In the BNC World Edition each word or "multiword unit" (such as of course) is "tagged" according to its word-class or part of speech (PoS) with one of the codes reproduced below. "Fused" forms – contractions and possessives written with apostrophe as well as the form cannot and a few others – are "tokenized" as separate units and receive tags as well. PoS tags are assigned automatically by computer and thus are subject to ambiguity (almost 4%), errors (about 1%) and inconsistencies. Consequently some occurrences of the same word form representing the same word class may appear under different PoS codes.

Normalization conventions were adopted for this database to limit its overall size and to allow important patterns to emerge more clearly. For more detailed study of forms and tags which are not distinguished in this database, please consult the British National Corpus directly. In cases of "portemanteau" or ambiguous word tagging, the BNC shows two possible tags; here the more likely first one has been chosen. All capital letters are converted to lower case; proper nouns are recognized by the NP0 tag. Numerals are mapped onto a single "#" regardless of their magnitude or precision. Multiword_units identified by the parser are joined with an underscore into a single "word".  Refer to the normalization conventions for further details on how the database was compiled. These links lead to the BNC's lists of multiword units and fused forms. Geoffrey Leech and Nicholas Smith's Manual to accompany The British National Corpus (Version 2) with Improved Word-class Tagging describes the CLAWS parser and explains the PoS codes in greater detail.