The cost of commas?
« previous post | next post »
My 1/2/2025 post "American health care in 1754" quoted at length from Benjamin Franklin's account of the founding of Pennsylvania Hospital. The main point was the striking difference between then and now in the attitudes of (some) business leaders. But since this is Language Log rather than Health Care Politics Log, I suggested "the obvious stylistic change in sentence length" as a linguistic angle, with a link to the slides for my presentation at SHEL12 in 2022, "Historical trends in English sentence length and syntactic complexity". And Julian reponded in the comments: "Clearly commas were cheaper, in those days".
Punctuation is definitely part of the picture — especially semicolons, as this figure from my 2022 presentation illustrates:
But commas, not so much. It's plausible that longer sentences also tend to use more commas, but I didn't bother to check for that. I could redo the 75-work semicolon analysis to count commas, and maybe I'll do that later if there's interest, but for now, here's a illustrative comparison between Franklin's 1754 publication and David Lodge's 1984 Small World:
| Title | Semicolons per 100k chars | Commas per 100k chars |
|---|---|---|
| Franklin 1754 | 145.3 | 186.5 |
| Small World | 1.677 | 138.3 |
Franklin's 145.3/100k semicolon rate (=1.453%) is on the low side for the middle of the 18th century, at least according to the graph above. But it's still more than 86 times David Lodge's 1.677 /100k (=0.01677%) rate (145.3/1.677 =86.64).
In contrast, Franklin's 186.5/100k comma rate (=1.865%) is only about 35% greater than David Lodge's rate of 138.3/100k (=1.383%) — 186.5/138.3 = 1.345.
There's some relevant discussion in "Death before syntax?", 10/20/2014.
And as a final note, it's worth thinking about whether there's an genuine analogy to the economic effects of pricing/supply/demand/etc., as suggested by Julian's metaphor. The typographical cost of semicolons and commas doesn't change, and there's no obvious sense in which the supply changes, though maybe we can think of cultural patterns as a kind of supply variable? But the desire for longer and more "periodic" sentences arguably increases the demand, I guess.
RfP said,
January 5, 2025 @ 4:44 pm
I would be interested in seeing figures for the difference in the relative use of periods between Franklin and Lodge, in relation to the use of commas and semicolons.
Hasn’t the period replaced both the semicolon and the comma in many different contexts in contemporary writing?
And the cost savings in ink and lead are nothing to sneeze at. I mean, given how small a period is compared to a comma—and especially a semicolon!
~flow said,
January 5, 2025 @ 4:49 pm
Well, longer sentences, one should think, do also, to a degree, increase the cost, as it were, for the reader, inclined or not, to correctly parse, and understand the text.
Philip Taylor said,
January 5, 2025 @ 4:52 pm
~flow, did you perhaps mean "to correctly parse, and understand, the text" or "to correctly parse and understand the text" ?
Guy Plunkett said,
January 5, 2025 @ 5:10 pm
I must be an old soul; I use semicolons (and parentheses, em-dashes, and ellipses) quite often, even in informal writing.
RfP said,
January 5, 2025 @ 5:51 pm
@ Guy
Likewise, I’m sure.
And, in fact, I’ve been using ellipses a lot more often lately as a way to enhance the comprehension of people who just aren’t used to parsing long sentences anymore.
There are times when I want to join complex clauses in a single sentence… and adding that small amount of punctuation seems (in my own mind, at any rate) to give people enough additional mental breathing room to help them navigate what for many has become unnavigable.
Andrew Usher said,
January 5, 2025 @ 6:48 pm
Like RfP, my first reaction is that the reduction in other punctuation must be compensated to some degree by an increase in the periods at the end of sentences. And how likely is _that_ sentence to have been written without commas in that 18th-century style? Sentence length and its correlates are not the only thing affecting punctuation usage.
k_over_hbarc at yahoo dot com
Viseguy said,
January 5, 2025 @ 7:10 pm
I've always favored semicolons, as an indispensable way of emphasizing the logical or semantic joinder of two thoughts that could otherwise stand alone as separate sentences. Lately, however, I find myself pausing to consider how much, if anything, is lost by putting a period and making two sentences instead, especially in informal writing. Asking myself whether — given the intended reader(s) — the semicolon does more to hinder comprehension than aid it? I dunno; I'm of two minds.
~flow said,
January 6, 2025 @ 4:56 am
@Philip Taylor perque no los dos?
Philip Taylor said,
January 6, 2025 @ 5:52 am
Well, I feel the construct needs two or none — what do others think ?
David W said,
January 6, 2025 @ 10:31 am
I believe that Philip Taylor is right.
"to correctly parse, and understand the text" lacks an object for "parse."
Robert Coren said,
January 6, 2025 @ 10:53 am
It's my impression, not backed by any study or analysis, that the concept that every comma performs a more-or-less well-defined grammatical function is a relatively modern one. Back in Franklin's time, one might reasonably expect to see a comma pretty much anywhere where, in speech, one might pause to draw breath.
@Guy: I don't know if I'm an old soul, but I definitely occupy an old body, and I use a lot of semicolons, even in highly informal contexts such as Facebook.
P.S.: I noticed, while proofreading the above, that I had put a comma after "function", which, according to my formulation, would have been perfectly acceptable in the 18th century, but would be technically "wrong" now.
Chris Button said,
January 6, 2025 @ 4:48 pm
@ RfP
Indeed! I have heard/read somewhere that some printing presses omitted optional commas for such savings, which included paper savings too. That could be entirely apocryphal though.
Yes, I think so. Today's journalism contains lots of periods and lots of paragraph breaks.
RfP said,
January 6, 2025 @ 6:26 pm
@ Chris
That was an attempt at dry humor. But now that you mention it, I wouldn’t be surprised at certain printers in certain times and places doing everything they could to pinch pennies in this way.
Barbara Phillips Long said,
January 6, 2025 @ 11:37 pm
During Franklin’s time, printers hand-set type and each typeface had limited numbers of letters, numbers, punctuation, and other symbols. So there was a cost per comma and a cost for setting each comma that modern computer type-setting technology does not have. The evidence of Franklin’s prose, however, seems to indicate printers were not editing texts based on comma costs.