Note that this directory includes samples of languages for which there
is no language model in the default database.  These will often pick a
closely-related language, but sometimes will appear to be selecting
languages at random because there are only a few n-gram matches with
varying languages; although the highest-scoring languages are picked,
their scores are in fact fairly low.

In addition, many of the Wikipedia-derived files contain some text in
other languages, particularly when the Creative Commons license notice
is included (this text is often in a major language related to the
minor language of the localized Wikipedia, e.g. Italian for one of
Italy's regional languages).

Files containing the substring .PNGscriptures. are extracts from Bible
translations distributed by http://pngscriptures.org/, and are
licensed under the Creative Commons Attribution-Noncommercial-
NoDerivatives v3.0 license (see http://pngscriptures.org/terms.htm for
details).
