documented languages. With relatively few exceptions (words known to be coined ab novo), the lexical items we study, or at least their morphological components, are indefinitely old.

Is there a way to refine our methods so that tracing words back to their source would not depend on our ability to reconstruct “protolanguages”? Words and morphemes are in principle immortal. Thanks to horizontal transfer, they can survive the death of languages they came from. For example, quite a few English words (ebony and ivory, among them) are certainly or possibly of Ancient Egyptian origin in the sense that we can follow their trail back to Ancient Egyptian via intermediaries such as French (or Spanish, or Arabic), Classical Latin, Ancient Greek, or Old Persian. Ancient Egyptian is as far as we go simply for lack of further clues. Perhaps there is a way to recover at least a handful of “global etymologies”, i.e. words whose reflexes occur in the language families of all the inhabited continents – not necessarily by virtue of being inherited from a common protolanguage but because they have managed to colonise the world thanks to a combination of inheritance and prehistoric borrowing?

I don’t think we have been introduced.
[Albrecht Dürer. Source: Wikimedia Commons]

To use a biological analogy, we know that the mitochondrial DNA of our species has a single source – an anonymous Paleolithic woman we dub “Mitochondrial Eve”. There was also a common male source of our Y chromosomes, a certain “Y-Chromosomal Adam”, who lived at a different time and place from Mitochondrial Eve. But that is not the end of the story: in the human genome there are a few hundred thousand regions for which family trees could be constructed, and each of them may be rooted in a different prehistoric individual, some “Adam”, “Eve”, “Jack”, “Jill” or “Algernon” who never met any of the others. Those trees, viewed separately, do not represent the ancestry of the whole species, or of identifiable subpopulations – just of selected genomic loci. Perhaps we can get round the “unreconstructible Proto-World” problem by concentrating on globally shared lexical units and leaving aside the question which particular languages they have passed through?

The discussion will be continued in the following posts.

[► Back to the beginning of the Proto-World thread]

The Uncertainty Principle in Linguistic Reconstruction

What is a “reconstructed protolanguage” like Proto-Indo-European? It’s customary to define it as the most recent common ancestor of a family of languages. If the structure of relationships within the family is represented as a clearcut phylogenetic tree, it is even possible to offer a formal definition based on a pair of languages, A and B, each belonging to a different primary branch of the family (produced by the oldest split in its history). Thus, Indo-Europeanists generally agree that the Anatolian group (Hittite, Luwian and their close relatives) had split from the “core” part the family before Core IE underwent further fragmentation. By putting it in this way we ascribe a “basal” status to Anatolian and tend to see the other branch as “PIE proper”. That’s because Core IE contains all the modern languages and all the familiar (and excellently documented) “classical” ones like Latin, Ancient Greek, and Vedic. We should realise, however, that the privileged position of Core IE with respect to Anatolian is merely an artifact of the history of IE studies and of the (accidentally) unequal written attestation of the two primary branches. It is more reasonable to say that the common ancestor bifurcated into a pair of daughter languages, Proto-Anatolian and Proto-Core (rather than to insist that either of them “split off first”). The fact that Proto-Anatolian has left no contemporary descendants is an accidental consequence of the vagaries of history, not of its “basal” or “less advanced” status.

Getting blurry
Hat tip: Jo Verrent

Hittite is a member of Anatolian; Greek is a member of Core IE. Therefore, we could define PIE as the most recent common ancestor or Hittite and Greek (or Luwian and Albanian, or Lycian and Welsh, if you prefer).  Any such definition is fine as far as we accept the family-tree model as a valid representation of the history of languages. It is worth reflecting, however, that we reconstruct protolanguages by comparing a large number of sets of related replicators (words, morphemes and their phonological building-blocks)  and tracing their history back, as if reversing the changes they have undergone. In a “perfect” family tree the genealogies of replicators simply follow the genealogies of languages. They all converge on the same point – the earliest bifurcation in the family tree. We have seen, however, that every replicator has its own family tree. For the oldest lexical layer of the Indo-European languages most of those family trees roughly coincide, producing the outline of a somewhat fuzzy but reasonably distinct language phylogeny. But do they all go back to the same time and place? Quite certainly not. We have no guarantee that all the features we call “Proto-Indo-European” co-occurred in one and the same speech community. Rather than that, their individual histories coalesce within a certain temporal range. The coalescent points may be considerably older than PIE as defined above. Such “deep coalescence” is especially likely when we combine the comparative method with internal reconstruction.

Deep coalescence

Let’s consider an example. We have a theoretical model of phonological alternations in PIE nominal paradigms. We know (or, to be precise, have reasons to believe) that some nouns originally had an “acrostatic” pattern of accentuation: the root syllable was accented throughout the declensional paradigm, and the quality of the root vowel alternated between *o and *e, whereas other nouns, accentually “mobile” ones, showed a parallel pattern in which an accented root vowel (normally *e) corresponded to the acrostatic *o, and the phonetic reduction of the vowel, caused by the shift of accent to a suffixal syllable, corresponded to the acrostatic *e. Traces of the acrostatic pattern are quite rare in the documented languages. We reconstruct the pattern on the basis of its scattered remains rather than fully preserved paradigms. Still, we are pretty confident that an alternation such as shown below occurred at some stage in the stem *dorw- ‘wood, tree’ (an acrostatic neuter):

• nom./acc.sg. *dóru, gen.sg. *déru-s

However, no IE language preserves anything remotely like this. On strictly comparative grounds, without trying to make sense of the paradigm and its relation to other similarly behaving nouns, one could reconstruct something like *dóru/*dorw-ós (Indo-Iranian disagrees, suggesting *dóru/*dréu-s instead), and leave it at that. However, we have reasons to believe that such forms are analogical, reflecting various attempts to level out vowel alternations or replace the unproductive acrostatic pattern with a more common mobile one. If the reconstruction *dóru/*déru-s is correct, it belongs to a deep chronological stratum in the prehistory of PIE itself. It was in all likelihood replaced by “regularised” variants like *dóru/*dérw-os (with a vowel inserted in the genitive suffix), *dóru/*dorw-ós (with the root alternation eliminated and the genitive ending accented, as in more productive mobile patterns), and possibly others, before the disintegration of PIE unity. Quite possibly by that time the “original” paradigm had already been abandoned and forgotten. The large amount of hesitation and paradigmatic inconsistency found even in the most conservative languages on either side of the Anatolian/Core divide suggests that much of that polymorphy was inherited from the common ancestor. Fully coalescent forms must therefore be older and can’t be dated with much accuracy. Reconstructed PIE is only roughly bounded in time (and even less so geographically, since even the approximate location of the IE homeland remains a moot question). The more we rely on abstract analyses to recover “the oldest” forms of alternating paradigms and to understand the origin of the alternations, the less precisely their chronology can be determined.

Note that we are talking of a well-studied language family with 400+ extant members and a written record beginning in the second millennium BC. The PIE reconstruction is a monumental intellectual achievement, and yet it isn’t “a language” that could be ascribed to any single speech community at any time. It’s a large set of coalescent reconstructions distributed in time and possibly in space as well. Other protolanguages, even relatively uncontroversial ones, are usually still more nebulous. If we ever manage to prove that the IE languages are related to some other established family, the reconstructed features of the common ancestor will naturally be even harder to constrain, and the protolanguage itself more elusive and fragmentary. It is hard to predict how far back in time our best reconstructive methods can take us before the notion od “protolanguage” becomes too vague to be meaningful. We can only resolve this question empirically, by putting our methods to extreme tests. If we consistently fail, it may mean that we have already reached the limit. Fortunately, there is no shortage of enthusiasts undaunted by the difficulties of long-range comparative research. Their efforts are necessary and praiseworthy, but the results so far have been rather disappointing. Only time can tell if further progress can be achieved. But even if some “superfamily” groupings eventually win general acceptance, we are still light-years away from reconstructing anything resembling Proto-World. My personal view, which I have tried to justify in this series of posts, is that linguistic lineages are too ill-defined to remain identifiable at great depths of time. Language phylogenies are not real objects but products of our analysis, with all its limitations and simplifying assumptions. They are defined from a certain point of view, relative to an arbitrarily chosen historical frame of reference. If you try to extend them far beyond that frame, their boundaries, fuzzy to begin with, just blur away. There is therefore no way to define Proto-World properly as “a language”, let alone reconstruct its features.

But perhaps we can talk meaningfully of global etymologies without insisting on the reconstruction of Proto-World? This will be the topic of the next post.

[► Back to the beginning of the Proto-World thread]