Nominalizations

Universität Stuttgart
Graduiertenkolleg "Sprachliche Repräsentationen und ihre Interpretation"
Guest lecture series

Nominalizations and thematic interpretation of genitive attributes

Nadine Aldinger - 28/10/2004

Contents

I.	General topic

	1.	Sortal readings of -ung nominalizations
	2.	Tests for sorts
	3.	Problems with sort tests
	4.	Aims and methods of my work

II.	Corpus study: Thematic interpretation of genitive attributes of -ung nominals

	5.	Theoretical views on verb-to-noun argument inheritance
	6.	Setup of the study
	7.	Results: Telic and atelic verbs, plurals, Content Objects
	8.	Summary
	9.	Discussion

I. General topic

	(1)		Um 18.30 Uhr war die Absperrung des Regierungsviertels vollzogen. At 18.30 the closing off of the government quarter was complete. (event)
	(2)		[D]ie Absperrung der Lurgiallee [...] läßt so manchen unverhofften Besucher scheitern. The closing off of the Lurgiallee causes many an unexpected visitor to fail. (state)
	(3)		Hinter den Absperrungen der Polizei sammelte sich im Verlauf des frühen Abends eine große Menge Schaulustiger[.] A large crowd of spectators gathered behind the police's barriers in the course of the early evening. (object)

(all sentences from Frankfurter Rundschau Corpus, 1992-1993, ~40 million words)

1. Sortal readings of -ung nominalizations (1)

Basic sorts

Eventuality ("verb-like")
- Process (P)
- Event (EV)
- State
  - Result State (RS)
  - Non-Resultative State (N-RS)
Object ("noun-like")
- Result Object (RO)

Verfolgung (tracking)

Vernichtung (annihilation)

Beklebung (sticking onto)

Bewunderung (admiration)

Umgebung
(surroundings)
Lieferung, Absperrung
(delivery, barrier)

(based on Ehrich/Rapp 2000)

1. Sortal readings of -ung nominalizations (2)

What kinds of base verbs can form nominalizations with particular readings?

Eventuality
- Process
- Event
- State
  - Result State
  - Non-Resultative State
Object
- Result Object

mainly atelic (Activity) verbs

telic verbs

telic verbs

stative verbs

mainly telic verbs

(based on Ehrich/Rapp 2000)

1. Sortal readings of -ung nominalizations (3)

Some refinements of base verb selection:
lexical semantic classes of telic verbs

		P	RS	RO
verbs of production	Herstellung, Errichtung production, building
verbs of annihilation	Vernichtung, Absetzung annihilation, dismissal
verbs of availability	Entdeckung, Ausgrabung discovery, excavation			X
verbs of change	Umgestaltung, Renovierung rearrangement, renovation	X	(X)
concrete / locative application verbs	Beklebung, Versiegelung sticking (onto), sealing	X	X	X (Applicatum!)
abstract / effecting application verbs	Bearbeitung, Übersetzung revision, translation	X		X

(based on Ehrich/Rapp 2000 - classes are defined via characteristic lexical-semantic structure)

2. Tests for sorts (1): Process / Event vs. State (a)

Process and Event nominalizations allow predicates that relate to parts of the event expressed by the nominal (4), States do not (5):
(4)	Die Verfolgung des Täters wird fortgesetzt. The pursuing of the perpetrator is continued.	(Process)
(5)	Die Absperrung des Geländes wird unterbrochen. The closing off of the premises is interrupted.	(Event / Result State)*

Processes and Events allow predicates which modify the progress of the event (6), States do not (7):
(6)	die vorsichtige / umständliche Verfolgung des Täters the careful / complicated pursuing of the perpetrator	(Process)
(7)	die vorsichtige / umständliche Absperrung des Geländes the careful / complicated closing off of the premises	(Event / Result State)*

(tests from Ehrich/Rapp 2000:252-254;
see also Vendler 1968b, Dowty 1991 on tests for verbal Aktionsart)

2. Tests for sorts (2): Process / Event vs. State (b)

States can be combined with stative predicates and predicates of evidence (8), Processes and Events cannot (9):
(8)	* die bestehende / vorgefundene Verfolgung des Täters the existing / found pursuing of the perpetrator	(Process)
(9)	die bestehende / vorgefundene Absperrung des Geländes the existing / found closing off of the premises	(Result State / Event)*

2. Tests for sorts (3): Process vs. Event

Process nominalizations (as well as Result States) allow time-span predicates (10), Events do not (11):
(10)	die tagelange Verfolgung des Täters the pursuing of the perpetrator for days	(Process)
(11)	die tagelange Absperrung des Geländes the closing off of the premises for days	(Process / Event)*

Events can be modified by time-frame predicates and incremental predicates (12), Processes cannot (13):
(12)	* die in zwei Tagen erfolgte / allmähliche Verfolgung des Täters the pursuing of the perpetrator in two days / the gradual pursuing of the perpetrator	(Process)
(13)	die in zwei Tagen erfolgte / allmähliche Absperrung des Geländes the closing off of the premises in two days / the gradual closing off of the premises	(Event / Process)*

3. Problems with sort tests (1)

Practical problems: Applying tests to real-life instances of nominalizations

Sortally ambiguous nominalizations (e.g. Absperrung):

automatic identification of sortal readings:
diagnostic contexts have to be searched and cannot be manipulated
- contexts are artificial (mostly DP-ung as subject + diagnostic predicate + nothing else)
- the predicates proposed are too few and occur too rarely in text corpora to be of great use (what predicates, e.g., can be "predicates of evidence"?)
semi-automatic identification:
(presumably) relevant context is extracted automatically and then checked by a human
- contexts may be changed, but only very carefully
- more effort

Sortally unambiguous nominalizations (e.g. Verfolgung):

possible readings depend on the verb's Aktionsart
-> additional knowledge needed -
from machine-readable semantic lexicons or human annotators

3. Problems with sort tests (2)

Theoretical problems

Acceptability judgments differ (e.g. die bestehende Verfolgung);
priming effects etc.
Acceptability is influenced by other contextual parameters, which are not always taken into account, e.g.
- number (Zerstörung destruction vs. Zerstörungen destructions)
- (in)definiteness
- attributes
  (die seit gestern bestehende Zerstörung the destruction existing since yesterday
  vs. * die seit gestern bestehende Zerstörung des Hauses the destruction of the house existing since yesterday)
-> Do the tests test what they are supposed to?
If there is no diagnostic context, is there no sort?
-> ambiguity, vagueness, underspecification?
If there is - which readings can be coerced under which circumstances?
-> status of unmarked vs. marked vs. impossible readings

4. Aims and methods of my work

To develop new tests for identification of sortal readings and improve existing ones
- suitable for (semi-)automatic identification from corpus text
- based on corpus data
Application: "weighted guessing" of sortal readings for particular instances
multi-parameter approach: no ultimately decisive feature
Focus on interpretation of genitive attributes of nominalizations (-> Part II)
- genitive interpretation can favor ( / decide?) sortal reading
- real ambiguities

II. Corpus study:
Thematic interpretation of genitive attributes of -ung nominals

	(1)		Um 18.30 Uhr war die Absperrung des Regierungsviertels vollzogen. At 18.30 the closing off of the government quarter was complete. (object genitive)
	(2)		[D]ie Absperrung der Lurgiallee [...] läßt so manchen unverhofften Besucher scheitern. The closing off of the Lurgiallee causes many an unexpected visitor to fail. (object genitive)
	(3)		Hinter den Absperrungen der Polizei sammelte sich im Verlauf des frühen Abends eine große Menge Schaulustiger[.] A large crowd of spectators gathered behind the police's barriers in the course of the early evening. (subject genitive)

5. Theoretical views on verb-to-noun argument inheritance (1)

Morphosyntactic view (e.g. Reis 1988):

nominal inherits syntactic argument structure of the verb
structural verbal cases (nominative, accusative) acquire structural genitive

Lexical-semantic view (e.g. Ehrich/Rapp 2000):

nominal inherits semantic argument structure of the verb
(arguments of decompositional predicates, thematic roles)
mapping of semantic argument structure onto syntactic argument structure (Linking) follows independent principles for nominals and verbs; perhaps differences between nominal sorts

"The issue is not, whether argument inheritance behaviour can be completely handled by word syntactic principles, but whether we can completely do without them." (Reis 1988, p.55)

5. Theoretical views on verb-to-noun argument inheritance (2)

Ehrich/Rapp (2000):

semantic argument structure of a regular -ung nominal depends on
- sort of the nominal,
- lexical-semantic structure of nominal and base verb
- additional, semantically driven processes
  (e.g. plurals of non-Process -ung nominals behave like Process nominals [Ehrich 2002])
Linking principles:
- Process nominals (= no BECOME predicate in LSS): all arguments in the nominal's lexical-semantic structure (LSS) can be linked
- nominals from telic verbs: only the most deeply embedded argument in LSS can be linked - all other arguments are suppressed
Consequences:
- transitive verbs: true subject genitives occur as attributes of Process nominals from Process verbs (Verfolgung pursuing) and from atelic variants of change-of-state verbs (Bemalung painting) and only there
- apparent subject genitives of other nominals are instances of genitivus auctoris (author's genitive) and are not related to verbal structure

6. Setup of the study (1)

Occurrence of object- and subject-like genitives, depending on

aspectual and lexical-semantic properties of the base verb
(Verfolgung pursuing _process vs. Absetzung dismissal _event;
Zerstörung destruction _{annihilation verb} vs. Umgestaltung rearrangement _{verb of change})
thematic properties of the base verb
(Behandlung treatment _{Agent-Patient}
vs. Befürchtung fear _{Experiencer-Relatum}
vs. Beobachtung observation _Agent-Theme)
sortal reading of the -ung nominal (if identifiable manually)
number and definiteness of the -ung nominal
(die Verfolgung _def.sg. vs. die Verfolgungen _def.pl. vs. eine Verfolgung _indef.sg. vs. Verfolgungen _indef.pl. vs. (nach) Verfolgung _{null sg.}
-> 5 categories
number and definiteness of the genitive attribute

text material: newspaper corpus - part-of-speech tagged, chunked (YAC)

6. Setup of the study (2): Items (a) - nominals from telic verbs

base verb properties >	aspect class	lexical-semantic class*	propositional argument
Herstellung production Errichtung raising	telic	production	-
Absetzung dismissal Vernichtung annihilation	telic	annihilation	-
Entdeckung detection; discovery	telic	availability	possible
Messung measuring	telic / atelic	availability?	possible
Umgestaltung rearrangement	telic / atelic	change	-
Absperrung closing off; barrier	telic / atelic	concrete application	-
Bearbeitung revision Übersetzung translation	telic / atelic	abstract application	-

* verb classes from Ehrich/Rapp (2000), cf. 1.3

6. Setup of the study (3): Items (b) - nominals from atelic verbs

base verb properties >	aspect class	propositional argument
Beobachtung observation	atelic	possible
Unterstützung support	atelic	possible
Verfolgung tracking; pursuit; persecution	atelic	-
Befragung questioning; survey	atelic / telic (task reading)	-
Behandlung handling; treatment	atelic / telic (med.)	-
Befürchtung fear	(Mental) State	obligatory

7. Results (1): Matches

number of evaluated matches per item (with genitive attribute):
between 13 (Absperrung) and ~800 (Unterstützung) - avg. 195
max. 200 matches per morphosyntactic category
(e.g. N def sg, G def) were evaluated
(classified according to genitive interpretation)

7. Results (2): Subject genitives (a) - nominals from telic verbs (1)

subject genitives do not occur with singular-number -ung nominals, except for
- 7 with Bearbeitung - (14), (15)
- several with Übersetzung - (16), all with a proper name as genitive attribute
- 1 with Entdeckung - (17)
with plural nominals, subject genitives consistently prevail over object genitives - (18)
- again with the exception of Bearbeitung (no subject genitive with plural) and Übersetzung (few)
(note: no plurals found with Herstellung, Vernichtung, Umgestaltung)

Influence of sortal reading - examples:

Absperrung: most plural nominals have Object reading, or are vague between Object and State (19); subject genitives: 5 of 6 genitive attributes of Absperrungen
Bearbeitung: all subject genitives except one (15) are embedded in a PP-in (like (14)); all nominals with subject genitives have Object or Object/State reading

7. Results (3): Subject genitives (a) - telic verbs (2)

(14)	20.30 Uhr: "Geschlossene Gesellschaft", Jean-Paul Sartres düsteres Kammerspiel über die Hölle auf Erden, in einer Bearbeitung des Theaters in Bornheim [...].
(15)	[E]s war nicht der originale Schumann, sondern die [...] Bearbeitung eines ungenannten Musiker-Komponisten.
(16)	Aber in der Personenkonstellation und im Charakter zeigt sich auch eine Nähe zu Goethes "Iphigenie auf Tauris", was durch den Ton und die Sprache der Übersetzung Wielands unterstützt wird.
(17)	Die Europäisierung nach der bösen Entdeckung jenes Indianers ist also noch keineswegs abgeschlossen.
(18)	Das lasse sich jedenfalls aus den Ergebnissen der jahrelangen Messungen seines Institutes nicht herauslesen.
(19)	Sie wurden aber durch großräumige Absperrungen der Polizei daran gehindert, an ihr Ziel zu gelangen.

7. Results (4): Subject genitives (b) - atelic verbs (1)

occasional subject genitives found: Verfolgung, Befragung, Behandlung
subject and object genitives balanced: Beobachtung, Unterstützung
almost only subject genitives: Befürchtung

Examples:

occasional subject genitives - Verfolgung:
N definite singular, Gen definite (die Verfolgung _def.sg. des Täters _def.):
subject genitive : object genitive = 2 : 76 (20)
N definite plural, Gen definite (die Verfolgungen _def.pl. des Täters _def.): 2 : 0 (21)
subject and object genitives balanced - Unterstützung:
N definite singular, Gen definite: 100 : 100 (22)
almost only subject genitives - Befürchtung:
N definite singular, Gen definite: 12 : 2
- all object genitives are Event nouns (deverbals or coerced Process interpretation) (23)

7. Results (5): Subject genitives (b) - atelic verbs (2)

(20)	a.	Laut Alvarez leben in den Widerstandsdörfern rund 45 000 Menschen, die Anfang der 80er Jahre vor der Verfolgung der Armee aus ihren Dörfern geflohen waren.
	b.	Ihr mißfällt , daß die Verfolgung des Geschlechts - so in Indien und in islamischen Staaten - in Deutschland als Asylgrund nicht anerkannt wird.
(21)		Michel Foucault hat einmal gesagt, das ruhige Wissen von den Ordnungen bei Linné habe seine Quellen in den blutigen Verfolgungen der Inquisition.
(22)	a.	Nach dem Krieg wurde einer der Brückenbögen mit der Unterstützung der Stadt zu einem Bootshaus ausgebaut.
	b.	Als letzter von insgesamt 13 Abschnitten des Wahlprogramms erscheint die Unterstützung der heimischen Wirtschaft und deren Beschäftigten.
(23)	a.	Obwohl Unger und Lang Verständnis für die Befürchtung der Anwohner von Asylunterkünften in Maintal äußerten, [...].
	b.	Noch dominieren Desinteresse oder die Befürchtung reglementierender Studienverschärfungen.

8. Summary (1): Explaining (away) subject genitives (a)

Subject genitive as default: Content Objects

verbs which subcategorize for a proposition as internal argument
form -ung nominals referring to that proposition,
which in turn can take the "owner" of the proposition as a (subject) genitive attribute
- (23)a, (24)

Subject and object genitives both unmarked: Plurals

plural -ung nominals denote a sum over Processes / Events - (18), (21);
any(?) thematic role-player of the verb can serve as common feature of the summed elements

8. Summary (2): Explaining (away) subject genitives (b)

More contexts that loosen restrictions on subject genitive (but less regularly -> "coercion"):

genitive attribute violates selectional or contextual restrictions on the object - (15), (19), (21)
object referent is mentioned previously in the sentence - (20)a, (25)

certain verbs (brauchen to need, finden to find, rechnen mit to reckon with) which take the internal argument of the nominal's base verb as subject and the -ung nominal as object:

[Der Entwurf]_subj	findet	[die Zustimmung	der Abgeordneten]_obj.
the bill	finds	the approval	of the MPs

[Die Abgeordneten]_subj	stimmen	[dem Entwurf]_iobj zu.
the MPs	approve	of the bill

- (26)

nominal + attribute are embedded in a PP - (14), (22), (27); especially for Content Objects (27)

8. Summary (3): Explaining (away) subject genitives (c)

(24)	Auf eine Beobachtung des Hamburger Polizeiexperten B. möchte ich noch hinweisen.
(25)	Seit einem Jahr arbeitet Halter an einem Film, der die Erinnerung an Menschen wachhält, die während des Krieges Juden vor der Verfolgung der Nazis retteten.
(26)	Der SKV braucht die Unterstützung seiner Fans.
(27)	In den neuen Bundesländern ist nach der Beobachtung des Verbandes der "Neugier-Reiseboom" vorbei.

8. Summary (4)

with nominals from atelic verbs, subject genitives are more restricted than Ehrich/Rapp's theory predicts; with some nominals from telic verbs, they seem less restricted (maybe due to Object reading)

-> my suspicion: nominals from telic and atelic verbs do not behave differently w.r.t. subject genitives

-> wanted: explanation for occurrence / non-occurrence of subject genitive (and genitivus auctoris) without telic/atelic-distinction
working hypotheses:
- default for Process / Event readings: only object genitive allowed
- reasons for frequent/unmarked occurrence of subject genitive with a given -ung nominal:
  - plural (sums)
  - Object reading in general
  - the nominal's base verb governs a propositional argument
    -> Content Object reading
  - static base verb?
- reason for occasional/marked occurrence of subject genitive: "coercion"

9. Discussion

"true" subject genitive vs. genitivus auctoris:
standard test (replacing by PP-durch by) seems to test only for Process / Event reading of the head nominal
-> does the distinction (only?) reflect the sortal reading of the head nominal?
Content Object hypothesis should be tested with telic verbs subcategorizing for a propositional argument (do such verbs exist?)
coercion seems less acceptable with most telic verbs
- cf. (26) vs. (28):

(26) Der SKV braucht die Unterstützung seiner Fans.

(28) ? Das Haus bedarf der Renovierung eines erfahrenen Architekten.

References

Bierwisch, Manfred (1989): Event nominalizations: Proposals and problems. In: Linguistische Studien A 194.

Dowty, David (²1991): Word meaning and Montague grammar: the semantics of verbs and times in generative semantics and in Montague's PTQ. Dordrecht: Reidel (u.a.). Reprint.

Ehrich, Veronika; Rapp, Irene (2002): Nominalizations and temporal prepositions. In: ZAS Papers in linguistics 27, pp.39-66.

Ehrich, Veronika; Rapp, Irene (2000): Sortale Bedeutung und Argumentstruktur: -ung-Nominalisierungen im Deutschen. In: Zeitschrift für Sprachwissenschaft 19, pp.245-303.

Ehrich, Veronika (2002): The thematic interpretation of plural nominalizations. In: ZAS Papers in linguistics 27, pp.23-38.

Lindauer, Thomas (1995): Genitivattribute: eine morphosyntaktische Untersuchung zum deutschen DP/NP-System. Tübingen: Niemeyer.

Grimshaw, Jane (1990): Argument structure. Cambridge, MA: MIT Press.

Reis, Marga (1988): Word structure and argument inheritance: how much is semantics? In: Linguistische Studien A 179, pp.53-67.

Teußer, Yvonne (2000): Zur Nominalisierung psychisch zuständlicher Verben. Zul.-Arb., Univ. Tübingen.

Vendler, Zeno (²1968): Verbs and times. In: Linguistics and philosophy. Cornell University Press.

(26)	Der SKV braucht die Unterstützung seiner Fans.
(28)	? Das Haus bedarf der Renovierung eines erfahrenen Architekten.