I think Gruber’s take is bad, and Microsoft’s decision is bad as well. I’ll explain my reasoning. Firstly, some historical context. For those who write in Latin script (including the entire...
Exemplary
I think Gruber’s take is bad, and Microsoft’s decision is bad as well.
I’ll explain my reasoning.
Firstly, some historical context. For those who write in Latin script (including the entire English-writing world), putting spaces between words is a relatively new thing. For the first ~1000 years of its existence, Latin script was written as either scriptio continua, or also with an interpunct (or "middle dot") between words. The invention of separating words with punctuation was very helpful as it disambiguates instances where a continuous stream of letters could be analyzed multiple ways—that is, separating words has the benefit of preventing graphical ambiguities.
E.g.: "penisland"
Is it "penis land" or "pen island"? The difference seems non-negligible.
So, the practice of an interpunct, or the more recent innovation of putting whitespace between characters, is helpful. However, I’ll argue that spaces are inferior to explicit marks, at least when working with digital text.
Whitespace, or negative space—i.e., the space between text—occurs between many orthographic units of text. For instance, in Latin script, in the lowercase I, ⟨i⟩, the shaft is separated from from the tittle with whitespace. Yes, the entire grapheme, ⟨i⟩, is composed in the Unicode code point U+0069 (as distinguished from the uppercase ⟨I⟩ which has code point U+0049). But, if you are not using a computer to write, or if you are using a computer, but are, say, using a writing implement, such as a stylus, pen, or pencil, or otherwise putting marks down on canvas, virtually or otherwise, but not using a text input system that composes the tittle for you, you will have to pick up your writing implement to, as the saying goes, "dot your i's".
In non-cursive writing, whitespace is also used between letters within words. To make some adjacent graphemes look nicer, some fonts and typesetting systems will present some pairs of adjacent graphemes with alternative forms that actually remove the space between letters using orthographic ligatures. Generally, though, in text written in the Latin alphabet, there is a clear hierarchy of space between orthographic atoms.
In ascending order based on the units of text that are separated, the hierarchy is as follows:
The space between components of individual graphemes. E.g., the stem of the lowercase ⟨i⟩ is separated from the tittle by space, or the dieresis is separated from the ⟨u⟩ in ⟨ü⟩.
The space between graphemes within words. E.g., the space between ⟨a⟩ and ⟨n⟩ in the word "an". (As mentioned before, sometimes such spaces are removed in favor of ligatures that connect adjacent graphemes, such as adjacent ⟨f⟩s composing to ⟨ff⟩ which is encoded in Unicode with its own code point as U+FB00. Depending on the typesetting presentation you use, this may be handled by the font with the underlying textual representation containing the source sequence ff—two instances of code point U+0066.
The space between words themselves. Though, recall that historically this space was either nonexistent, such as in many surviving written works of Classical Latin, or was instead marked with explicit interpuncts delimiting word boundaries.
The space between sentences. For languages that use the Latin alphabet, sentences are customarily marked with other kinds of sentence-final punctuation, but after such punctuation, there is some space before subsequent text.
And, finally, there is space between larger units of text, such as the space between paragraphs or the space between list items (such as in this very list).
When writing text, it is highly desirable to separate the concerns of tracking the content and the presentation of the content. This is a goal of languages like HTML or LaTeX. As such, the exact magnitude of the space occupied by each of the semantic separations in the hierarchy I’ve laid out above is a presentational concern. When you are writing HTML (as Gruber rightly points out), contiguous whitespace will be collapsed.
However, when typing text on a computer, there is utility in making a semantic distinction between word-separating space and sentence-separating space within the content, regardless of how it is ultimately presented. Items 1 and 2 do not need explicit markings, though one could devise special non-printing characters that could represent these. In practice, they can remain implicit. And in Unicode, we already have special non-printing characters we use to represent the fifth level of spacing. Namely, when you press the return key on your keyboard, it usually invisibly adds a \n—Unix new line U+000a—or \n\r—Unix new line and a carriage return U+000aU+000dto your text.
But, in the model of the hierarchical semantics of textual spacing I introduced, it is still desirable to distinguish 3 from 4, and 4 from 5! And we have yet to discuss a mechanism by which we can make this distinction. A convention arose naturally to make this distinction (after the use of interpuncts fell out of favor). This convention is to use a single space to delimit words, and two spaces to delimit sentences within paragraphs!
What benefit do you get from this? Well, if you are typing text in a software that attempts to merge the concerns of content and presentation (such as MS Word, or other softwares), the benefits are non-obvious, and in fact will be immediately lost. So, while Microsoft’s decision is unsurprising, it is no less disappointing and even frustrating.
If you are creating an HTML document, or LaTeX document, etc., where your aim is to maintain a separation between content and presentation, or even if you are just using a plain-text editor that is capable of indicating non-printing characters, you will get a huge benefit from making an explicit distinction between word- and sentence-delimiting whitespace!
Let me show you the difference with a pair of screenshots from TextMate (my preferred text editor):
In the first screenshot, the "invisible" non-printing characters are displayed: spaces are indicated with ⟨⸱⟩ and newlines are indicated with ⟨¬⟩. With this configuration, all text is visible and explicit. This is a highly preferable way to work with text than in the second screenshot where implicit negative space may hide innumerable non-printing characters. For instance, this mode of editing text lets me see clearly when something bogus is happening, such as if I have a rogue byte-order mark in the middle of some text.
Now, when we go on to publish documents for human eyes to consume, you usually don’t want all the extra visual indicators that you see when editing the source text. It ends up being quite noisy. Luckily, software can do all this for us. But, if we conflate presentation and content with bad WYSIWYG software models for text, we end up with all sorts of problems, not the least of which is that we can’t distinguish word- and sentence-delimiters in Latin text.
My God, man, have you introduced grammatical prescriptivism to typesetting? Is the world not crazy enough? (╯°□°）╯︵ ┻━┻ In all seriousness, though, English is not Latin. It is so far removed from...
My God, man, have you introduced grammatical prescriptivism to typesetting? Is the world not crazy enough?
(╯°□°）╯︵ ┻━┻
In all seriousness, though, English is not Latin. It is so far removed from that clearly dead language it practically has nothing to do with it. There is literally no reason why we should be looking to a language that is quite literally missing a millennium of evolution to influence the English language. Quite simply, Latin is no longer an accurate description of the way people think, which means it is a poor choice to model communication upon.
It has nothing to do with grammar and everything to do with textual semantics! Microsoft is the one being prescriptive! No, but Latin script is the current standard system for written English. I...
My God, man, have you introduced grammatical prescriptivism to typesetting?
It has nothing to do with grammar and everything to do with textual semantics! Microsoft is the one being prescriptive!
In all seriousness, though, English is not Latin.
No, but Latin script is the current standard system for written English.
There is literally no reason why we should be looking to a language that is quite literally missing a millennium of evolution to influence the English language.
I don’t think you’ve understood what I wrote if that’s what you took away. The idea of explicitly visualizing text when we input it into a computer is a good idea that Latin script innovated, but then lost. Other orthographies include the middle dot as an explicit word delimiter, such as Katakana. It has nothing to do with English, specifically. Some writing systems have entirely different, yet totally expressive orthographic systems for explicitly encoding the semantics of document structure. E.g., Tibetan script.
Latin is not a script that we write in. We write in one of a number of English scripts. Latin script is actually a fairly modern invention which came as a result of the rise of movable type, where...
Latin is not a script that we write in. We write in one of a number of English scripts. Latin script is actually a fairly modern invention which came as a result of the rise of movable type, where one don't could represent a number of languages. The fact we don't use Latin script is actually fairly self-evident - the English language does not use a number of symbols and characters that are necessity for other languages written in "Latin script", such as ß or ø.
I understand what you are trying to say, but you are completely ignoring the history of typography and why things look the way they do today. We have completely different symbols today specifically because they take advantage of the way we read, such as ligatures.
You are also discounting how influencial type is on how people write. We are so used to seeing type that we now almost entirely write in block letters.
There's also a problem with your insurance of blending content and presentation. First of all, we can't actually type the full range of English script. They don't have a keyboard layout that includes all they keys you would need. To use the example at hand, two spaces are not how we end sentences when writing, we use a single longer space. And to be frank, the entire reason why people used two spaces on typewriters - it was a hack to produce that longer space.
It is literally called “Latin script”. You can also call it Roman script, but that’s a bit revisionist. And yes, the Latin script includes other letters beyond what are commonly used in English...
Latin is not a script that we write in.
It is literally called “Latin script”. You can also call it Roman script, but that’s a bit revisionist. And yes, the Latin script includes other letters beyond what are commonly used in English writing today. None of that has any bearing on how we use spaces between words and sentences.
I understand what you are trying to say, but you are completely ignoring the history of typography and why things look the way they do today.
Rather than making vague claims, could you please explain what I’m ignoring?
You are also discounting how influencial type is on how people write. We are so used to seeing type that we now almost entirely write in block letters.
No, I’m aware of how, for instance, German typesetting influenced the loss of the thorn in English writing among other things. None of that is really relevant to the discussion of the use of space delimiters, though.
There's also a problem with your insurance of blending content and presentation. First of all, we can't actually type the full range of English script. They don't have a keyboard layout that includes all they keys you would need. To use the example at hand, two spaces are not how we end sentences when writing, we use a single longer space. And to be frank, the entire reason why people used two spaces on typewriters - it was a hack to produce that longer space.
I don’t want to blend content and presentation! We can use a double-space as a symbol for the presentation software to typeset appropriately. I really don’t understand what’s so difficult to understand. You’re acting like we don’t do similar things in markup languages all the time. E.g., there is special semantics of ending a line two spaces in Markdown. Or we use special entities in HTML, or use elements like <p>.
I think you and I are both misunderstanding each other. And honestly this argument is over something so meaningless in the greater scheme of things, I'd rather just apologise for if I upset you...
I think you and I are both misunderstanding each other. And honestly this argument is over something so meaningless in the greater scheme of things, I'd rather just apologise for if I upset you and call it a night.
I’m not upset, I just don’t understand. If you’d like to drop it, that’s fine with me. Sorry if my tone seemed combative, but my question for clarification was genuinely out of curiosity, not...
I’m not upset, I just don’t understand. If you’d like to drop it, that’s fine with me. Sorry if my tone seemed combative, but my question for clarification was genuinely out of curiosity, not defensiveness.
I may be drastically missing the point, but wouldn't it be reasonable to consider U+0020U+002E (dot-space) as the sentence delimiter, compared to space alone (or, I suppose, space preceded by...
I may be drastically missing the point, but wouldn't it be reasonable to consider U+0020U+002E (dot-space) as the sentence delimiter, compared to space alone (or, I suppose, space preceded by anything that isn't a dot) as the word delimiter?
"Hello, Mr. Smith." is one sentence. That's also one of the things that makes searching for a sentence delimiter more difficult in text that doesn't end sentences with two spaces.
"Hello, Mr. Smith." is one sentence. That's also one of the things that makes searching for a sentence delimiter more difficult in text that doesn't end sentences with two spaces.
It’s reasonable to consider anything. To be clear, I’m not advocating for spaces, alone, to be the delimiter. I’m advocating for two spaces to follow the conventional sentence-final punctuation....
It’s reasonable to consider anything. To be clear, I’m not advocating for spaces, alone, to be the delimiter. I’m advocating for two spaces to follow the conventional sentence-final punctuation. The point of the two spaces is to contrast with the single space between words (which may also follow punctuation, such as after abbreviations). The criteria for deciding what is a best practice ought to include what you can reasonably get people to adopt as a de facto standard, what is easy to type on standard keyboard layouts, and also what is unambiguous.
The example given above of "Hello, Mr. Smith." includes such a sequence. In general, where periods are used for abbreviations, they should be followed by a word-space, not a sentence-space. LaTeX...
The example given above of "Hello, Mr. Smith." includes such a sequence. In general, where periods are used for abbreviations, they should be followed by a word-space, not a sentence-space.
LaTeX resolves this by having you code that sentence as Hello, Mr.~Smith., using a sequence other than period-space for non-sentence-ending periods.
You’d have to ask a professional typesetter. I just go by what looks good to my personal aesthetics, which, if I’m being honest, isn’t particularly sensitive to the presentation. Tildes, for...
Is it ever presentationally correct
You’d have to ask a professional typesetter. I just go by what looks good to my personal aesthetics, which, if I’m being honest, isn’t particularly sensitive to the presentation. Tildes, for instance, collapses my double-space delimited text in my Markdown source and presents them as single spaces, and I think it looks OK. Again, I care more about how words and sentences are delimited in the source text, so that is where I have a strong opinion.
I think I understand. I want my document processor to make every typographic decision for me. So I find "two spaces or one" kind of trivial since if the processor can't make it look good, it has...
I think I understand.
I want my document processor to make every typographic decision for me. So I find "two spaces or one" kind of trivial since if the processor can't make it look good, it has failed as a typesetter. But that never happens, because as you mention it always looks good anyway.
When I hand write, I have never consciously put space between sentences. Usually the smallest units I think of are words; when I'm at the end of a sentence I mark a dot, leave space, and continue. I sort of think of sentences as an emergent property of the punctuation, rather than the sentences prescribing the punctuation.
So when I'm editing a document I don't really care about making the space between sentences different than the space between words, because I'm not thinking on that level. It's like how I don't really care about the encoding.
I don't know why I wrote so much. Guess I'm curious what you think about those details.
My whole point is that when putting text into a computer, I am ideally writing the document source text. As such, the distinction between word- and sentence-delimiting spacing is a semantic...
So I find "two spaces or one" kind of trivial since if the processor can't make it look good, it has failed as a typesetter. But that never happens, because as you mention it always looks good anyway.
My whole point is that when putting text into a computer, I am ideally writing the document source text. As such, the distinction between word- and sentence-delimiting spacing is a semantic distinction.
So when I'm editing a document I don't really care about making the space between sentences different than the space between words, because I'm not thinking on that level. It's like how I don't really care about the encoding.
Right, and I do care about the encoding! The semantic distinction between word and sentence boundaries is purely about the encoding of text and ideally has nothing to do with the presentation. In software that conflates content and presentation, however, the desiderata come into conflict. Unfortunately, presentation often wins in that clash and then we potentially end up with totally broken documents.
Not to be glib ― if the distinction is purely semantic and it's not possible to see the difference, then what's the point of it? I guess what I'm asking is, if I don't think "end of sentence" when...
Not to be glib ― if the distinction is purely semantic and it's not possible to see the difference, then what's the point of it?
I guess what I'm asking is, if I don't think "end of sentence" when I conceive a thought, and I don't notate it when hand writing, and I shouldn't see it in printed material, then why should it be meaningful when I type into a computer?
Sorry, I know this conversation is kind of tiring. I'm pulling because I feel like you might be able to clarify something for me, but I don't know what.
It is possible to see the difference in the source with appropriate software (see my comparative screenshots). There are myriad reasons, but an obvious one would be to facilitate automatic...
Not to be glib ― if the distinction is purely semantic and it's not possible to see the difference, then what's the point of it?
It is possible to see the difference in the source with appropriate software (see my comparative screenshots).
I guess what I'm asking is, if I don't think "end of sentence" when I conceive a thought, and I don't notate it when hand writing, and I shouldn't see it in printed material, then why should it be meaningful when I type into a computer?
There are myriad reasons, but an obvious one would be to facilitate automatic counting of the number of sentences in a document.
You might like LyX.
Actually, I really dislike WYSIWYG editors. LyX is based on LaTeX, which I do like a lot! I just use LaTeX directly by writing the source in my text editor, TextMate, with the LaTeX bundle.
LyX isn't WYSIWYG, it's very highly structured. It doesn't let you make bad markup like "boldface" spaces, and it doesn't even let you change font size, unless you indicate "this text is a chapter...
LyX isn't WYSIWYG, it's very highly structured. It doesn't let you make bad markup like "boldface" spaces, and it doesn't even let you change font size, unless you indicate "this text is a chapter title" or whatever. It forces the semantic structure to make sense by default. You really might like it! :-)
(I switched from LyX to plain LaTeX at some point. Now I'm on Markdown-with-TeX-formulas which I find really cozy.)
Wow, awesome post. Some years ago I decided to start separating sentences using single newlines in LaTeX. This way I know that each line is a sentence, so most of the time navigating vertically is...
Wow, awesome post.
Some years ago I decided to start separating sentences using single newlines in LaTeX. This way I know that each line is a sentence, so most of the time navigating vertically is moving between sentences. This obviously works because in LaTeX paragraphs are marked by having 2 newlines.
That’s a nice way to do it in LaTeX source (and should work in some other languages as well). The issue with reusing something like Unix new lines for sentence separation, is that then that is now...
That’s a nice way to do it in LaTeX source (and should work in some other languages as well). The issue with reusing something like Unix new lines for sentence separation, is that then that is now ambiguous with paragraph or other unit separators. I think just using two spaces after sentences is a more universal/portable convention, because there is no other semantic use of two spaces in a row that I’m aware of (other than in Markdown, but that is two spaces followed by a Unix new line, which isn’t ambiguous with that added context). Whatever works for you, though!
Can you explain this part a bit better? Do you mean in general or in my use case? Since paragraphs, like I said, are 2 newlines.
The issue with reusing something like Unix new lines for sentence separation, is that then that is now ambiguous with paragraph or other unit separators.
Can you explain this part a bit better? Do you mean in general or in my use case? Since paragraphs, like I said, are 2 newlines.
It’s only ambiguous outside of LaTeX, but it also prevents you being able to use a new line as a “hard break” in your LaTeX source, since you’re imbuing a single new with your idiosyncratic semantics.
It’s only ambiguous outside of LaTeX, but it also prevents you being able to use a new line as a “hard break” in your LaTeX source, since you’re imbuing a single new with your idiosyncratic semantics.
I linked to Gruber's post on this purely because it's such a satisfying schadenfreude-esque take on the situation, especially for him, as he's a long time opinionated grammar snob. If you'd like a...
I linked to Gruber's post on this purely because it's such a satisfying schadenfreude-esque take on the situation, especially for him, as he's a long time opinionated grammar snob. If you'd like a more objective take on the situation, check out The Verge's article.
This hurts my soul. I don't use Word so it shouldn't, but for whatever reason this feels like a personal attack. Seems like a strange thing for a word processor to be opinionated about, though....
This hurts my soul. I don't use Word so it shouldn't, but for whatever reason this feels like a personal attack. Seems like a strange thing for a word processor to be opinionated about, though. Even if I thought this was best practice, I don't want best practice beat into me by default in something like a word processor. I don't want a bunch of squiggly lines appearing because the developers don't like how I format things. I want it to get out of my way unless I tell it otherwise or there is something very obviously wrong.
A member of general population is more likely to put two spaces in by mistake. Not a single person teaching me language even mentioned the possibility of two spaces after a period. I've never...
A member of general population is more likely to put two spaces in by mistake. Not a single person teaching me language even mentioned the possibility of two spaces after a period. I've never encountered it in the wild (except as a mistake in an otherwise-single-spaced text). This is the first I'm even hearing about the possibility.
Basically anyone who was in the working world back when typewriters were a thing still uses two spaces. It hate it. I make a point of doing a find-and-replace for all “.__ “ with “._” before I...
Basically anyone who was in the working world back when typewriters were a thing still uses two spaces.
It hate it. I make a point of doing a find-and-replace for all “.__ “ with “._” before I start tracking changes any time I have to edit their work. It’s one of those things that’s way easier to do than it is to undo. 😈
Yeah, it seems to be a generational thing. We use a lot of shared documents at work, and almost all of the older workers use two spaces after a period, while all of the younger workers use one....
Yeah, it seems to be a generational thing. We use a lot of shared documents at work, and almost all of the older workers use two spaces after a period, while all of the younger workers use one. Collaborative documents often have a smattering of one and two spaces, depending on which people edited which parts.
Of all the people on staff, I think I'm the only person who notices or cares about this. I've pointed it out a couple times and the response has always been some form of "huh, I didn't notice" or "wow, you can really tell just by looking?"
My memory is fuzzy, so the timing might be off, but I definitely was taught to do this on a computer. It was either in late elementary school or middle school during a class about word processors....
My memory is fuzzy, so the timing might be off, but I definitely was taught to do this on a computer. It was either in late elementary school or middle school during a class about word processors.
I know it's not standard any more and it may have been wrong when I was taught too! But honestly it's so ingrained in me at this point I'm not sure I can stop doing it.
That's alright if you use it for yourself. If you send your documents to other people, then it should. Double space is a minor thing, but I still pray for the day to finally come, when Word will...
I don't want best practice beat into me by default in something like a word processor
That's alright if you use it for yourself. If you send your documents to other people, then it should. Double space is a minor thing, but I still pray for the day to finally come, when Word will open up the skies and strike down with thunder all those who move stuff around the page with multiple spaces and line breaks instead of proper formatting.
I think writing a text and putting in to paper or the screen are two separate matters and there's no need to intertwine both. For example text alignment is something I don't see many people doing...
I don't want a bunch of squiggly lines appearing because the developers don't like how I format things
I think writing a text and putting in to paper or the screen are two separate matters and there's no need to intertwine both.
For example text alignment is something I don't see many people doing manually. Even when writing code people try to use text formatters so they don't have to do it themselves.
You can always turn it off. I don't see how enforcing best practice with sane defaults instead of an obsolete standard is hurtful. https://practicaltypography.com/one-space-between-sentences.html
You can always turn it off. I don't see how enforcing best practice with sane defaults instead of an obsolete standard is hurtful.
I wanna say that you're actually right. It's fine to feel like this is exactly not what you'd want your software to do, the whole point of there being multiple word processors is to allow you to...
I wanna say that you're actually right. It's fine to feel like this is exactly not what you'd want your software to do, the whole point of there being multiple word processors is to allow you to choose, after all.
I use Word a lot, mostly to write essays for classes or documents for organizations. It being opinionated is useful for me there because those are exactly the kind of texts where you'd have an editor/supervisor come in and tell that, no, we don't use Oxford commas and this part is too lengthy and needs to be shortened.
But if you want a word processor for creative writing, quick note-taking, or really any kind of writing where you want to be in completely in charge: Word's probably not the best choice. It has a bunch of options you don't need and most options you might need are available elsewhere.
Ah I see, because people hardwired putting two spaces from typewriter days. I thought It would be something like that, but had to check the comments here.
Ah I see, because people hardwired putting two spaces from typewriter days. I thought It would be something like that, but had to check the comments here.
idk. I've never typed on a typewriter. got to say that QWERTY is the more painful legacy from the typewriter days than an extra space between sentences. This is similar to if Microsoft were coming...
idk. I've never typed on a typewriter. got to say that QWERTY is the more painful legacy from the typewriter days than an extra space between sentences. This is similar to if Microsoft were coming at python programmers with the old spaces versus tabs argument, and insisting on one side as the right answer. It's cute they think they get to decide. Microsoft has a long legacy of being preachy bastards though, so not terribly surprising.
Subtext: I'm not salty at all that Microsoft autocorrects a single hyphen to be a dash character, which is available in utf-8 encoding but not available in ASCII, thereby potentially breaking some of the older code I have to work with on a regular basis if the user accidentally lets Microsoft decide their delimiter for them.
I have typed on a typewriter. The worst part of it is not the keyboard layout, but the way the keys are angled upward. Horrible for ergonomics. The flat keyboards that are predominant nowadays...
I have typed on a typewriter. The worst part of it is not the keyboard layout, but the way the keys are angled upward. Horrible for ergonomics. The flat keyboards that are predominant nowadays aren't great for ergonomics, but they're 10x better than that atrocity.
I found out about this change in writing style (standard?) And try to only single space, going as far as to do a search/replace for it. Writing a recent paper, I was surprised at it being flagged.
I found out about this change in writing style (standard?) And try to only single space, going as far as to do a search/replace for it. Writing a recent paper, I was surprised at it being flagged.
I think Gruber’s take is bad, and Microsoft’s decision is bad as well.
I’ll explain my reasoning.
Firstly, some historical context. For those who write in Latin script (including the entire English-writing world), putting spaces between words is a relatively new thing. For the first ~1000 years of its existence, Latin script was written as either scriptio continua, or also with an interpunct (or "middle dot") between words. The invention of separating words with punctuation was very helpful as it disambiguates instances where a continuous stream of letters could be analyzed multiple ways—that is, separating words has the benefit of preventing graphical ambiguities.
E.g.: "penisland"
Is it "penis land" or "pen island"? The difference seems non-negligible.
So, the practice of an interpunct, or the more recent innovation of putting whitespace between characters, is helpful. However, I’ll argue that spaces are inferior to explicit marks, at least when working with digital text.
Whitespace, or negative space—i.e., the space between text—occurs between many orthographic units of text. For instance, in Latin script, in the lowercase I, ⟨i⟩, the shaft is separated from from the tittle with whitespace. Yes, the entire grapheme, ⟨i⟩, is composed in the Unicode code point
U+0069(as distinguished from the uppercase ⟨I⟩ which has code pointU+0049). But, if you are not using a computer to write, or if you are using a computer, but are, say, using a writing implement, such as a stylus, pen, or pencil, or otherwise putting marks down on canvas, virtually or otherwise, but not using a text input system that composes the tittle for you, you will have to pick up your writing implement to, as the saying goes, "dot your i's".In non-cursive writing, whitespace is also used between letters within words. To make some adjacent graphemes look nicer, some fonts and typesetting systems will present some pairs of adjacent graphemes with alternative forms that actually remove the space between letters using orthographic ligatures. Generally, though, in text written in the Latin alphabet, there is a clear hierarchy of space between orthographic atoms.
In ascending order based on the units of text that are separated, the hierarchy is as follows:
The space between components of individual graphemes. E.g., the stem of the lowercase ⟨i⟩ is separated from the tittle by space, or the dieresis is separated from the ⟨u⟩ in ⟨ü⟩.
The space between graphemes within words. E.g., the space between ⟨a⟩ and ⟨n⟩ in the word "an". (As mentioned before, sometimes such spaces are removed in favor of ligatures that connect adjacent graphemes, such as adjacent ⟨f⟩s composing to ⟨ff⟩ which is encoded in Unicode with its own code point as
U+FB00. Depending on the typesetting presentation you use, this may be handled by the font with the underlying textual representation containing the source sequenceff—two instances of code pointU+0066.The space between words themselves. Though, recall that historically this space was either nonexistent, such as in many surviving written works of Classical Latin, or was instead marked with explicit interpuncts delimiting word boundaries.
The space between sentences. For languages that use the Latin alphabet, sentences are customarily marked with other kinds of sentence-final punctuation, but after such punctuation, there is some space before subsequent text.
And, finally, there is space between larger units of text, such as the space between paragraphs or the space between list items (such as in this very list).
When writing text, it is highly desirable to separate the concerns of tracking the content and the presentation of the content. This is a goal of languages like HTML or LaTeX. As such, the exact magnitude of the space occupied by each of the semantic separations in the hierarchy I’ve laid out above is a presentational concern. When you are writing HTML (as Gruber rightly points out), contiguous whitespace will be collapsed.
However, when typing text on a computer, there is utility in making a semantic distinction between word-separating space and sentence-separating space within the content, regardless of how it is ultimately presented. Items 1 and 2 do not need explicit markings, though one could devise special non-printing characters that could represent these. In practice, they can remain implicit. And in Unicode, we already have special non-printing characters we use to represent the fifth level of spacing. Namely, when you press the return key on your keyboard, it usually invisibly adds a
\n—Unix new lineU+000a—or\n\r—Unix new line and a carriage returnU+000aU+000dto your text.But, in the model of the hierarchical semantics of textual spacing I introduced, it is still desirable to distinguish 3 from 4, and 4 from 5! And we have yet to discuss a mechanism by which we can make this distinction. A convention arose naturally to make this distinction (after the use of interpuncts fell out of favor). This convention is to use a single space to delimit words, and two spaces to delimit sentences within paragraphs!
What benefit do you get from this? Well, if you are typing text in a software that attempts to merge the concerns of content and presentation (such as MS Word, or other softwares), the benefits are non-obvious, and in fact will be immediately lost. So, while Microsoft’s decision is unsurprising, it is no less disappointing and even frustrating.
If you are creating an HTML document, or LaTeX document, etc., where your aim is to maintain a separation between content and presentation, or even if you are just using a plain-text editor that is capable of indicating non-printing characters, you will get a huge benefit from making an explicit distinction between word- and sentence-delimiting whitespace!
Let me show you the difference with a pair of screenshots from TextMate (my preferred text editor):
In the first screenshot, the "invisible" non-printing characters are displayed: spaces are indicated with ⟨⸱⟩ and newlines are indicated with ⟨¬⟩. With this configuration, all text is visible and explicit. This is a highly preferable way to work with text than in the second screenshot where implicit negative space may hide innumerable non-printing characters. For instance, this mode of editing text lets me see clearly when something bogus is happening, such as if I have a rogue byte-order mark in the middle of some text.
Now, when we go on to publish documents for human eyes to consume, you usually don’t want all the extra visual indicators that you see when editing the source text. It ends up being quite noisy. Luckily, software can do all this for us. But, if we conflate presentation and content with bad WYSIWYG software models for text, we end up with all sorts of problems, not the least of which is that we can’t distinguish word- and sentence-delimiters in Latin text.
My God, man, have you introduced grammatical prescriptivism to typesetting? Is the world not crazy enough?
(╯°□°）╯︵ ┻━┻
In all seriousness, though, English is not Latin. It is so far removed from that clearly dead language it practically has nothing to do with it. There is literally no reason why we should be looking to a language that is quite literally missing a millennium of evolution to influence the English language. Quite simply, Latin is no longer an accurate description of the way people think, which means it is a poor choice to model communication upon.
It has nothing to do with grammar and everything to do with textual semantics! Microsoft is the one being prescriptive!
No, but Latin script is the current standard system for written English.
I don’t think you’ve understood what I wrote if that’s what you took away. The idea of explicitly visualizing text when we input it into a computer is a good idea that Latin script innovated, but then lost. Other orthographies include the middle dot as an explicit word delimiter, such as Katakana. It has nothing to do with English, specifically. Some writing systems have entirely different, yet totally expressive orthographic systems for explicitly encoding the semantics of document structure. E.g., Tibetan script.
Latin is not a script that we write in. We write in one of a number of English scripts. Latin script is actually a fairly modern invention which came as a result of the rise of movable type, where one don't could represent a number of languages. The fact we don't use Latin script is actually fairly self-evident - the English language does not use a number of symbols and characters that are necessity for other languages written in "Latin script", such as ß or ø.
I understand what you are trying to say, but you are completely ignoring the history of typography and why things look the way they do today. We have completely different symbols today specifically because they take advantage of the way we read, such as ligatures.
You are also discounting how influencial type is on how people write. We are so used to seeing type that we now almost entirely write in block letters.
There's also a problem with your insurance of blending content and presentation. First of all, we can't actually type the full range of English script. They don't have a keyboard layout that includes all they keys you would need. To use the example at hand, two spaces are not how we end sentences when writing, we use a single longer space. And to be frank, the entire reason why people used two spaces on typewriters - it was a hack to produce that longer space.
It is literally called “Latin script”. You can also call it Roman script, but that’s a bit revisionist. And yes, the Latin script includes other letters beyond what are commonly used in English writing today. None of that has any bearing on how we use spaces between words and sentences.
Rather than making vague claims, could you please explain what I’m ignoring?
No, I’m aware of how, for instance, German typesetting influenced the loss of the thorn in English writing among other things. None of that is really relevant to the discussion of the use of space delimiters, though.
I don’t want to blend content and presentation! We can use a double-space as a symbol for the presentation software to typeset appropriately. I really don’t understand what’s so difficult to understand. You’re acting like we don’t do similar things in markup languages all the time. E.g., there is special semantics of ending a line two spaces in Markdown. Or we use special entities in HTML, or use elements like
<p>.I think you and I are both misunderstanding each other. And honestly this argument is over something so meaningless in the greater scheme of things, I'd rather just apologise for if I upset you and call it a night.
I’m not upset, I just don’t understand. If you’d like to drop it, that’s fine with me. Sorry if my tone seemed combative, but my question for clarification was genuinely out of curiosity, not defensiveness.
I may be drastically missing the point, but wouldn't it be reasonable to consider
U+0020U+002E(dot-space) as the sentence delimiter, compared to space alone (or, I suppose, space preceded by anything that isn't a dot) as the word delimiter?"Hello, Mr. Smith." is one sentence. That's also one of the things that makes searching for a sentence delimiter more difficult in text that doesn't end sentences with two spaces.
It’s reasonable to consider anything. To be clear, I’m not advocating for spaces, alone, to be the delimiter. I’m advocating for two spaces to follow the conventional sentence-final punctuation. The point of the two spaces is to contrast with the single space between words (which may also follow punctuation, such as after abbreviations). The criteria for deciding what is a best practice ought to include what you can reasonably get people to adopt as a de facto standard, what is easy to type on standard keyboard layouts, and also what is unambiguous.
Is it ever presentationally correct to have a low dot and space between two letters, where the space is smaller than the space between two sentences?
The example given above of "Hello, Mr. Smith." includes such a sequence. In general, where periods are used for abbreviations, they should be followed by a word-space, not a sentence-space.
LaTeX resolves this by having you code that sentence as
Hello, Mr.~Smith., using a sequence other than period-space for non-sentence-ending periods.You’d have to ask a professional typesetter. I just go by what looks good to my personal aesthetics, which, if I’m being honest, isn’t particularly sensitive to the presentation. Tildes, for instance, collapses my double-space delimited text in my Markdown source and presents them as single spaces, and I think it looks OK. Again, I care more about how words and sentences are delimited in the source text, so that is where I have a strong opinion.
I think I understand.
I want my document processor to make every typographic decision for me. So I find "two spaces or one" kind of trivial since if the processor can't make it look good, it has failed as a typesetter. But that never happens, because as you mention it always looks good anyway.
When I hand write, I have never consciously put space between sentences. Usually the smallest units I think of are words; when I'm at the end of a sentence I mark a dot, leave space, and continue. I sort of think of sentences as an emergent property of the punctuation, rather than the sentences prescribing the punctuation.
So when I'm editing a document I don't really care about making the space between sentences different than the space between words, because I'm not thinking on that level. It's like how I don't really care about the encoding.
I don't know why I wrote so much. Guess I'm curious what you think about those details.
My whole point is that when putting text into a computer, I am ideally writing the document source text. As such, the distinction between word- and sentence-delimiting spacing is a semantic distinction.
Right, and I do care about the encoding! The semantic distinction between word and sentence boundaries is purely about the encoding of text and ideally has nothing to do with the presentation. In software that conflates content and presentation, however, the desiderata come into conflict. Unfortunately, presentation often wins in that clash and then we potentially end up with totally broken documents.
Not to be glib ― if the distinction is purely semantic and it's not possible to see the difference, then what's the point of it?
I guess what I'm asking is, if I don't think "end of sentence" when I conceive a thought, and I don't notate it when hand writing, and I shouldn't see it in printed material, then why should it be meaningful when I type into a computer?
Sorry, I know this conversation is kind of tiring. I'm pulling because I feel like you might be able to clarify something for me, but I don't know what.
You might like LyX.
It is possible to see the difference in the source with appropriate software (see my comparative screenshots).
There are myriad reasons, but an obvious one would be to facilitate automatic counting of the number of sentences in a document.
Actually, I really dislike WYSIWYG editors. LyX is based on LaTeX, which I do like a lot! I just use LaTeX directly by writing the source in my text editor, TextMate, with the LaTeX bundle.
LyX isn't WYSIWYG, it's very highly structured. It doesn't let you make bad markup like "boldface" spaces, and it doesn't even let you change font size, unless you indicate "this text is a chapter title" or whatever. It forces the semantic structure to make sense by default. You really might like it! :-)
(I switched from LyX to plain LaTeX at some point. Now I'm on Markdown-with-TeX-formulas which I find really cozy.)
Thanks for the conversation!
I’ve also experimented with Markdeep as a half-measure between Markdown and full LaTeX.
Wow, awesome post.
Some years ago I decided to start separating sentences using single newlines in LaTeX. This way I know that each line is a sentence, so most of the time navigating vertically is moving between sentences. This obviously works because in LaTeX paragraphs are marked by having 2 newlines.
That’s a nice way to do it in LaTeX source (and should work in some other languages as well). The issue with reusing something like Unix new lines for sentence separation, is that then that is now ambiguous with paragraph or other unit separators. I think just using two spaces after sentences is a more universal/portable convention, because there is no other semantic use of two spaces in a row that I’m aware of (other than in Markdown, but that is two spaces followed by a Unix new line, which isn’t ambiguous with that added context). Whatever works for you, though!
Can you explain this part a bit better? Do you mean in general or in my use case? Since paragraphs, like I said, are 2 newlines.
It’s only ambiguous outside of LaTeX, but it also prevents you being able to use a new line as a “hard break” in your LaTeX source, since you’re imbuing a single new with your idiosyncratic semantics.
I linked to Gruber's post on this purely because it's such a satisfying schadenfreude-esque take on the situation, especially for him, as he's a long time opinionated grammar snob. If you'd like a more objective take on the situation, check out The Verge's article.
This hurts my soul. I don't use Word so it shouldn't, but for whatever reason this feels like a personal attack. Seems like a strange thing for a word processor to be opinionated about, though. Even if I thought this was best practice, I don't want best practice beat into me by default in something like a word processor. I don't want a bunch of squiggly lines appearing because the developers don't like how I format things. I want it to get out of my way unless I tell it otherwise or there is something very obviously wrong.
Pretty sure that, like almost every grammar option, there is a setting you can turn on or off.
As it says in the other article, there is, which is why I'm talking about defaults specifically and not as if I can never type two spaces again.
That's fine. But if I send the document to others, will theirs now be full of red squigglies instead?
A member of general population is more likely to put two spaces in by mistake. Not a single person teaching me language even mentioned the possibility of two spaces after a period. I've never encountered it in the wild (except as a mistake in an otherwise-single-spaced text). This is the first I'm even hearing about the possibility.
Basically anyone who was in the working world back when typewriters were a thing still uses two spaces.
It hate it. I make a point of doing a find-and-replace for all “.__ “ with “._” before I start tracking changes any time I have to edit their work. It’s one of those things that’s way easier to do than it is to undo. 😈
Yeah, it seems to be a generational thing. We use a lot of shared documents at work, and almost all of the older workers use two spaces after a period, while all of the younger workers use one. Collaborative documents often have a smattering of one and two spaces, depending on which people edited which parts.
Of all the people on staff, I think I'm the only person who notices or cares about this. I've pointed it out a couple times and the response has always been some form of "huh, I didn't notice" or "wow, you can really tell just by looking?"
My memory is fuzzy, so the timing might be off, but I definitely was taught to do this on a computer. It was either in late elementary school or middle school during a class about word processors.
I know it's not standard any more and it may have been wrong when I was taught too! But honestly it's so ingrained in me at this point I'm not sure I can stop doing it.
That's alright if you use it for yourself. If you send your documents to other people, then it should. Double space is a minor thing, but I still pray for the day to finally come, when Word will open up the skies and strike down with thunder all those who move stuff around the page with multiple spaces and line breaks instead of proper formatting.
I think writing a text and putting in to paper or the screen are two separate matters and there's no need to intertwine both.
For example text alignment is something I don't see many people doing manually. Even when writing code people try to use text formatters so they don't have to do it themselves.
https://en.wikipedia.org/wiki/Separation_of_content_and_presentation
You can always turn it off. I don't see how enforcing best practice with sane defaults instead of an obsolete standard is hurtful.
https://practicaltypography.com/one-space-between-sentences.html
I wanna say that you're actually right. It's fine to feel like this is exactly not what you'd want your software to do, the whole point of there being multiple word processors is to allow you to choose, after all.
I use Word a lot, mostly to write essays for classes or documents for organizations. It being opinionated is useful for me there because those are exactly the kind of texts where you'd have an editor/supervisor come in and tell that, no, we don't use Oxford commas and this part is too lengthy and needs to be shortened.
But if you want a word processor for creative writing, quick note-taking, or really any kind of writing where you want to be in completely in charge: Word's probably not the best choice. It has a bunch of options you don't need and most options you might need are available elsewhere.
I love Microsoft’s response to Tom Warren’s article on The Verge about this.
In Portuguese two spaces after period is not even a thing, for anyone. So that is news to me.
Ah I see, because people hardwired putting two spaces from typewriter days. I thought It would be something like that, but had to check the comments here.
idk. I've never typed on a typewriter. got to say that QWERTY is the more painful legacy from the typewriter days than an extra space between sentences. This is similar to if Microsoft were coming at python programmers with the old spaces versus tabs argument, and insisting on one side as the right answer. It's cute they think they get to decide. Microsoft has a long legacy of being preachy bastards though, so not terribly surprising.
Subtext: I'm not salty at all that Microsoft autocorrects a single hyphen to be a dash character, which is available in utf-8 encoding but not available in ASCII, thereby potentially breaking some of the older code I have to work with on a regular basis if the user accidentally lets Microsoft decide their delimiter for them.
I have typed on a typewriter. The worst part of it is not the keyboard layout, but the way the keys are angled upward. Horrible for ergonomics. The flat keyboards that are predominant nowadays aren't great for ergonomics, but they're 10x better than that atrocity.
I found out about this change in writing style (standard?) And try to only single space, going as far as to do a search/replace for it. Writing a recent paper, I was surprised at it being flagged.