IP and other clause types

The IP is the basic clause, containing the verbal complex, its arguments, and adjuncts. Because there is no VP in our system, many sub-clause constituents containing a verb and argument(s) must be labelled IP (e.g. in coordinated clauses). Participial and fragmentary clauses are also treated below. (Subordinate and interrogative clauses are embedded in a CP and are treated on the page for CP types).

(This annotation guide is not exhuastive, but is intended to help distinguish the most frequent clause types and highlight differences between our annotations and those of other corpora. For detailed descriptions, see the manuals for PPCHE and HeliPaD.)

Notes for all types of clauses:

-SPE: attached to all IPs and CPs that are part of direct speech.
Every CP must contain an IP-SUB and every IP-SUB must be housed in a CP.
IPX and CPX: coordinated clauses with elided material. Not to be confused with IP-X below.

IPs

IP-MAT

Finite IPs (IP-MAT & IP-SUB) other than imperatives must contain a subject (even if null) and a finite verb (unless elided, as is possible with the perfect and passive auxiliaries haben, sein, andwerden in ENHG (Breitbarth 2004), especially in coordinated clauses.
Imperatives are labelled IP-MAT following HeliPaD (not IP-IMP). They need not (but may) have a subject.
In most cases, each IP-MAT should be the root node of a sentence token. (For now, we indicate what will later be split into new sentence tokens as IP-MAT-CONJ.) Coordinated IP-MATs are only allowed in rare cases (in which one conjunct clause has elision).
As in HeliPaD, direct speech after verb of saying begins a new sentence token.
IP-MAT-PRN indicates a parenthetical main clause, usually contained within another clause, and either elaborates on the matrix clause or is an inquit:

( (IP-MAT ...
          (ADVP-RSP (ADV so=so))
          (MDPI^3^SG muosß=müssen)
          (ADVP (ADV zuovor=zuvor))
          (NP-SBJ (NPR Adam=adam))
          (RP vnder=unter)
          (VB gon=untergehen)
          (, .=_)
          (IP-MAT-PRN (NP-SBJ (D^N^SG dz=der))
                      (BEPI^3^SG ist=sein)
                      (, /=/)
                      (CP-THT (C 0) 
                              (IP-SUB (CP-FRL-SBJ (WNP-PRD-1 (WPRO^N^SG was=was))
                                                 (C 0)
                                                 (IP-SUB (NP-PRD *T*-1)
                                                         (NP-SBJ (PRO^N^PL wir=wir))
                                                         (BEPI^1^PL (BEPI^1^PL seind=sein) ...)))
                                      (MDPI^3^SG muosß=müssen)
                                      (, /=/)
                                      (NP-SBJ-PRN (Q^N^SG alles=all))
                                      (PP (P zuo=zu)
                                          (NP (N^D^SG grund=grund)))
                                      (VB gon=gehen))))
                      (, /=/)
   ...
  (ID 1523_Butzer_Rel-sermon_Alsace.,199))

( (IP-MAT-SPE (NP-SBJ (D^N^SG der)
		      (ADJ^N^SG khuenest)
		      (NP-POS (PRO$^G^SG deines) (N^G^SG Haeres)))
	      (IP-MAT-PRN-SPE (VBPI^1^SG denck)    <- parenthetical clause 
			      (NP-SBJ (PRO^N^SG ich)))
              …
	      (MDPI^3^SG wil)
	      (ADVP (ADV auch))
	      (NEG nit)
	      (ADVP (ADV widerum))
	      (RP heym)
	      (VB ziehen)
	      …
  (ID 1533_Rodler_Fierrabras_Moselfrk.,16))

( (IP-MAT-SPE (PUNCQL ")
	      (CONJ dan)
	      (NP-SBJ (PRO^N^SG ich))
	      (HVPI^1^SG hab)
	      (ADVP (ADV wol))
	      (, /)
	      (PUNCQR ")
	      (IP-MAT-PRN (VBDI^3^SG sprach)    <- parenthetical inquit
			  (NP-SBJ (D^N^SG der) (N^N^SG Keyser))
			  (ADVP (ADV weitter)))
	      (, /)
	      (PUNCQL ")
	      (PP (P auß)
		  (NP (PRO$^D^PL seinen) (N^D^PL reden)))
	      (VBN vernomen)
	      …
  (ID 1533_Rodler_Fierrabras_Moselfrk.,19))

IP-MAT-PRN is usually attached at the clausal level (as in the two examples above), but may be embedded within a different phrase type, e.g. when the parenthetical clause elaborates on an NP:

... (NP-PRN (NPR^N^SG Cupido)
	    (IP-MAT-PRN (NP-SBJ (D^N^SG das))
		        (BEDI^3^SG was)
		        (NP-PRD (NP-COM (D^G^SG der) (N^G^SG lieb))
			        (N^N^SG schücz)))) 
     ...
  (ID 1480_Troyana_Lit-prose_Austria.,64))

IP-SUB

IP-SUB is used in questions (CP-QUE; direct or indirect) and subordinate clauses (all other CP types). Every IP-SUB must be inside a CP*, and every CP* must contain an IP-SUB.

IP-X

As in CHLG, wan(de) clauses that are V1 or V2 (or are too short to diagnose) are tagged as IP-X (with wande a CONJ), as these are ambiguous between matrix and subordinate clauses. Unambiguously V-late wan(de) clauses are tagged CP-ADV with wande in C:

 ...  (IP-X-CONJ (CONJ wan)       <-  *-CONJ because should be new token
                 (NP-SBJ (PRO er))
                 (BEPI^3^SG ist)   <- IP-X because 'wan' followed by V2
                 (NEG nit)
                 (ADJ-PRD (ADJR groesser))
                 ...
  (ID 1533_Rodler_Fierrabras_Rhine.,155))

( (FRAG-SPE (INTJP (INTJ We)
	           (NP-OB2 (PRO^D^SG mir)))
	    (CP-ADV-SPE (C wan) <- subordinate because 'hab' is V-late
			(IP-SUB-SPE (NP-SBJ (PRO^N^SG ich))
				    (ADVP (ADV nun))
				    (NP-OB1 (PRO$^A^SG meinen) 
                                            (N^A^SG suone))
				    (VBN verlorn)
				    (HVPI^1^SG hab)))
	    (, /)
  (ID 1533_Rodler_Fierrabras_Moselfrk.,90))

Non-finite IP (IP-INF)

IP-INF is required when an infinitive occurs, except when selected by a modal (see wil auch nit widerum heym ziehen above) or auxiliary sein/haben (e.g. es ist nicht zu verwundern).

IP-INF is frequently selected by the verbs anfangen, beginnen, lassen, meinen, pflegen, sehen, hören (as in CHLG). NPs within the infinitival should be sub-tagged according to their grammatical function (as if the verb were finite). With verbs of perception/AcI, the agent of the infinitive is NP-SBJ but case tagged as accusative:

(IP-MAT (NP-SBJ (NPR Fierrabras))
	(VBDI^3^SG hort)
	(IP-INF (NP-SBJ (PRO^A^SG jn))    <- agent of reden
	        (ADVP (ADV als)
	              (ADV dapffer))
	        (VB reden))	 
   …
  (ID 1533_Rodler_Fierrabras_Rhine.,108))

If the accusative agent of an ACI verb is moved out of IP-INF, the moved agent is tagged as a bare NP (to prevent the matrix clause from appearing to have two subjects):

  ... (CP-DEG (C das)
	      (IP-SUB (NP-SBJ (PRO^N^SG er))
                      (NP-1 (PRO^A^SG yn))    <- moved subject is bare NP
		      (PP (P fur)
			  (ADJP (ADJ^A^SG dot)))
		      (VBDI^3^SG ließ)
		      (IP-INF (NP-SBJ *ICH*-1)  <- trace of ACI subject
	                      (VB ligen))))))
    (, .))
  (ID 1430_NN_Karrenritter_SRhFrk.,133))

Infinitivals that are the complement of a degree-modified ADJ/ADV/Q (zu viel gesehen, um umzudrehen) should be sister of ADJ/ADV/Q:

...  (NP-SBJ (D dißer)
             (N walt))
     (BEPI^3^SG ist)
     (NP-OB1 (PRO^D^PL vns))
     (ADJP-PRD (ADV allzu)
               (ADJ dick)
               (IP-INF-SPE (TO zu)
                           (VB jostiern)))
     (CODE .))
  (ID 1430_NN_Karrenritter_SRhFrk.,106))

Purpose clauses are extended with IP-INF-PRP (replacing HeliPaD's IP-INF-NCO) unless um is overt:

( (IP-MAT (NP-SBJ *con*)
          (VBDI^1^PL schickten)
          (NP-OB1 (PRO$ vnsere)
                  (ADJ kleynen)
                  (N^A^PL schiff))
          (PP (P nach)
              (NP (NPR Lissebona)))
          (, /)
          (IP-INF-PRP (NP-OB1 (D solches))
                      (NP-OB2 (D dem)
                              (N kuenige))
                      (RP an)
                      (TO zu)
                      (VB zeygen)
          ...
  (ID 1557_Staden_Historia_Hesse.,133))

See also IPX for incomplete IPs, and NP-SMC / ADJP-SMC for small clauses (replacing HeliPaD's IP-SMC).

Other clause-like constituents

PTP

PTP (participial phrase) houses a participle that is not part of the clause's verbal complex, nor attributive to an N (which are treated as adjectives before N and reduced relatives when immediately following N). PTPs are thus are either absolutives, or modify Ns without being adjacent to N, or are parenthetical as in the following:

(IP-SUB (NP-SBJ *T*-5)
	(PP (P zo)
	(NP (D^D^PL den) (N^D^PL zijden)))
	(VBN geachtet)
	(RDDI^3^SG wart)
        (PTP (NP-SBJ (ONE^N^SG eyn))     <- parenthetical participial
             (PP (P zo)
	         (NP (D^D^SG deme) (Q^D^SG andren)))
	     (VBN gerechent))
	(NP-OB1 ... (NUM^A^SG CM.) (N^A^SG gulden)))))))
  (ID 1360_Hauwe_NeuesBuch_Cologne.,52))

FRAG

A fragment is a group of two or more constituents that do not form a constituent, or a non-IP root node (see also Helipad for the use of FRAG in restarts). Root FRAG is frequently an NP (e.g. a chapter title), or a sentence token consisting of only a subordinate clause:

( (FRAG (NP (N Auszfart)
            (NP-POS (PRO$ meiner)
                    (ADJ ersten)
                    (N schifffart)
                    (PP (P von)
                        (NP (NPR Lissebona)))
                    (PP (P auß)
                        (NP (NPR Portugal)))
                    (, .))))
  (ID 1557_Staden_Historia_Hesse.,116))

( (FRAG (ADVP (ADV Vort))
	(CP-THT (C dat)
		(IP-SUB (NP-SBJ (D die)
				(N Amptlude)
				(PP (P in)
				    (NP (D den) (N geburhuseren))))
			(VB richten)
			(MDPI^3^PL solen)
			(PP (P mit)
			    (NP (NUM vier) (N geboiden))
                        ...
  (ID 1360_Hauwe_NeuesBuch_Cologne.,97))

On the other hand, a verb-final root node that is not introduced by an obvious subordinator is simply IP-MAT:

( (IP-MAT (PP (ADV+P Darunb))
          (NP-SBJ (NPR Philager))
          (, /)
          (NP-OB1 (D^D^SG dem)
                  (, /)
                  (CP-REL (WNP-SBJ-1 0)
                          (C so)
                          (IP-SUB (NP-SBJ *T*-1)
                                  (NP-OB1 (PRO^A^SG ynenn))
                                  (VBDI^3^SG fraget) ...   
          (ADVP (ADV recht))
          (VBN geantwort)
          (HVPI^3^SG hatt)
          ...
  (ID 1537_JungeFursten_Politics_Cologne.,10))

QTP

Any direct speech that is smaller than a clause (which would be IP-*-SPE) is housed in QTP ('quotation phrase'):

( (IP-MAT (VBDI^3^PL sagten)
          (NP-SBJ (PRO^N^PL sie))
          (, /)
          (QTP (PP (P in)         <- the quote is just a PP
                   (NP (NUM 28.)
                       (N gradus))))
          (, /)
          (CP-ADV (C wie)
                  (IP-SUB (NP-SBJ *pro*)
                          (ADJP-PRD (ADJ wahr))
                          (BEPI^3^SG ist)))
          (, .))
  (ID 1557_Staden_Historia_Hesse.,310))

( (IP-MAT-SPE (CODE-QL ")
              (QTP (INTJ Ja)
                   (NP-SBJ (PRO er)))
              (CODE-QR ")
              (CODE ,)
              (IP-MAT-PRN (VBDI^3^SG sprach)
                          (NP-SBJ (D der)
                                  (N konig))
                          (CODE ,))
   ...
  (ID 1430_NN_Karrenritter_SRhFrk.,90))