#*kʷ *gʷ *kʷʰ (< pie *gʷʰ)

[protoindoeuropean]

There are two examples (from my field of Indo-European comparative linguistics) that immediately spring to mind when it comes to the reliability of the comparative method and the method of reconstruction, which show that the methods of historical linguistics can successfully predict what will be found, even if predicting history might at first seem like an oxymoron:

In 1878, Ferdinand de Saussure (yes, that guy) proposed – based strictly on structural grounds [which seems fitting for the founding figure of linguistic structuralism] – that the Indo-European proto-language featured sounds that had an effect on the other sounds around them (notably vowels, but also consonants), but which then disappeared and were not visible as such in any of the actually attested Indo-European languages, no matter how ancient. His interpretation of sonant coefficients (coefficients sonantiques) was developed into what is today known as the laryngeal theory, which is sometimes considered the crowning achievement of Indo-European comparative linguistics. And the reason for that is that starting in 1905, excavations at an archaeological site in Turkey revealed thousands of cuneiform clay tablets, written in a language that was in 1917 accepted as belonging to the Indo-European language family – Hittite. Ten years later, one of the de Saussure's proposed sonant coefficients was identified as corresponding to a sound explicitly attested in Hittite: thus, for example, Proto-Slavic *pa̋sti 'to pasture', Latin pāscō 'I pasture', which with their long root vowels indicate a past presence of a laryngeal (what would be previously reconstructed as Proto-Indo-European *pās- and *pāska-, de Saussure would write *pa₁ᴀs- and *pa₁ᴀska₁- and today we would write *peh₂-s- and *peh₂-sk̑e-, the sequence *a₁ᴀ / *eh₂ producing long PSl. *a̋ and Lat. ā) have a cognate in Hitt. paḫšari '(s)he guards': de Saussure's *ᴀ / modern *h₂ = Hitt. ḫ. What de Saussure only hypothesized was identified in reality – 49 years later.

The other example is the decipherment of Linear B and with it Mycenaean Greek: this earliest stage of Greek displays certain characteristics that the other, younger Greek dialects do not. For example, at the time of its decipherment it had long been recognized that the Indo-European proto-language included labiovelars (*kʷ, *gʷ, *gʷʰ), which however do not appear as such in the attested Greek dialects and furthermore have divergent reflexes (e.g. PIE *kʷetu̯ores > Boeotian Gk. πέτταρες, Attic Gk. τέτταρες, with p- vs. t-, cf. Lat. quattuor 'four'). It was thus suggested that the intermediate stage, Proto-Greek, should still have the (mostly) unmodified labiovelars. And lo and behold, the stage of Greek closest to Proto-Greek still has the labiovelars: the quintessentially Greek βασιλεύς 'king' appears in Mycenaean as qa-si-re-u 'official' (< PGk. *gʷasileus). What was previously only assumed for an earlier stage of Greek was thus then also confirmed.

on reconstruction and historical linguistics

to follow up on today's reblog, i want to comment briefly on the apparent misapprehension that linguistic reconstruction is just guesswork with a fancy name, because that's not accurate!

reconstruction is based on specific, well-attested constraints of linguistic development. we know from centuries of investigation that languages tend to change in predictable ways. we also have a decent understanding of the complexities introduced by phenomena like language contact, which can result in borrowing on multiple structural levels. our methods are well established and borne out by evidence.

comparative reconstruction involves applying these known constraints ("rules") in reverse on a collected body of words in related descendant languages. when possible, we also incorporate historical written evidence, which often provides midpoint references for changes in progress. it is always recognized by historical linguists that reconstruction can be imperfect; we cannot know what information has been lost.

the results of reconstruction can be mixed, but i'll let campbell (2013:144) explain:

How Realistic are Reconstructed Proto-languages? The success of any given reconstruction depends on the material at hand to work with and the ability of the comparative linguist to figure out what happened in the history of the languages being compared. In cases where the daughter languages preserve clear evidence of what the parent language had, a reconstruction can be very successful, matching closely the actual spoken ancestral language from which the compared daughters descend. However, there are many cases in which all the daughters lose or merge formerly contrasting sounds or eliminate earlier alternations through analogy, or lose morphological categories due to changes of various sorts. We cannot recover things about the proto-language via the comparative method if the daughters simply do not preserve evidence of them. In cases where the evidence is severely limited or unclear, we often make mistakes. We make the best inferences we can based on the evidence available and on everything we know about the nature of human languages and linguistic change. We do the best we can with what we have to work with. Often the results are very good; sometimes they are less complete. In general, the longer in the past the proto-language split up, the more linguistic changes will have accumulated and the more difficult it becomes to reconstruct with full success. (emphasis mine)

or, to quote labov's (1982:20) pithier if less optimistic approach:

Historical linguistics may be characterized as the art of making the best use of bad data, in the sense that the fragments of the literary record that remain are the results of historical accidents beyond the control of the investigator.

in sum, historical linguists are very realistic about what we can achieve, but the confidence we do have is genuinely well earned, because linguistics is a scientific field and we treat our investigations with rigor.

---

Campbell, Lyle. 2013. Historical Linguistics: An Introduction. Cambridge, MA: MIT Press.

Labov, William. 1982. "Building on Empirical Foundations." In Perspectives on Historical Linguistics. Winifred P. Lehmann and Yakov Malkiel, eds. Pp. 17-92. Amsterdam: John Benjamins.