4

Prompted by the questions about "despite"/"in spite of" on ELL and EL&U I played in N-gram for in spite of, despite even though, although, however.

enter image description here

After 1750 there is a sharp rise in most of them. Did the general language change to start including negative clauses or did some word fall out of favor to increase popularity of all the rest?

I tried comparing the chart with "but" but despite the steady decline, it didn't register any sharp drops between 1730 and 1800. So what expression was in popular use in the role of these conjunctions, that they replaced?

(footnote: The graph for "However" may be wrong; Ngram refuses to acknowledge "However" can be a conjunction; (however_CONJ) yields a flat zero; so does (however - however_ADJ))

Community
  • 1
SF.
  • 11,386
  • They were in use well before 1.750. *However: (adv., conj.) is from late 14c., from how + ever.http://www.etymonline.com/index.php?term=however. Although*: is from early 14c., althagh, compound of all + though, showing once-common emphatic use of all.http://www.etymonline.com/index.php?allowed_in_frame=0&search=although&searchmode=none –  Sep 11 '15 at 09:33
  • @Josh61: So they functioned for 400 years before at some point of 18c they began rapidly gaining popularity; although almost quadrupled , however reaching 5x the original over a 100 years period. – SF. Sep 11 '15 at 09:40
  • @Josh61: Ngram is normalized over the number of books published in given year, so the number of books doesn't change the profile. Literacy might - but then what kinds of words/constructs were replaced? – SF. Sep 11 '15 at 09:45
  • I'm not sure I understand what you are looking for, however was used before 1.750, why are you assunimg it replaced a similar expression? –  Sep 11 '15 at 09:48
  • 1
    @Josh61: It was used LESS. Across the language, without some kind of revolutionary social changes that affect the way people think, the percentage of words fulfilling given function - like conjunctions expressing contrast - remains relatively similar over the time, even if proportions between different words of that function change. So unless in 1750 people suddenly began contrasting things far more often than before, the total usage of all these conjunctions should be roughly the same; rise of one means decline of other. – SF. Sep 11 '15 at 11:24
  • Where do you get 'a sharp rise after 1500' from? The ngram starts in 1700. Also ngram has very spotty collection of texts from before 1800, so inferences from that earlier time will be very questionable. – Mitch Sep 11 '15 at 15:19
  • @Mitch Me bad, I meant 1750, I have no clue how 1500 got there. – SF. Sep 13 '15 at 22:53
  • Do you know what books Google used as part of their Ngram database? The type of books used (in other words context) could make a difference...in fact, I'm sure it does. – michael_timofeev Sep 14 '15 at 01:00
  • @michael_timofeev actually, given the magnitude of the change, it wouldn't at all, statistically speaking, assuming the observed phenomena is in fact a thing. – Please stop being evil Sep 14 '15 at 06:07
  • @michael_timofeev: apples to apples, while the effect could be biased by the kind of books, the shift and the counterpart would be affected identically by the same bias. – SF. Sep 14 '15 at 07:53
  • @SF. That's a point I'm not capable of answering. I do feel this needs to be seriously explored given the heavy reliance on Google – michael_timofeev Sep 14 '15 at 08:59
  • @michael_timofeev: The analysis is correct within the ngram. All the internal errors cancel each other out so we can with full confidence conclude the frequency of 'however' within books analyzed by Google rose replacing the frequency of 'yet' in these books. The only true potential for error is the selection bias; how that reflects on common language of these times. Both Google's selection bias on which books were scanned, and medium selection bias, books not accurately reflecting the language. (a.k.a. "psychoanalysis is the study of the mind of friends and family of psychoanalysts") – SF. Sep 14 '15 at 09:48
  • @SF. I did some looking at the quality of the OCR from some of the older books (I suggest you do, too) and found some pretty bad renderings of text. For example one book said "In spite of all my human dignities," and Google rendered it as "In spite of all my hitman dignity." some words were just rendered as Mcx,jut or other such garbage. I think we really need to start reconsidering the quality of Ngram. Also, many of the sources seemed to be religious commentary...nothing against religion but isn't this source going to have different occurrences of words? – michael_timofeev Sep 15 '15 at 16:14

2 Answers2

6

Adding the word "yet" in seems to show a fairly large decline in its usage - could this be the word you're looking for?

Edd
  • 504
1

It's not exactly a synonym, but notwithstanding seems to have been on the wane since the early 18th Century, just as 'however' was taking off.

JHCL
  • 5,096
  • It could have contributed, but the volume by which it declined is disproportionally small comparing to the volume "However" grew by. – SF. Sep 13 '15 at 22:56
  • @SF. you're putting a lot of faith in Ngram and the results it provides. How many books were catalogued before 1850 and how accurate was the OCR they used? – michael_timofeev Sep 14 '15 at 01:02
  • @michael_timofeev The OCR would be just as inaccurate for the word with positive trend as for the one with negative; the biases would cancel each other out. 1750's ngram is composed of 105214 pages; 19mln words for that year alone. That's 15,000 occurrences of "yet" and 2000 occurrences of "However". Plenty enough for statistically valid results. – SF. Sep 14 '15 at 08:37