I have often read that the vocabulary of the English language consists of words of roughly 60% Latin origin and of roughly 40% words of Germanic origin, give or take. This to me seems accurate as a relatively crude approximation, as almost every English word can either be found in a modern Latin-based language like French or Spanish or in a Germanic-based language like German.

My question is: If I wanted to get crude, statistically provable results for the lexical origin in other languages, what would be the best way to go about it? It should be a process that can be automated.

Let's say for example I wanted to define the lexical origins of a language like modern Russian. What would be the best way to go about it?

I have found that what makes this hard to answer in a generalized fashion is, among other things, the fact that the result can highly depend on what kind of source material it is based upon. For example, if you were to analyze a text about medicine or technology in Russian, you may get the impression that Russian is Latin or English-based to a much higher degree than if you were to analyze texts about topics that use less homogenized vocabulary.

On the other hand, analyzing the entirety of a language's vocabulary might also result in skewed results, as that would include obsolete words, as well as highly technical and specific ones, again resulting in an output that may not accurately reflect the average use of the language. Perhaps it would then make sense to take into account all words of a language but apply a score to words that is based on their overall frequency?

What are good ways to go about measuring this? For instance, with the English example, how was this done? Because it does seem to relatively accurately reflect the reality for standard use cases.

  • You are going to have to decide (a) whether you are looking at word lists in a dictionary or words as used in a corpus (the latter weighting towards more common words) and (b) what to do about words which moved between Germanic and Romance languages or the other way round before coming into English. For the first, consider Wikipedia saying "around 70 percent of words in any text are Anglo-Saxon" at the same time as saying one study gave about 26% from Germanic sources, 29% Latin and 29% French, 16% others
    – Henry
    Dec 19, 2021 at 12:59
  • The questions would then be: 1. Where could one find an appropriate / unbiased corpus that reflects modern, "standard" usage of a given language? Perhaps one based on popular newspapers in that language? 2. Are there tools that automate the process of recognizing the origins of the words used? Dec 19, 2021 at 16:53
  • 1
    Don't forget, written sources are badly biased. Almost nothing gets written and published, compared to what gets spoken. And in searching for "sources", how far back to you want to go? Russian, for example, has two etymological sources for Slavic-origin words, resulting in doublets like gorod and grad, Rossiya and Rus, etc.
    – jlawler
    Dec 19, 2021 at 17:52
  • Would you say that this holds true even today, with so much being written online? If there is such a thing, I could easily include a social media / forum / comment section corpus as well, to reflect that part of language. I feel like it wouldn't be far off from spoken vocabulary. As to your other question, for what I'm trying to do, it would completely suffice to just have crude categories like "Slavic", "Latin", "Germanic". Dec 19, 2021 at 21:22


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Browse other questions tagged or ask your own question.