Making use of Relations CG-3 can also work with generic relations. These are analogous to dependency relations, but can have any name, overlap, are directional, and can point to multiple cohorts.
ADDRELATION, ADDRELATIONS ADDRELATION ADDRELATIONS [wordform] ADDRELATION <name> <target> [contextual_tests] TO|FROM <contextual_target> [contextual_tests] ; [wordform] ADDRELATIONS <name> <name> <target> [contextual_tests] TO|FROM <contextual_target> [contextual_tests] ; ADDRELATION creates a one-way named relation from the current cohort to the found cohort. The name must be an alphanumeric string with no whitespace. ADDRELATION (name) targetset (-1* ("someword"))   TO (1* (@candidate)) (2 SomeSet) ; ADDRELATIONS creates two one-way named relation; one from the current cohort to the found cohort, and one the other way. The names can be the same if so desired. ADDRELATIONS (name) (name) targetset (-1* ("someword"))   TO (1* (@candidate)) (2 SomeSet) ;
SETRELATION, SETRELATIONS SETRELATION SETRELATIONS [wordform] SETRELATION <name> <target> [contextual_tests] TO|FROM <contextual_target> [contextual_tests] ; [wordform] SETRELATIONS <name> <name> <target> [contextual_tests] TO|FROM <contextual_target> [contextual_tests] ; SETRELATION removes all previous relations with the name, then creates a one-way named relation from the current cohort to the found cohort. The name must be an alphanumeric string with no whitespace. SETRELATION (name) targetset (-1* ("someword"))   TO (1* (@candidate)) (2 SomeSet) ; SETRELATIONS removes all previous relations in the respective cohorts with the respective names, then creates two one-way named relation; one from the current cohort to the found cohort, and one the other way. The names can be the same if so desired. SETRELATIONS (name) (name) targetset (-1* ("someword"))   TO (1* (@candidate)) (2 SomeSet) ;
REMRELATION, REMRELATIONS REMRELATION REMRELATIONS [wordform] REMRELATION <name> <target> [contextual_tests] TO|FROM <contextual_target> [contextual_tests] ; [wordform] REMRELATIONS <name> <name> <target> [contextual_tests] TO|FROM <contextual_target> [contextual_tests] ; REMRELATION destroys one direction of a relation previously created with either ADDRELATION or SETRELATION. REMRELATION (name) targetset (-1* ("someword"))   TO (1* (@candidate)) (2 SomeSet) ; REMRELATIONS destroys both directions of a relation previously created with either ADDRELATION or SETRELATION. REMRELATIONS (name) (name) targetset (-1* ("someword"))   TO (1* (@candidate)) (2 SomeSet) ;
Existing Relations in Input Relational attachments are in the forms of ID:id and R:name:id tags. The ID tags are assumed to give the cohort a globally unique numeric ID, and this number is what the id from R tags refer to. The name part must be alphanumeric and must not start with a number. It is normal for ID tags to exist without R tags for one-way relations, but any line with an R tag must have its ID along. For example: "<There>"   "there" <*> ADV @F-SUBJ ID:1056 "<once>"   "once" ADV @ADVL "<was>"   "be" <SVC/N> <SVC/A> V PAST SG1/3 VFIN IMP @FMV "<a>"   "a" <Indef> ART DET CENTRAL SG @>N "<man>"   "man" N NOM SG @SC ID:1060 R:Beginning:1056 "<$.>"
Contextual Tests Once you have relations to work with, you can use that in subsequent contextual tests as seen below. These positions can be combined with the window spanning options.
Specific Relation r:rel The 'r:rel' position asks for cohorts found via the 'rel' relation. 'rel' can be any name previously given via ADDRELATION or SETRELATION. Be aware that for combining positional options, 'r:rel' should be the last in the position; 'r:' will eat anything following it until it meets a space. (r:rel (ADJ))
Any Relation r:* The 'r:*' position asks for cohorts found via any relation. (r:* (ADJ))
Self S The 'S' option allows the test to look at the current target as well. Used in conjunction with p, c, cc, s, or r to test self and the relations. (Sr:rel (ADJ))
Left/right of, Left/rightmost l r ll rr lll rrr llr rrl The r, l, rr, ll, rrr, lll, rrl, llr options documented at Dependencies also work for Relations. (rrrr:rel (ADJ)) (lr:rel (ADJ))
All Scan ALL The 'ALL' option will require that all of the relations match the set. For example, position 'ALL r:rel' requires that all of the 'rel' relations match the set. (ALL r:rel (ADJ))
None Scan NONE The 'NONE' option will require that none of the relations match the set. For example, position 'NONE r:rel' requires that none of the 'rel' relations match the set. (NONE r:rel (ADJ))