The TELL search routine is written in Perl and inherits Perl's rich regular expression pattern matching. We have customized Perl's regular expressions for phonological searches, which means that we have added some features and broken others.
Metacharacters that correspond to phonological classes of segments, e.g. 'C' for consonants, 'V' for vowels. You may view the entire set of metacharacters.
Metacharacter that represents a syllable.
Metacharacter that represents a word boundary (Perl's word boundary metacharacter doesn't work properly on the TELL data set).
NB: Unlike Perl's metacharacters, TELL's custom metacharacters are not escaped with '\', so 'CVC' is a valid sequence of metacharacters meaning 'consonant-vowel-consonant sequence'.
NB: All custom metacharacters are upper-case letters or non-alphabetic characters. Note that '@' counts as an alphabetic character in the TELL dataset, as it is used in digraphs.
Perl metacharacters that contain a capital letter may or may not work, e.g. '\W', '\S', as upper-case letters are reserved for TELL's custom metacharacters. Assume that they don't work. Mostly they aren't useful in searching the TELL database anyway.
Perl's word boundary metacharacter '\b' doesn't work as you expect. This is because '@' is an alphabetic character in the TELL database but Perl interprets it as a non-alphabetic character. Use '<' instead.