specifies that the TAN file contains codes and rules for the parts of speech for a language, the codes for those parts, and the rules for combining them. Root element. contains a string that serves as an identifier for <feature>. The hyphen is reserved for "not applicable"; spaces and (),| reserved for @feature-test expressions [^\-\+\s]|[^\+\s]+ takes a regular expression. The condition is true only if an <m> matches the pattern. This is one of four tests for determining whether to process a <report> or <assert> takes a regular expression. When validating a given <m>, a test will be made against each companion <tok> (i.e., those <tok>s that have the same ancestral <ana>). The condition will be treated as true only for those matches with the resolved token value. This is one of four tests for determining whether to process a <report> or <assert> specifies one or more features. Values must be separated by spaces or the padded + (a plus sign with surrounding spaces). A plain space is treated as meaning "or" and the padded + as meaning "and." If there is at least one match between the list and the value of the codes in an <m> then the condition will be treated as true. The padded plus sign specifies that all the items need to be found in the <m>. For example, feature-test="A B + C D + E" means that the condition will be true for a given <m> only if that <m> has A or (B and C) or (D and E) present as values. This is one of four tests for determining whether to process a <report> or <assert> [^\s\+]+(\s(\+\s)?[^\s\+]+)* specifies a range or sequence of integers (e.g., 2-4). If the quantity of features in an <m> matches a number from that sequence, the condition is true, and false otherwise. If the keyword 'last' or 'max' is used, the total number of features will be substituted. This test is useful for non-category based grammars to put a limit on the number of features that can be declared in an <m>. This is one of four tests for determining whether to process a <report> or <assert> names a pattern that, if found to be false in any <m> in a dependent TAN-A-lm file, will return the enclosed message upon validation of the dependent file, along with an error or warning. Modeled on Schematron <report>. The sequence of consecutive <assert>s is immaterial. groups a set of features that share a common grammatical trait such as gender, number, etc. The sequence of consecutive <category>s is important, dictating the sequence in which codes should appear in the values of <m> in a dependent TAN-A-lm file. names a pattern that, if found to be true in any <m> in a dependent TAN-A-lm file, will return the enclosed message upon validation of the dependent file, along with an error or warning. Modeled on Schematron <report>. The sequence of consecutive <report>s is immaterial. encloses asserts and reports that should be evaluated provided that the conditions specified in the attributes (or children <where> attributes) are true. The sequence of consecutive <rule>s is immaterial. names, through its IRI + name pattern, a grammatical feature or concept (e.g., plural, subjunctive, 1st person) that is part of a language. In the context of <category> a @code is always included. The first <feature> of a <category>, however, never has a @code, because it describes the grammatical class of the <category>. The sequence of consecutive <feature>s is immaterial. See main.xml#vocabularies-features for a list of standard features.