;;; -*- Mode: tdl; Coding: utf-8; -*-

;;;
;;; Copyright (c) 2009 -- 2011 Stephan Oepen (oe@ifi.uio.no); 
;;; see `LICENSE' for conditions.
;;;

;;
;; now, with +CLASS information available, optionally make any token a proper
;; NE that is (a) capitalized and not initial, (b) spelled in mixed case
;; (|LinGO|), or (c) initial all-caps (a sub-set of capitalized).  note that
;; we want these rules to also fire on single-character tokens, as e.g. in
;; |the A Team| or |the I and J columns| (dan may tame the |I| proper name by
;; adding a non-nominative native lexical entry for it).
;;
;; _fix_me_
;; the ERG lexicon includes a few entries (e.g. titles like |Mr.| and |Jr.|)
;; with capitalized orthography.  currently capitalized NEs are about the only
;; class of generics that can survive alongside a native entry (in the lexical
;; filtering phase), hence it might make sense to prune unwanted tokens here,
;; even though that means knowledge about the ERG lexicon is applied at token
;; mapping time already?                                       (23-jan-09; oe)
;;
;; come to think of it, i suspect the |I| special case mentioned above would
;; fall into this class too?                                    (6-aug-11; oe)
;;
capitalized_name_tmr := add_ne_tmt &
[ +CONTEXT < [ +CLASS alphanumeric & [ +INITIAL -, +CASE capitalized ] ] >,
  +OUTPUT < [ +CLASS proper_ne ] > ].   

mixed_name_tmr := add_ne_tmt &
[ +CONTEXT < [ +CLASS alphanumeric & [ +CASE mixed ] ] >,
  +OUTPUT < [ +CLASS proper_ne ] > ].   

upper_name_tmr := add_ne_tmt &
[ +CONTEXT < [ +CLASS alphanumeric & 
                      [ +INITIAL +, +CASE capitalized+upper ] ] >,
  +OUTPUT < [ +CLASS proper_ne ] > ].