Author Topic: Accuracy of Historical Linguistics Reconstructions?  (Read 36 times)

Offline EliteoftheRad

  • New Linguist
  • *
  • Posts: 2
Accuracy of Historical Linguistics Reconstructions?
« on: February 07, 2018, 09:22:06 PM »
I have long been interested in historical linguistics, and unwraveling the mysteries of ancient languages, particularly their pronunciation, has been an endless source of curiosity. However, I am not a linguist, and so one question that has bugged me for awhile is: How certain are historical linguists of the accuracy of their reconstructions?

As we all know, attempts at a reconstruction of PIE began with comparisons between Latin, Greek, and Sanskrit. Nothing of PIE exists in the archaeological record, so reconstructions must rely solely on comparative evidence. But since, because of this, there is nothing to compare PIE reconstructions against, the accuracy of those reconstructions will forever be in question to some extent. In order to test the validity of comparative models of reconstruction, have there ever been any linguists who have tried to reconstruct, say, Latin solely using current methods with modern Romance languages? It seems to me that if Latin could be so recreated with a reasonable degree of accuracy, then the validity of these methods would be without question. However, if these methods were unable to reproduce Latin, then all reconstructions using current methods would be rendered invalid. After all, unlike ancient languages that mostly exist in fragments, modern Romance languages contain vast oceans of texts, and Latin is one of the best preserved ancient languages, so comparison of a reconstruction against the actual language would be simple.


Offline Daniel

  • Administrator
  • Experienced Linguist
  • *****
  • Posts: 1642
  • Country: us
    • English
Re: Accuracy of Historical Linguistics Reconstructions?
« Reply #1 on: February 07, 2018, 10:14:29 PM »
Yes, the methods can be used to reconstruct Latin (and other known languages). However, there are some practical difficulties, specifically that reconstruction results in Proto-Romance, which is similar to Vulgar Latin (spoken), not Classical Latin (written), so the connection is not direct. For example, *v is reconstructed as /v/, rather than as /w/ as it was pronounced in Classical Latin, because that innovation is shared throughout Romance. Doing a completely unbiased reconstruction of Latin is also difficult because most historical linguists know at least something about Latin, so it's hard to really use this as a blind verification of the methodology. But, yes, overall it works.

As for checking accuracy, one interesting case is how the laryngeal/glottalic theory of PIE was proposed without any direct evidence before the discovery of Hittite (as being related to PIE at least), which somewhat verified the existence of distinct H-like sounds. (The details are still somewhat complicated/controversial, but that is an example of new evidence being discovered in support of an existing hypothesis.)

Overall, the confidence in the results of comparative reconstruction are all relative. It is almost completely certain that the idea of PIE is correct-- that there is some common ancestor for all of those languages. But it is much less certain exactly where it was spoken, by who, or when. And we likewise have a pretty good idea of many of the phonemes but much less of an idea (if any) about the allophones and actual specific pronunciation of the phonemes. Really, all that reconstruction gives us is a mapping between modern languages that shows connections, and then we imagine that as a sort of artificial stage in the proto-language. Reconstructions for PIE, for example, range over maybe 2000 years, so it's really not the case that there even is a literal "Proto-Indo-European" to reconstruct. If we went back in time with a time machine and tried to speak with them, what exactly would it sound like? Well, that's tricky. When? But it is likely that we are right about, in general, most of the phoneme inventory and that would correspond to something at that point in time. So the reconstruction is something like an outline of the language, and indeed some aspects are probably quite reliable, while others are much less so.

So, in the end, it depends. The whole methodology should certainly not be abandoned. But many of the details are far from certain.
Welcome to Linguist Forum! If you have any questions, please ask.