Syntacticon: Syntax rule engine

SYNTAXRULE SIMPLE RULENAME: COMMAND / CONDITION // FLAGS COMPLEX TRY SIMPLE OR SYNTAXRULE RULENAME: optional (but required if a ":" appears later) Spaces are allowed. COMMAND may be any of the following. Tree modification: MoveFeature (NODES₁)(NODES₂) The features in the NODES₁ move to the NODES₂. Sources must be head-type nodes (no P, no ʹ). FLAGS may include 'atEnd'; otherwise, the features are placed at the start of the feature list, but after any roots. MoveToSpecifier (NODES)(NAME) FLAGS implicitly include "once". Move the first valid NODE to the specifier position of NAME, which must be an intermediate constituent (NAME must have ʹ) and it must not have a specifier. By default the NODE goes to the left of NAME, but if NAME has a right-hand sister called [Spec], then NODE is moved to the right-hand position replacing [Spec]. DIR-Adjoin (NODES)(NAME) DIR is either L or R. FLAGS implicitly include "once". AdjoinFeature(FEATURE)(NODES) Adds FEATURE as a label on the NODES. MergeRoot(ROOT)(NODES) Adds the root (adding √) at the start of the content of the NODES. NODES must be head-type nodes (no P, no ʹ). MergeFeature(FEATURE)(NODES) CopyFeature(MP,NODES₁)(MP,NODES₂) Copies the contents of the given MP collected from all nodes in NODES₁ to the MP in all the nodes in NODES₂, subject to any conditions. DeleteFeature(NODES, FeatureName) NODES must be head-type nodes (no P, no ʹ). FeatureName is either an MP (like Agr) or an MPS (like 3sg). DeleteNode(NODES) NODES must be phrases (P). Spellout Each head node containing MP:MPS invokes the RULEBLOCK for MP. FLAGS include "global" and NAME. By default, the input features to MP are all those in the head node. NAME collects all features in head nodes in the subtree rooted at NAME as input to the RULEBLOCK. "global" is equivalent to the NAME of the entire tree. Apply RULEBLOCK to NODES Usually invoked after Spellout to perform phonological transformations. Implicitly checks isHead. Result generation Flatten Preserve only the «seg» content of the head-type nodes. Debugging Expect PATTERN WHERE PATTERN is the expected current syntax pattern (without arrows) WHERE (optional) is the group number Crash (MSG) Skip all subsequent syntax rules except Spellout and Flatten. Check RELATIONS For debugging; results in "yes" or "no", but does not affect the current syntax tree. / CONDITION is optional A list of RELATIONs (no spaces allowed), each of which must hold. If a RELATION has "_" instead of a NAME, it refers to the source. If a RELATION has "‗" instead of a NAME, it refers to the destination for syntax rules that have both source and destination such as CopyFeature.. // FLAGS is optional echo ― Show the command in the "Errors and messages" section. once ― Only run the command once if NODES specifies several; stop after the first effective use. atEnd ― Move feature to the end of the existing features (for Move) global ― On Spellout, first collect the features in the whole tree as the MPS. NAME ― On Spellout, first collect all the features in the tree headed by NAME. Glossary NAME: The name of a node. HEAD: The name of a head-type node (no P, no ʹ) NODES: Either (1) a NAME, (2) a regexp matching NAMEs, such as /DP.*/, (3) an MPS specification, such as {3sg}{past,fut}, matching the content of head-type nodes, or (4) a relation with an underscore, such as isSister(_, NAME). FEATURE: of the form MP:MPS, where MPS is comma-separated, such as Agr:apart,3sg PATTERN: a balanced syntax-tree pattern, delimited by square brackets. RELATIONS: a comma-separated list of relations. isSister(NAME, NAME) contains(UPPER_NAME, LOWER_NAME) ccommands(UPPER_NAME, LOWER_NAME) isSpecifier(UPPER_NAME,LOWER_NAME) containsFeature(NAME, REGEXP) The REGEXP is applied to the contents of all head-type nodes under NAME. isOccupied(HEAD) True iff the HEAD contains a feature or a root. headPhrase(HEAD_NAME, PHRASE_NAME) isHead(HEAD_NAME)