February 27, 2007

The Theory of Date Formats

Arnold's discussion of the clever forensic use of date order by the Iranian government made me think a little about date formats. An analysis along the lines of Optimality Theory seems appropriate.

The constraints, in descending rank order, are:

Monotonicity
Are the units in a consistent increasing or descending order?
Adjacency
Are the month and day adjacent?
Endianness
Of the first two units, does the larger come first?

We obtain the following tableau, which seems to predict fairly nicely the attested distribution.

Pattern Attestation Monotonicity Adjacency Endianness
y-m-dEast Asia and ISO8601
d-m-yinternational
m-d-yUS
y-d-mUS Fish and Wildlife Service
m-y-dunattested?
d-y-munattested?

Empirical questions remain. The topic of date format not having been the focus of much research, one has to wonder whether some obscure tribe in the Amazon will turn out to write its dates in m-y-d or d-y-m order. Indeed, my colleague Geoff Pullum informs me that Pirahã dates are of the form "?-?-?". I'm not sure whether this means that they are unable to specify any of the three components, which would be consistent with the reports that their language has no numbers, or whether this is an ASCII version of IPA [ʔ ʔ ʔ], which doesn't seem any more informative.

It is also interesting to note that these properties appear to have analogues in word order:

  • Monotonicity ∼ Head-finality
  • Month-Day Adjacent ∼ Presence of Verb Phrase
  • Largest First ∼ Subject Precedes Object

[Addendum: as a reader has pointed out, "Endianness" here is not the same as in its use in computer science. I didn't intend it to be - it just seemed like a convenient term and I couldn't think of anything better. Proponents of OT will recognize that the analysis here is not true OT. It isn't intended to be. Another reader has reported that the US Fish and Wildlife Service uses dates in the y-d-m format for some purposes.]

Posted by Bill Poser at February 27, 2007 01:43 PM