Overview

I have recently been wondering about the use of abbreviations which are built from the first letter of a word followed by a number and the last letter. The number represents the count of letters between the start and end letter. Thus accessibility becomes a11y. This came to light (to me anyway) during an email exchange with Mike Ray regarding the accessibility issues on the tag index page on the HPR site. The website issues were resolved, but I was left wondering how useful the term a11y is, or whether it just jars with me!

According to the Wikipedia article this type of word is known as a numeronym, but they may also be referred to as alphanumeric acronyms, alphanumeric abbreviations, or numerical contractions.

As the Wikipedia article notes these types of abbreviations are almost always used to refer to their computing sense — such as g11n for globalisation — in the context of computing, not the general context.

Looking at a11y as an example

While I sympathise with the motivation behind using 'a11y' to mean accessibility, I do find it odd and counter-intuitive. I often find myself pondering the acceptability of this type of abbreviation. How many other words in common English fit patterns like this I wonder? Quite a few I would expect. How does this affect the admissibility of such abbreviations?

Not only are they adventurously strange to my simple brain, but I find them to be aesthetically displeasing. My experiments with the standard Linux dictionary looking for words that fit this pattern I find affirmatively supportive of this view. I describe this experiment later.

Algebraically, it is to be expected that there are many dictionary words of 13 characters which start with 'a' and end with 'y'. Looking at them allegorically, such numeronyms convey little meaning except in very limited contexts since the motivation seems to be to reduce the need to type long words. Alternatively, if they were accepted by data entry software and expanded automatically a better case could be made for applicability, but only one word could be assigned to a numeronym.

In my mind there is a certain artificiality in the use of these abbreviations.

You might wonder at the weird rambling nature of the above section - this was my (small) joke to try and use many of the words that match the a11y pattern.

Here’s the result of transforming them:

While I sympathise with the motivation behind 'a11y' to mean accessibility, I do find it odd and counter-intuitive. I often find myself pondering the a11y of this type of abbreviation. How many other words in common English fit these patterns I wonder? Quite a few I would expect. How does this affect the a11y of such abbreviations?

Not only are they a11y strange to my simple brain, but I find them to be a11y displeasing. My experiments with the standard Linux dictionary looking for words that fit this pattern I find a11y supportive of this view. I describe this experiment later.

A11y, it is to be expected that there are many dictionary words of 13 characters which start with 'a' and end with 'y'. Looking at them a11y, such numeronyms convey little meaning except in very limited contexts since the motivation seems to be to reduce the need to type long words. A11y, if they were accepted by data entry software and expanded a11y a better case could be made for a11y, but only one word could be assigned to a numeronym.

In my mind there is a certain a11y in the use of these abbreviations.

Make your own numeronyms

The following piece of Bash scripting scans the file /usr/share/dict/words and picks out words which match the a11y pattern (after removing those ending in 's). It writes the word and the numeronym generated from it, which it computes, though it’s unnecessary in this case because they all generate the same numeronym. I did it this way because I wanted to apply the algorithm to other words:

while read -r word; do
    printf '%-20s %s\n' "$word" "${word:0:1}$((${#word}-2))${word: -1}"
done < <(grep -E -v "'s$" /usr/share/dict/words | grep -E '^a.{11}y$')

Here’s a variant which selects all words which are 8-20 letters long, and picks 20 at random to which to apply the numeronym algorithm:

while read -r word; do
    printf '%-20s %s\n' "$word" "${word:0:1}$((${#word}-2))${word: -1}"
done < <(grep -E -v "'s$" /usr/share/dict/words | grep -E '^.{8,20}$' | shuf -n 20)

Here is a sample:

Aconcagua            A7a
semiweeklies         s10s
broadened            b7d
enlisting            e7g
nonpolitical         n10l
recessional          r9l
reorganizing         r10g
optimizations        o11s
taunting             t6g
subservience         s10e
dinosaur             d6r
hydroelectric        h11c
mellowing            m7g
perching             p6g
Winnebago            W7o
bunghole             b6e
mundanely            m7y
noisemaker           n8r
rattlings            r7s
microprocessors      m13s

Have fun with this - if you are so inclined!

Extremely long word (fake)

In researching for this episode I came upon an extremely long word, with information about it on Wikipedia. The word is:

Pneumonoultramicroscopicsilicovolcanoconiosis

Click to hear it spoken on Wikipedia

This is a made-up (possibly nonsensical) word, but I thought I could try my algorithm on it:

$ word="Pneumonoultramicroscopicsilicovolcanoconiosis"
$ printf '%-20s %s\n' "$word" "${word:0:1}$((${#word}-2))${word: -1}"
Pneumonoultramicroscopicsilicovolcanoconiosis P43s

Conclusion

Numeronyms don’t appeal to me. Notwithstanding my little jokes above, I know the proposal is not to replace all longer words with them; this would cause chaos! However, as a means of denoting long words this seems wrong.

I assume that their evolution occurs like this:

• We use a word often in a particular context
• The word is long and not easy to type
• For the sake of speed and to avoid typographic errors we make a numeronym
• We then tell the world that "i18n" (as an example) means internationalisation.
• Those in the know have no problems with it but many people who encounter it later puzzle over it - as I am doing here!

It seems fair to say that this obscure process has fulfilled the need to abbreviate this awkwardly long word - in the limits of the context where it has evolved. However it has not conveyed information very well; it has mainly benefited those who write (or read) documentation relating to the context.

Many editor and word processor applications have the facility of expanding abbreviations like this, in my experience. I would prefer to use this rather than embed the coded abbreviation into the language.

On the other hand, I’m OK with Pneumonoultramicroscopicsilicovolcanoconiosis being replaced by P43s!

I will confess that I had a similar reaction to XKCD’s “Up Goer Five” idea. He explains the Saturn 5 - “Explained using only the ten hundred words people use the most often”.

Maybe you disagree with me! If so, feel free to add a comment to this show — or indeed, record a show of your own!

Links

• Wikipedia article on Numeronyms

• XKCD:
  • Comic 1133: Up Goer Five
  • XKCD Explained - 1133: Up Goer Five
  • The Up-Goer Five text editor

• Fake words:
  • Lung disease Pneumonoultramicroscopicsilicovolcanoconiosis
  • Welsh village Llanfairpwllgwyngyllgogerychwyrndrobwllllantysiliogogogoch