Plurals

Short Example

The english language is very simple when it comes to plurals:

You have 0 elephants, 1 elephant, or 2+ elephants.

So basically you have 2 different forms: one singular and one plural.

But for some other languages this is quite a bit more difficult. Let’s take the Bosnian language as another example:

You have [1/21/31] slon, [2/3/4] slona, [0/5/6] slonova and [7/8/9/11] …

The problem with the old system of plurals was that you would have needed to specify them all, and get a loop in there.

As we are not the first developers facing this problem, it was not really hard to find a suitable solution. We decided to use the system from Unicode.org, which is e.g. used by Mozilla.

Plural Rules

So we defined the following 16 rules for plurals. First point is the language family, afterwards there are a number of rows with the following format: <key> - <rule>: <example-numbers>

Note

0 is handled as a special case. If you add a key 0 to your array, that will be used in case of 0 independent of the plural rule.

Rule #0

  • Families: Asian (Chinese, Japanese, Korean, Vietnamese), Persian, Turkic/Altaic (Turkish), Thai, Lao

  • 1 - everything: 0, 1, 2, …

Rule #1

  • Families: Germanic (Danish, Dutch, English, Faroese, Frisian, German, Norwegian, Swedish), Finno-Ugric (Estonian, Finnish, Hungarian), Language isolate (Basque), Latin/Greek (Greek), Semitic (Hebrew), Romanic (Italian, Portuguese, Spanish, Catalan)

  • 1 - 1

  • 2 - everything else: 0, 2, 3, …

Rule #2

  • Families: Romanic (French, Portuguese, Brazilian Portuguese)

  • 1 - 0, 1

  • 2 - everything else: 2, 3, …

Rule #3

  • Families: Baltic (Latvian)

  • 1 - 0

  • 2 - ends in 1, excluding 11: 1, 21, … 101, 121, …

  • 3 - everything else: 2, 3, … 10, 11, 12, … 20, 22, …

Rule #4

  • Families: Celtic (Scottish Gaelic)

  • 1 - is 1 or 11: 1, 11

  • 2 - is 2 or 12: 2, 12

  • 3 - others between 3 and 19: 3, 4, … 10, 13, … 18, 19

  • 4 - everything else: 0, 20, 21, …

Rule #5

  • Families: Romanic (Romanian)

  • 1 - 1

  • 2 - is 0 or ends in 01-19, excluding 1: 0, 2, 3, … 19, 101, 102, … 119, 201, …

  • 3 - everything else: 20, 21, …

Rule #6

  • Families: Baltic (Lithuanian)

  • 1 - ends in 1, excluding 11: 1, 21, 31, … 101, 121, …

  • 2 - ends in 0 or ends in 10-20: 0, 10, 11, 12, … 19, 20, 30, 40, …

  • 3 - everything else: 2, 3, … 8, 9, 22, 23, … 29, 32, 33, …

Rule #7

  • Families: Slavic (Bosnian, Croatian, Serbian, Russian, Ukrainian)

  • 1 - ends in 1, excluding 11: 1, 21, 31, … 101, 121, …

  • 2 - ends in 2-4, excluding 12-14: 2, 3, 4, 22, 23, 24, 32, …

  • 3 - everything else: 0, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 25, 26, …

Rule #8

  • Families: Slavic (Slovak, Czech)

  • 1 - 1

  • 2 - 2, 3, 4

  • 3 - everything else: 0, 5, 6, 7, …

Rule #9

  • Families: Slavic (Polish)

  • 1 - 1

  • 2 - ends in 2-4, excluding 12-14: 2, 3, 4, 22, 23, 24, 32, … 104, 122, …

  • 3 - everything else: 0, 5, 6, … 11, 12, 13, 14, 15, … 20, 21, 25, …

Rule #10

  • Families: Slavic (Slovenian, Sorbian)

  • 1 - ends in 01: 1, 101, 201, …

  • 2 - ends in 02: 2, 102, 202, …

  • 3 - ends in 03-04: 3, 4, 103, 104, 203, 204, …

  • 4 - everything else: 0, 5, 6, 7, 8, 9, 10, 11, …

Rule #11

  • Families: Celtic (Irish Gaeilge)

  • 1 - 1

  • 2 - 2

  • 3 - is 3-6: 3, 4, 5, 6

  • 4 - is 7-10: 7, 8, 9, 10

  • 5 - everything else: 0, 11, 12, …

Rule #12

  • Families: Semitic (Arabic)

  • 1 - 1

  • 2 - 2

  • 3 - ends in 03-10: 3, 4, … 10, 103, 104, … 110, 203, 204, …

  • 4 - ends in 11-99: 11, … 99, 111, 112, …

  • 5 - everything else: 100, 101, 102, 200, 201, 202, …

  • 6 - 0

Rule #13

  • Families: Semitic (Maltese)

  • 1 - 1

  • 2 - ends in 01-10: 0, 2, 3, … 9, 10, 101, 102, …

  • 3 - ends in 11-19: 11, 12, … 18, 19, 111, 112, …

  • 4 - everything else: 20, 21, …

Rule #14

  • Families: Slavic (Macedonian)

  • 1 - ends in 1: 1, 11, 21, …

  • 2 - ends in 2: 2, 12, 22, …

  • 3 - everything else: 0, 3, 4, … 10, 13, 14, … 20, 23, …

Rule #15

  • Families: Icelandic

  • 1 - ends in 1, excluding 11: 1, 21, 31, … 101, 121, 131, …

  • 2 - everything else: 0, 2, 3, … 10, 11, 12, … 20, 22, …

How to use the rules

The first thing your language package needs, is a definition, which rule to use for your package. This is done in the language/xy/common.php language file at the beginning of the array, (Rule #1 is the rule for the English language and will be used by default, if you don’t specify one):

'PLURAL_RULE' => 1,

The following example is using rule #13:

It has the following rows:

  • 1 - 1

  • 2 - ends in 01-10: 2, 3, … 9, 10, 101, 102, …

  • 3 - ends in 11-19: 11, 12, … 18, 19, 111, 112, …

  • 4 - everything else: 20, 21, …

While the English language only has 2 rows in its array:

'EXAMPLE' => [
    1 => '1 example',
    2 => '2 or more examples',
],

You need to specify the zero-row and 4 rows for the “plurals”:

'EXAMPLE' => [
    1 => '1 example',
    2 => '[0 or number ending with 01-10] examples',
    3 => '[number ending with 11-19] example',
    4 => 'even more examples',
],

If you require separate handling for 0, you can simple add the 0-case:

'EXAMPLE' => [
    0 => 'No example',
    1 => '1 example',
    2 => '[zero is not handled here anymore! Only number ending with 01-10] examples',
    3 => '[number ending with 11-19] example',
    4 => 'even more examples',
],

If you forget a line the system will automatically use the row before. So if you forget to add the 3-row, it will use 2-row for 11-19 as well. If there is no previous row, it uses the last row of the array.

Ensure your cases are in ascending order, otherwise the system may produce unexpected results if any keys are missing or out of order.

Credits

The system is based on Mozilla, which uses the “Plural Rules and Families” from GNU gettext documentation.