Difference between revisions of "Empty categories"

From Icelandic Parsed Historical Corpus (IcePaHC)
Jump to: navigation, search
(Empty subjects)
 
(33 intermediate revisions by 3 users not shown)
Line 2: Line 2:
  
 
*[[Traces]]
 
*[[Traces]]
 +
[[Extraction out of NP]]
 +
 +
[http://linguist.is/icelandic_treebank/ADJP#Extraction_out_of_an_empty_ADJP Extraction out of empty ADJP]
 +
 
*[[Pied piping]]
 
*[[Pied piping]]
 
*[[Missing copula]]
 
*[[Missing copula]]
*[[Expletives]] (*exp*): Note that (NP-SBJ *exp*) is not used in the Icelandic corpus for the experiencers of known quirky-case-verbs; see the documentation on [[Quirky Case]]. [EFS: athuga hvað maður gerir við (þó pro/exp hásumar væri) en (það rigndi í gær) er bara með venjulegt NP-SBJ]
+
 
 +
 
 +
==Empty subjects==
 +
 
 +
===Subjects elided under conjunction (Conjunction Reduction) ===
 +
[[*con*]]: subjects elided under conjunction, by default.
 +
 
 +
*con*-D, *con*-A: oblique subjects elided under conjunction
 +
 
 +
*con*-N: nominative subjects elided under conjunction which are controlled by a previous oblique subject
 +
 
 +
===Expletives===
 +
[[Expletives]] (*exp*): Note that (NP-SBJ *exp*) is not used in the Icelandic corpus for the experiencers of known quirky-case-verbs; see the documentation on [[Quirky Case]].   Note that in early (Old Icelandic) texts, it may be difficult to distinguish between *exp* and *pro* in cases where there is an empty subject and it has a roughly impersonal (i.e. "maður") interpretation.  In searches where you want to include all such sentences or all impersonals in early texts, it is wise to include both *exp* and *pro* in your searches.
 +
 
 +
===Arbitrary PRO===
 +
[[*arb*]]: arbitrary PRO in the [[IP-INF]] complements of inherent ECM verbs.
 +
 
 +
===pro subject===
 +
[[*pro*]]: other empty subjects.  Note that in early (Old Icelandic) texts, it may be difficult to distinguish between *exp* and *pro* in cases where there is an empty subject and it has a roughly impersonal (i.e. "maður") interpretation.  In searches where you want to include all such sentences or all impersonals in early texts, it is wise to include both *exp* and *pro* in your searches.
 +
 
 +
Sometimes, it's ambiguous whether a quantifier is the subject of the clause or if it modifies a [[*pro*]] subject. Then the context may indicate the right parse.
 +
 
 +
<pre>
 +
( (IP-MAT (VBDI Sátu)
 +
  (ADVP-LOC (ADV þar))
 +
  (NP-SBJ (Q-N allar))
 +
  (PP (P um)
 +
      (ADJP (ADJ-A kyrt)))
 +
  (. ,-,)))
 +
</pre>
 +
 
 +
Topicalization or stylistic fronting of quantifiers (or other elements, e.g. D, N, P): indicated with *ICH* traces according to [http://www.ling.upenn.edu/~beatrice/annotation/syn-empty.htm#stylistic_fronting_of_quantifiers Penn Guidelines].
 +
 
 +
Empty objects of verbs: not indicated in annotation.
 +
 
 +
Empty objects of prepositions: in this case we follow the [http://www-users.york.ac.uk/~lang22/YCOE/doc/annotation/YcoeRef.htm#prep_empty_object guidelines of the York Corpus of Old English Prose (YCOE)].  See also Particles under [[Head Types]].
 +
 
 +
EINS OG ER
 +
 
 +
===Empty oblique subject===
 +
 
 +
A case extension is added to empty subjects if and only if the empty element is non-nominative. This can most frequently occur in cases of conjunction reduction, like *con*-D (empty dative subject), but also in other cases where the subject has been omitted, like *pro*-D.
 +
 
 +
<pre>
 +
( (IP-MAT (CONJ og-og)
 +
  (NP-SBJ *con*-D)
 +
  (VBDI þótti-þykja)
 +
  (NP-SBJ (N-N skömm-skömm))
 +
  (IP-INF (TO að-að)
 +
  (VB skilja-skilja)
 +
  (PP (P við-við)
 +
      (NP (PRO-A hann-hann))))
 +
  (. .-.)))
 +
</pre>
 +
 
 +
We also use *arb*-D for any instance of quirky PRO (whether or not it is truly "arbitrary" PRO or controlled PRO).
 +
 
 +
===Parasitic Gaps===
 +
 
 +
Parasitic gaps are marked with the same trace (same index) as the real gap on which they are parasitic, as in the example below: 
 +
 
 +
<pre>
 +
  (CP-QUE-SPE (IP-SUB-SPE (BEPI Er-vera)
 +
  (NP-SBJ (PRO-N það-það))
 +
  (NEG ekki-ekki)
 +
  (NP-PRD (D-N sá-sá)
 +
  (CP-REL-SPE (WNP-2 0)
 +
      (C sem-sem)
 +
      (IP-SUB-SPE (NP-OB1 *T*-2)
 +
  (NP-SBJ (PRO-N þeir-hann))
 +
  (VBDI sóktu-sækja)
 +
  (PP (P til-til)
 +
      (IP-INF-PRP-SPE (NP-OB1 *T*-2)
 +
      (TO að-að)
 +
      (VB lífláta-lífláta)
 +
      (CODE {COM:parasitic_gap}))))))))
 +
  (. ?-?))
 +
  (ID 1540.NTJOHN.REL-BIB,200.545))
 +
</pre>
 +
 
 +
We frequently also put a comment in the token indicating that the gap is parasitic.
 +
 
 +
==A generic empty category (X *)==
 +
 
 +
Traces of A-movement are marked with an asterisk " * ", including all pseudo-passives.
 +
 
 +
 
 +
===BE *===
 +
We do not use (BE *) after a modal verb and before a participle (VAN or VBN):
 +
 
 +
<pre>
 +
  (IP-MAT-SPE (NP-SBJ (Q-N Lítil-lítill) (N-N tilraun-tilraun))
 +
      (MDPI mun-munu)
 +
      (VAN birt-birta)
 +
      (PP (P til-til)
 +
  (NP (PRO-G þess-það))))
 +
</pre>

Latest revision as of 08:56, 3 September 2019

Empty categories

Extraction out of NP

Extraction out of empty ADJP


Empty subjects

Subjects elided under conjunction (Conjunction Reduction)

*con*: subjects elided under conjunction, by default.

  • con*-D, *con*-A: oblique subjects elided under conjunction
  • con*-N: nominative subjects elided under conjunction which are controlled by a previous oblique subject

Expletives

Expletives (*exp*): Note that (NP-SBJ *exp*) is not used in the Icelandic corpus for the experiencers of known quirky-case-verbs; see the documentation on Quirky Case. Note that in early (Old Icelandic) texts, it may be difficult to distinguish between *exp* and *pro* in cases where there is an empty subject and it has a roughly impersonal (i.e. "maður") interpretation. In searches where you want to include all such sentences or all impersonals in early texts, it is wise to include both *exp* and *pro* in your searches.

Arbitrary PRO

*arb*: arbitrary PRO in the IP-INF complements of inherent ECM verbs.

pro subject

*pro*: other empty subjects. Note that in early (Old Icelandic) texts, it may be difficult to distinguish between *exp* and *pro* in cases where there is an empty subject and it has a roughly impersonal (i.e. "maður") interpretation. In searches where you want to include all such sentences or all impersonals in early texts, it is wise to include both *exp* and *pro* in your searches.

Sometimes, it's ambiguous whether a quantifier is the subject of the clause or if it modifies a *pro* subject. Then the context may indicate the right parse.

( (IP-MAT (VBDI Sátu)
	  (ADVP-LOC (ADV þar))
	  (NP-SBJ (Q-N allar))
	  (PP (P um)
	      (ADJP (ADJ-A kyrt)))
	  (. ,-,)))

Topicalization or stylistic fronting of quantifiers (or other elements, e.g. D, N, P): indicated with *ICH* traces according to Penn Guidelines.

Empty objects of verbs: not indicated in annotation.

Empty objects of prepositions: in this case we follow the guidelines of the York Corpus of Old English Prose (YCOE). See also Particles under Head Types.

EINS OG ER

Empty oblique subject

A case extension is added to empty subjects if and only if the empty element is non-nominative. This can most frequently occur in cases of conjunction reduction, like *con*-D (empty dative subject), but also in other cases where the subject has been omitted, like *pro*-D.

( (IP-MAT (CONJ og-og)
	  (NP-SBJ *con*-D)
	  (VBDI þótti-þykja)
	  (NP-SBJ (N-N skömm-skömm))
	  (IP-INF (TO að-að)
		  (VB skilja-skilja)
		  (PP (P við-við)
		      (NP (PRO-A hann-hann))))
	  (. .-.)))

We also use *arb*-D for any instance of quirky PRO (whether or not it is truly "arbitrary" PRO or controlled PRO).

Parasitic Gaps

Parasitic gaps are marked with the same trace (same index) as the real gap on which they are parasitic, as in the example below:

	  (CP-QUE-SPE (IP-SUB-SPE (BEPI Er-vera)
				  (NP-SBJ (PRO-N það-það))
				  (NEG ekki-ekki)
				  (NP-PRD (D-N sá-sá)
					  (CP-REL-SPE (WNP-2 0)
						      (C sem-sem)
						      (IP-SUB-SPE (NP-OB1 *T*-2)
								  (NP-SBJ (PRO-N þeir-hann))
								  (VBDI sóktu-sækja)
								  (PP (P til-til)
								      (IP-INF-PRP-SPE (NP-OB1 *T*-2)
										      (TO að-að)
										      (VB lífláta-lífláta)
										      (CODE {COM:parasitic_gap}))))))))
	  (. ?-?))
  (ID 1540.NTJOHN.REL-BIB,200.545))

We frequently also put a comment in the token indicating that the gap is parasitic.

A generic empty category (X *)

Traces of A-movement are marked with an asterisk " * ", including all pseudo-passives.


BE *

We do not use (BE *) after a modal verb and before a participle (VAN or VBN):

	  (IP-MAT-SPE (NP-SBJ (Q-N Lítil-lítill) (N-N tilraun-tilraun))
		      (MDPI mun-munu)
		      (VAN birt-birta)
		      (PP (P til-til)
			  (NP (PRO-G þess-það))))