The WWP indicates the language of words in a foreign language using the lang= attribute. This attribute is placed on whatever element is most appropriate to encode the words in question (e.g. <quote>). If the words are renditionally distinct solely by virtue of being in a foreign language, we encode them with the <foreign> element and encode the language with the lang= attribute.
The WWP provides a lang= attribute for any foreign word that is renditionally distinct. In addition, in a text where the foreign language is foregrounded as part of the content (in particular, travel narratives or works focusing on linguistics), all words which are clearly foreign will also be encoded (with <foreign> if there is no other applicable element) and given a lang= attribute.
In some cases, it may be difficult to determine whether a word or phrase is actually in a foreign language, since short phrases of foreign origin (such as “deja vu” or “et cetera”) may enter the English language gradually (and indeed different authors at any given point might disagree on whether they are fully naturalized). Where phrases like this occur in a renditionally distinct context (such as quotation or linguistic emphasis) it may be very difficult to tell whether the text is treating the phrase as foreign or not. In these cases, encoders will need to judge foreignness on the basis of usage and context.