Density of information

« previous post | next post »

The first public sign I noticed after arriving at Hong Kong last week was this one embedded in the floor near the  conveyor belt (visible at the top of the photo):

Qǐng wù jiāng xínglǐ chē tuī rù huáng xiàn nèi 請勿將行李車推入黃線內
("please do not push baggage cart inside the yellow line")
11 characters — 92 strokes

No trolley beyond the yellow line
6 words — 42 strokes*

The next two public signs I encountered were similarly embedded in the floor at the exit from customs and just outside the door at the entry from the waiting room:

qǐng wù děnghòu 請勿等候
("please don't wait")
4 characters — 40 strokes

No waiting
2 words — 11 strokes

bù zhǔn jìnrù 不准進入
("entry not permitted")
4 characters — 28 strokes

No entry
2 words — 12 strokes

At the exit from the express train arrival area:

kuòzhájī 闊閘機
("wide turnstyles")
3 characters — 46 strokes

Wide gates
2 words — 13 strokes

At the queue for taxis:

qǐng zài cǐ děnghòu 請在此等候
("please wait here")
5 characters — 48 strokes

Please wait here
3 words — 20 strokes

*stroke totals for English words are as I write them; totals for others may differ slightly

We have previously seen that, in comparison with many other languages, equivalent English expressions tend to be shorter.

"French vs. English " (8/2/15)

Contrary to common opinion, by a variety of measurements English also often turns out to be shorter than Mandarin.  Density of information conveyed per stroke is a separate metric from density of spatial layout, and even when it comes to actual, customary usage (recording of equivalent linguistic utterances), English is frequently more economical than Mandarin in its density of spatial layout.

[Thanks to Zizi Li]


  1. maidhc said,

    December 2, 2015 @ 3:57 am

    What I guess you could call "headline English" is pretty concise.

  2. flow said,

    December 2, 2015 @ 7:17 am

    One should add that while the characters themselves do have a high density of strokes, what is being expressed by 請勿將行李車推入黃線內 is closer to your "please do not push baggage cart inside the yellow line" than to the (more appropriate in the context of an airport) equivalent "No trolley beyond the yellow line"; for the more literal English, I count 10 words with 72 strokes, still a tad shorter than the Chinese, but not that much. So the effect is due both to differences in the writing systems and to differences in the particular wordings.

    Some Chinese written expressions *can* be quite terse, and I wonder whether, as time goes by, phrases like 請勿將行李車推入黃線內 are going to be shortened, a la 禁煙.

  3. Tim Martin said,

    December 2, 2015 @ 9:28 am

    Similar to what flow said, I think the possibility of different wordings makes it difficult to compare which languages are shorter.

    I was thinking of a Japanese example. Lots of signs in Japanese advocate for a certain behavior using wording that directly translates to "Let's do [X]". This is a common way of softening a command in Japanese (e.g. "Let's keep the river clean" or 川をきれいにしましょう) . The English text on these signs will sometimes avoid that translation, going instead with a more typical command, "Keep the river clean." This is shorter, but which translation is the one we should use in our comparison? It's true that we don't often use "let's" in English signs to urge people to behave a certain way, but that knowledge is more "cultural translation" than "semantic translation," no?

  4. liuyao said,

    December 2, 2015 @ 10:06 am

    Interesting that 請勿 and 禁止, even 嚴禁, would all be shortened to No in free translation (only 3 strokes!). Some signs in US would go further: No trespassing, violators will be persecuted (which always seems a bit extreme until one sees it everywhere).

  5. Christopher South said,

    December 2, 2015 @ 11:50 am

    Maidhc referred to headline English, as in newspaper headlines. In newspaper headlines, a comma (,) is often used to replace "and." For example, "City workers, city agree on labor contract." In news releases we receive from outside sources, we also drop "Please" from directions, such as, "Please call XXX-XXXX for more information." We also omit "that" whenever possible. It's overused and often unnecessary. It's also incorrectly used where "who" is appropriate.

  6. JS said,

    December 2, 2015 @ 5:42 pm

    As always with "Written Chinese," there are few unambiguous indications of a particular spoken variety here. Maybe fut3/kuo4 闊 as 'wide' is the one case where a Mandarin speaker would sense something 'off' (as this seems to draw transparently on Cantonese)? Yep6/ru4 入 as a complement is more common in colloquial Cantonese I think, but doesn't faze a Mandarin speaker in the slightest due to its ubiquity in Written… there must be more…

  7. Ross Presser said,

    December 3, 2015 @ 1:31 am

    Is stroke count really the right way to measure density, though? These signs are not being written by hand; they're either machine printed, or sprayed using a stencil. And they aren't being *read* stroke by stroke either. The English words are, for fluent readers, recognized as a single gestalt — this pattern of marks means one word, seen all at once. Likewise, I would think, for the Chinese writing, although I don't read it myself — the character is recognized at one glimpse. So I *would* measure information density by spatial extent, not by stroke count.

  8. flow said,

    December 3, 2015 @ 11:18 am

    @Ross In my experience stroke count is indeed a rough measure for visual density, which somewhat correlates with ease of recognition (although the relationship is certainly not 1:1; for example, there are character pairs like 予子 and 于干 which are not always easy to tell apart). Of course, when doing a comparison across writing systems, it would be a good idea to apply a single standard to all glyphs—it is not immediately clear how many strokes to count for 'N' (1? 3?) or `e` (2? 1?), and whether the bent strokes of Chinese should count as single or multiple elementary strokes (e.g. 了 has been counted as two strokes in its printed (Song) form from at least the Kangxi dictionary onwards, although it should probably be counted as a single stroke in its Regular Script (Kaishu) / handwritten form; complexity-wise, it would make sense to count it as three elementary strokes).

    What I really like about the picture is that, for once, a pleasing typographic balance between the Chinese and the English parts of the message has been found; indeed, the larger size of the Chinese characters is fully appropriate given their visual complexity. Only the appearance of the top of the character 入 (and the 內, which I find is a hypercorrected version of 内) is a bit awkward; it looks like a little knob that was added as an afterthought. The way it appears here, it almost looks like 人.

  9. BZ said,

    December 3, 2015 @ 1:30 pm

    What I noticed in this sign was "trolley" for baggage cart. In fact, I believe I've seen this usage at airports in Europe and Israel as well. Yet, never in the US. Is this a US/UK English difference?

  10. Eidolon said,

    December 3, 2015 @ 1:54 pm

    Speaking of density, it might be useful to examine the stroke counts of the pinyin, as well. I only have time for one:

    Qǐng wù jiāng xínglǐ chē tuī rù huáng xiàn nèi

    8 7 11 11 5 6 5 11 10 6

    = 80 total strokes, still less than the characters, but way above the English

  11. Victor Mair said,

    December 5, 2015 @ 9:44 am

    Thanks for counting those strokes of the pinyin, Eidolon. The result is revealing.

  12. Victor Mair said,

    December 5, 2015 @ 10:34 am

    I'm very grateful to flow's clear and precise response to Ross Presser.

RSS feed for comments on this post