Apologies in advance for any ignorance, I'm a non-linguist hoping to better understand the methods in the field (if any) to answer a question I have. In particular, I want to know when a word first appeared in a language tree.

For instance, if we take the word "iron," I see that the earliest version in the Wiktionary (https://en.wiktionary.org/wiki/iron#Etymology) is from Proto-Celtic, derived from the PIE for "blood." So, perhaps I could naively date the word to around 1000 BC, the rough age of Proto-Celtic (https://en.wikipedia.org/wiki/Proto-Celtic_language).

My question is what the most proper and formal way to do this is for a large number of words and languages. Should I:

  1. Be looking at etymological dictionaries / estimated origin dates for languages?
  2. Looking at some computational model conducted at the word level?
  3. Give up because we don't know enough to know these sorts of things? Of course, I don't need to know the word appeared on January 12th, 1427 BC. The question is, is there any method that improves beyond naive guesses?

Appreciate very much the help in this area--thanks a lot!

2 Answers 2


Before trying to figure out how to date a word, you need to first decide how to detect a specific word. Taking the example of "iron", first we have to decide whether to believe the Wiktionary story, versus Calvert Watkins' story (is-(e)ro "powerful, holy"). Either way, it's pretty clear that the word hasn't always meant "iron", so then you have to decide whether you set the birth of the word at the time when the sound-sequence took on the meaning "iron", or do you focus on the sound sequence and discard subtleties of meaning change?

If you want a stricter semantic relationship, i.e. you want to pin it down to the point where the word "iron" took on the meaning ferrum, then words will have a much later "origin". You will also need to sharpen up the theory of meaning-sameness, because the word "chicken" has gone through a number of subtle meaning-changes starting with "young fowl", then specifically "young chicken" and finally "any chicken". Except that "chicken" also includes the meaning "cowardly", thus you need to be clear on whether you are looking for all of the current meanings, or just some of them.

On the other side of the equation is the possibility of tracing a word-qua-pronunciation back in time, so Modern "iron", Old English iren from isærn from isarnan back to 9perhaps is-(e)ro. The problem is that most words ultimately originate from a language for which there are no records and we generally can't say much about when it was spoken, though we can at least know that Proto Indo-European was spoken further in the past than 4,000 years ago. That doesn't mean that the origin of the word is whenever PIE was spoken, it means that the origin of the word was at least that far back, and possibly much further back, it's just that we can't point to specific evidence proving that it had to be older than that.

Of the two approaches, the more semantically-restricted approach is "more practical" because then there is at least a potential empirical question. It makes sense to ask when the word "dog" came to mean dog (a general term, not a specific breed if that is what it used to mean), or when the meaning of "meat" changed to its current "flesh" meaning.

  • Thanks very much for this--it's extremely helpful. I had in mind the "strict semantic relationship" question. If I were to go about answering that question, what is the best approach? Is this done by "human expertise" word by word, by some quantitative model? What are some of the best papers/books/projects that have done this already? Once again, I very much appreciate the advice.
    – Cory
    Jul 11, 2016 at 18:47
  • I think the Oxford English Dictionary (full version) is the best resource, since it traces different senses over time. It's impossible to know w.r.t. "ordinary usage" since we don't have particularly old sources on ordinary talk, just what is written down. It is also limited to "English" i.e. Old English and modern representatives, so if there was a semantic change from PIE to Germanic, that may be beyond their reach. www.etymonline.com is also a fairly good resource that engages pre-historic states of English.
    – user6726
    Jul 11, 2016 at 19:15
  • Thanks for the suggestion; I think the www.etymonline.com is what I'm looking for (which I'll have to combine with estimates of language age). This is because I'm looking for the very ancient occurrences of words. They also seem to list their raw sources, which would be quite helpful for me. I'll also be on the lookout for similar sources for non-English / non-Indo European words since I am interested in those as well. Thanks once again for all your help!
    – Cory
    Jul 12, 2016 at 17:57

As you already sketched out, it is a difficult question. Usually (e.g., in the OED) the age of a word in a certain language is defined by its first attestation. This works well for the large part of vocabulary that was introduced rather recently.

Only for the core words of the language that are inherited from ancient times, different methods are needed: Reconstructions of proto-languages, time estimates when those proto-languages were presumably spoken, evidence from archaelogical findings (e.g., what technologies, domesticated plants and animals were available at what time) all play a role for dating those words.

  • As with the other answer, I find this extremely helpful. Let's say I am interested in the words you term "core words from ancient times." Are there existing academic or other works (e.g. wiktionary) that attempt to make these estimates for a large number of words? Of course, we cannot know for sure what the answer is. But I'm mostly wondering if there's a way an outsider could peruse existing estimates or apply some existing method to make his own. Or is it the sort of thing where I'd have to get a PhD and make my thesis about dating the single word "iron"?
    – Cory
    Jul 11, 2016 at 18:53
  • @Cory Dating the single word "iron" will not be a PhD thesis unless you develop a new method for this purpose—which is definitely possible—but it can be a conference paper or a journal paper. Mar 26, 2019 at 15:56
  • Thanks for following up (even after so long). Are there standard methods to date a word like "iron" or a repository of dates for such words? In my googling I found estimated dates for (e.g.) PIE but no variation within different PIE words.
    – Cory
    Mar 26, 2019 at 19:24
  • The repositories are currently the etymological dictionaries available. The dates are kind of coded by giving the name of the oldest layer of proto-languages where the word can be reconstructed. Dating the proto-languages is another difficult issue, see linguistics.stackexchange.com/questions/25830/… Mar 27, 2019 at 10:12
  • Makes sense. I'm correct in understanding, though, that there is no accepted methodology to date words within a language (like PIE)? So we might estimate "iron" & "silver" as having roughly the same age range if both were part of PIE?
    – Cory
    Mar 27, 2019 at 17:53

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.