Documentation of Matxin 1.0 - Revision history

Francis Tyers: Francis Tyers moved page Documentation of Matxin to Documentation of Matxin 1.0

2016-05-14T11:11:33Z

2009-12-01T15:09:48Z

Interchunk movement grammar

2009-11-28T21:45:33Z

Basque interchunk ordering grammars

2009-11-28T21:13:20Z

Basque interchunk ordering grammars

2009-11-28T21:11:52Z

Basque interchunk ordering grammars

2009-11-28T21:10:27Z

Basque interchunk ordering grammars

2009-11-28T20:13:02Z

Basque interchunk ordering grammars

2009-11-28T19:58:48Z

Basque interchunk ordering grammars

2009-11-28T19:17:37Z

Basque interchunk ordering grammars

2009-11-28T19:16:52Z

Basque interchunk ordering grammars

@@ Line 975: / Line 975: @@
 not '<code>[DAT]</code>' (dative), it will pass the information in the attribute '<code>mi</code>' to
 the chunk which is above it (direction '<code>up</code>') overwriting the contents of the
-attribute '<code>objMi</code>', proving that the chunk is of type '<code>adi-kat</code>'.
+attribute '<code>objMi</code>', but only if the chunk is of type '<code>adi-kat</code>'.
 ===Basque morphological preprocessing grammar===

@@ Line 1,105: / Line 1,105: @@
 ::::--x1-- => --[x2 x1]--
+* '''x1+x2'''  -- As above, only the child chunk (x2) is put immediately ''after'' the parent chunk (x1)
-(Note: '''x1+x2''' is not defined. Yet.)
+::: Note: only implemented in branches/matxin as of yet.
 * '''x2.x1.x2'''  -- The first chunks of the child chunk (x2) are put before the head, with the very last one being put right after the head (the order of the child chunk depends on whether earlier rules have reordered it or not)

← Older revision		Revision as of 21:13, 28 November 2009
Line 1,108:		Line 1,108:

	* '''x2.x1.x2''' -- The first chunks of the child chunk (x2) are put before the head, with the very last one being put right after the head (the order of the child chunk depends on whether earlier rules have reordered it or not)		* '''x2.x1.x2''' -- The first chunks of the child chunk (x2) are put before the head, with the very last one being put right after the head (the order of the child chunk depends on whether earlier rules have reordered it or not)
	::: eg. [vblex [pr [n]] where all three are chunks, ordered like "vblex n pr" before the rule applies, would give the order "n vblex pr"		::: eg. [vbser [vblex [adv] [n]] where all three are chunks, ordered like "vbser vblex adv n" before the rule applies, would give the order "vblex adv vbser n"

@@ Line 1,108: / Line 1,108: @@
 * '''x2.x1.x2'''  -- The first chunks of the child chunk (x2) are put before the head, with the very last one being put right after the head (the order of the child chunk depends on whether earlier rules have reordered it or not)
-::: eg. [vblex [pr [n]] where all three are chunks, ordered like "vblex n pr" before the rule applies, would give the order
+::: eg. [vblex [pr [n]] where all three are chunks, ordered like "vblex n pr" before the rule applies, would give the order "n vblex pr"
-::: n vblex pr

@@ Line 1,106: / Line 1,106: @@
 (Note: '''x1+x2''' is not defined. Yet.)
+* '''x2.x1.x2'''  -- The first chunks of the child chunk (x2) are put before the head, with the very last one being put right after the head (the order of the child chunk depends on whether earlier rules have reordered it or not)
+::: eg. [vblex [pr [n]] where all three are chunks, ordered like "vblex n pr" before the rule applies, would give the order
+::: n vblex pr

← Older revision	Revision as of 11:11, 14 May 2016
(No difference)

@@ Line 1,166: / Line 1,166: @@
 </pre>
+You can also say eg. =3, etc.
-(It seems <code><1, =1, >1</code> are the only possible values for <code>relativePosition</code>.)
 ==Program design==

@@ Line 1,151: / Line 1,151: @@
 </pre>
+Note: the first matching rule is applied, ie. above rules have priority.
 To sum up, these are the ordering relations:

@@ Line 1,115: / Line 1,115: @@
 then we can number them like this:
-    -2        -1           0          1        2
+    -2     -1       0      1       2
  child_A child_B parent child_C child_D

@@ Line 1,107: / Line 1,107: @@
 (Note: '''x1+x2''' is not defined. Yet.)
-relativePosition gives the position where the child chunk will be found with
+<code>relativePosition</code> gives the position where the child chunk will be found with
-respect to the parent chunk in the source language sentence.
+respect to the parent chunk in the source language sentence. If we have the following parse:
+ child_A child_B parent child_C child_D
+then we can number them like this:
+    -2        -1           0          1        2
+ child_A child_B parent child_C child_D
+so the <code>relativePosition</code> of child_C is 1, etc.
 ;Example:
@@ Line 1,140: / Line 1,150: @@
  (yo) (el pan)   (corto)    (en casa) (para tí)
 </pre>
+To sum up, these are the ordering relations:
+<pre>
+# if child is before head (<1), put child before the head:
+head	child 	<1	x2.x1
+# if child is immediately after head (=1), put child immediately before head:
+head	child 	=1	x2+x1
+# if child is after head, but not immediately after (>1), put child after head:
+head	child	>1	x1.x2
+# we should've exhausted the possibilities by now, but put child after head:
+head	child	.*?	x1.x2
+</pre>
+(It seems <code><1, =1, >1</code> are the only possible values for <code>relativePosition</code>.)
 ==Program design==