EESE 6/1997

Prepositional Attributes:
Syntax or Semantics? Grammar or Lexicon?

Stefan J. Schierholz (Göttingen)

1. The subject under discussion
2. Preliminary works
3. Characteristics of the PA-construction
4. Future Research

1. The subject under discussion

A Prepositional Attribute is to be defined as a prepositional phrase connected to a noun phrase (= NP) and governed by the particular noun of the NP. The constituents of a prepositional attribute are the preposition and the attached NP.

In order to keep the terminology here at least to a certain extent independent of other theories, the constituents appearing in this study are to be defined as follows:

With these kinds of constructions, the definitions assume an idealised and drastically simplified PA-construction which can arise in this form in normal language texts, but can also appear in a more complex version. The relations between the predecessor-NP<PA>, the preposition, and the postposition-NP<PA> are bound by definite syntactic and semantic features which will have to be analysed in detail at a later stage.

As it is important for a more precise definition of PA-constructions that similar syntactic constructions such as adverbials should also be defined, further terms will have to be defined as follows:

By means of these technical definitions, a certain area of independence from other theoretical approaches has been attained.

2. Preliminary works

The PA-construction is handled in a variety of ways in the relevant literature, in the grammars, and in both general and specialist German dictionaries. Formal grammar models are based mainly on examples of particular cases,1 and German valency dictionaries2 rarely give a coherent picture of the relevant features so that the syntactical and semantic relations are not systematically displayed and the German nouns are not treated comprehensively, but only in an 'ad hoc' fashion.3 Both traditional German grammars and studies concentrating on PA-constructions4 fail to make the fundamental defintion of PA-constructions to distinguish them from related syntactic constructions and in particular, from AAP-constructions. It is also certainly the case that analyses aimed at making this distinction can be criticised on the following grounds:5

1. PA-constructions and AAP-constructions which come after nouns are always defined in a contrastive manner, whereas the possible combinations with each other are not even discussed.
2. The position and function of the preposition - i.e. whether it belongs to the predecessor-NP or to the postposition-NP - still remains unclear.
3. Statements resembling rules are made concerning the internal structure of the PA-construction. The examples in the grammars selected for this purpose are not applied to test the statements and are not intended to be used for an inductive inference to generalisations.
4. The PA-construction has to be described with considerably more complex syntactic and semantic features than has been the case so far.

3. Characteristics of the PA-construction

PA-constructions will have to be deliberately formulated to check the following features or tests for the PA-construction so that the features will be either restricted or rejected to remain within their area of validity. With the help of text corpus analyses6, the relationships between actual language use and possible regularities should be clearly demonstrated. It should also be shown that there are only a few possible generalisations with regard to PA-constructions and that the main emphasis concerns idiosyncratic phenomena of the predecessor-noun<PA> which are the main factors in determining the internal structure of the PA-construction.

1. The postposition-NP takes a different case marker in the PA from the attributive adverbial phrase (= AAP).

The AAP (5) referring to a particular place (location in answer to the question where?) takes "auf" with the dative whereas the postposition-NP<PA> (6) following the preposition "auf" is in the accusative.
However, this distinction depends on the respective prepositions in their prepositional phrase.

Both prepositions "unter" ((7) and (8)) and "bei" ((9) and (10)) take the dative in German so that the case marker cannot be applied as a distinguishing criterion.

2. In the PA-construction, the preposition and the postposition-NP<PA> can be replaced by an adverbial pronoun formed from the relevant preposition (da-/wo- + preposition or as appropriate, dar-/wor- + preposition for prepositions beginning with a vowel).

In the AAP, a simple adverb stands as a possible substitute, a pro-form, which can be used either in the attached phrase referring to the AAP or instead of the AAP.

In fact, the application of this test depends on the content of the postposition-NP<PA>.

Adverbial pronouns can only be substituted for nouns referring to inanimate object and standing in a PA.7

In German, there are a few cases when the adverbial pronoun attains homonymy with an AAP:

With many prepositions, there is homonymy not only with regard to the adverbial pronoun. All the predecessor-nouns taking "nach" can also be joined to an AAP of time by "nach" and/or a PPA beginning with "nach".

(16) Die FRAGE nach dem Examen DANACH? [da'na:-]
(16') Die Frage NACH DEM EXAMEN WANN? DANACH? [da'na:-]
(16a') The QUESTION ABOUT the exam WHEN?

If the meaning is not clear from the context, the ambiguity of the example can only be solved by giving a different stress to "danach".

In many cases, the semantic reference is, however, the decisive factor to distinguish between AAP and PA.

(17) Die SUCHE NACH dem Unfall dauerte ewig. ?DANACH?
(17') Die Suche NACH DEM UNFALL dauerte ewig. WANN?
(17a) The SEARCH FOR the accident took an endless amount of time. FOR THAT
(17a) The search AFTER THE ACCIDENT took an endless amount of time. WHEN
(18) Die SUCHE NACH dem Grappaglas dauerte ewig. DANACH
(18') #Die Suche NACH DEM GRAPPAGLAS dauerte ewig.# #WANN?#
(18a) The SEARCH FOR the grappa glass took an endless amount of time. FOR THAT
(18a') #The search AFTER THE GRAPPA GLASS took an endless amount of time.# #WHEN?#

Whereas the PA-construction in (17) seems to be highly improbable for semantic reasons, the AAP-construction of time in (18) is excluded because there is no action or activity with regard to the "glass of grappa".

These examples, however, bring another problem to light. In the text corpus, no information on the lexemes "Frage" and "Suche" is offered by using an AAP of time.8 In many text examples, ambiguity is avoided by inserting adjectives or adding a temporal AAP to the postposition-NP<PA>. With the word "Suche", the phrase "auf der Suche nach" (= "in search of") occurs very frequently, thus excluding the possibility of a direct link with an AAP of time. In the PA-construction with "Suche" as a predecessor-nounPA, animate objects are usually in the postposition-NP<PA> which similarly excludes the possibility of it being understood as an AAP of time. From this, we can recognise that examples (16) to (18) are merely a case of offering illustrative examples. Although the possibilities of having a PA-construction (the semantic function of the postposition-NP<PA>) in German have been demonstrated, this construction rarely appears in actual language usage.9

3. In PA-constructions, no specifiers (e. g. geradewegs / directly, genau / right, exactly) can be inserted in front of the preposition,10 because the preposition is directly connected to the predecessor-noun<PA>.

In an AAP, these kinds of specifiers which are within the semantic scope of the relevant AAP, can be placed in front of the preposition.11 This rule seems to be equally true in all three languages.

This criterion can be used to identify the above PA-constructions ((19) and (20)); however, even with AAPs, specifiers cannot always be placed in front.

(21) Der Ärger IN ERFURT ==> #Der Ärger GENAU IN ERFURT#
(21a) The anger IN ERFURT ==> #The anger EXACTLY IN ERFURT#
(22) Die Reise nach Lissabon ==> #Die Reise GENAU NACH LISSABON#
(22a) The journey TO LISBON ==> #The journey EXACTLY TO LISBON#.

The more exact the formulation for location within the AAP is, the fewer the specifiers that can be used.

Even with the application of the preposition "bei", a specifier cannot be placed in front of the AAP-construction.

In addition, within the preposition "bei", even the distinction between the PA- and the AAP-construction is still unclear. The main feature of the PA-construction consists in there being abstract verbs or reference to persons in the post-position-NP<PA>.

If an AAP of place beginning with the preposition "bei" is added to a PA-construction, the AAP does not refer to the predecessor noun as in (27), but to the postposition-NP<PA> or to the entire PA-construction.

The reason for this is to be found neither in the syntax of the whole phrase nor in the semantics of the postposition-NP<PA> but more in the capacity of the predecessor-noun<PA> to make connections govern. Almost all the predecessor-nounsPA governing a PA with the preposition "bei" can be directly connected to a local adverbial phrase, an AAP of place, which, however, cannot be introduced by the preposition "bei" although in English the construction seems to be possible.

From this, it follows that AAP-constructions concerning the local place must be subcategorised by searching for various types of formulations constructed on the basis of the prepositions to be inserted. In addition, the combination posssibilities of PA und AAP will thereby be limited. The examples demonstrate that direct combination of an AAP beginning with "bei" is excluded in the case of predecessor-nouns<PA> governing the preposition "bei".

5. If a deverbative or a deadjective is involved in the predecessor-noun<PA>, the PA-construction can be paraphrased by means of the underlying verb or adjective. In this case, the preposition can be adopted.

However, the preposition cannot always be taken over from the underlying verb or adjective (e.g. in German Forderung an / claim on). With regard to the prepositional object (= PPO), the preposition in the PA may be a different one or the underlying verb or adjective takes either a genitive, dative or accusative object.12

(35) Bianka achtet den Kerl. ==> Biankas ACHTUNG VOR dem Kerl
(35a) Bianka respects the fellow ==> Bianka's RESPECT FOR the fellow
(36) Er ist seinen Prinzipien treu. ==> Die TREUE ZU seinen Prinzipien
(36a) He is true to his principles. ==> #The TRUTH TO his principles#
==> #The FIDELITY TO his principles#

Similarly, nouns which are not derived are also able to bind a PA (Appetit / appetite, Attentat / assassination, Chance / chance) or the nouns can only bind a PA in a particular semantic variant which does not exist in any way for the underlying verb/adjective (Anschlag auf / anschlagen).

In addition, there are predecessor-nouns<PA> in which there is a basic connection to the verb/adjective but also in these cases, the postposition-NP<PA> can vary in its meaning to such an extent that it is no longer possible to paraphrase the underlying verb/adjective.

(37) Der VERDACHT AUF den Dieb ==> Jemand verdächtigt den Dieb.
(37a) The SUSPICION OF/TOWARDS the thief ==> Someone suspects the thief.
(38) Der VERDACHT AUF Gelbsucht ==> #Jemand verdächtigt (die) Gelbsucht.#
(38a) The SUSPICION OF jaundice ==> #Someone suspects (the) jaundice.#

There seems to be no rule for the various possible derivations. In the AAP, there can be no paraphrase as the preposition in this case belongs to the postposition-NP<AAP>.

(39) Der Ärger IN DER SCHULE = Jemand ärgert sich/jemanden in der Schule.
(39a) The anger AT SCHOOL =/= Someone feels angry at school.

This does not, however, apply to all AAPs:

(40) Der Ärger AM SONNTAG = Jemand ärgert sich am Sonntag.
(40a) The änger ON SUNDAY = Someone feels angry on Sunday.
(41) Die Verabredung IM OKTOBER = Jemand verabredet sich im Oktober.
(41a) The date IN OCTOBER = Someone has a date in October.

The paraphrase test works only when distinguishing an AAP of location from a PA, and not with AAPs of time. Thus, the test is only valid for prepositions which can only be used in AAPs of location or there must be unambiguous criteria for defining AAPs of time and location.

The definition criteria as demonstrated show that which PA or AAP can be attached depends mainly on the constituent in front of the preposition. This leads to the conclusion that nouns functioning as predecessor-nouns reveal syntactic gaps which are, however, of an optional nature. With generalisations it cannot be decided which preposition can be governed by a noun.

5. If the preposition in a PA-construction is bound to the predecessor-noun<PA> and is thus syntactically required by the predecessor-noun<PA>, the preposition belongs to the predecessor-noun<PA> although this is also at the same time a constituent of the PA where it governs the case of the postposition-NP<PA>. In an AAP-construction, the preposition belongs solely to the PP and the type of AAP is established by the choice of the preposition. For this reason, students of German have to learn the prepositions in conjunction with the predecessor-nouns<PA> ((42)-(45)) and the AAP-constructions must be learnt in their entirety ((46)-(48)):

(42) Mein DANK AN
(42a) My THANKS TO
(43) Er hat APPETIT AUF
(43a) He has APPETITE FOR
(44) Eberhards WUT AUF Bayern
(44a) Eberhard's RAGE AT Bayern FC
(45) Seine WARNUNG AN
(45a) Seine WARNUNG AN
(46a) The Situation Where? AT THE UNIVERSITY
(47) Der Weg Wohin? NACH ERFURT
(47a) The route Where to? TO ERFURT
(48) Der Favorit Wann? IN DIESEM JAHR
(48a) The favourite When? THIS YEAR

Although the various functions of the prepositions can be demonstrated in this way, the governing capacity of the predecessor-noun<PA> or predecessor-noun<AAP> is not sufficiently taken into consideration. When adding an AAP, the predecessor-noun<AAP> also in fact determines what type of AAP can be connected to the noun.

(46') # Die Lage NACH HAMBURG#
(46a') The Situation TO HAMBURG#
(48') Der Favorit NACH BUXTEHUDE#
(48a') The favourite TO BUXTEHUDE#.

6. If it is the case that the constituent preceding the preposition determines whether a PA or an AAP can be added, then it can be established that nouns reveal syntactical gaps which are of optional manner. With Generalisations it cannot be decided which preposition is to be required by the relevant noun. The prepositions take different cases in both the AAP and the PA's and in some cases, the numerus of the postposition-NP is also determined (e.g. after zwischen / between or unter / among). Whereas successive addition of different AAP's is relatively flexible in the case of many nouns (the trip on Sunday from Buxtehude to Erfurt via Göttingen). the use of PA's is more restricted. Although there are occasions when several PA's can follow each other with various prepositions, it is also the case that various prepositions are mutually exclusive. If two PA's follow a predecessor-nounPA, different meanings are conveyed by the appropriate prepositions; whatever the case may be, it is quite clear that the possibilities of combinations of the PA's has proven to be a very complex matter.

Agreement can govern the prepositions with, about or between. PAs with with or between coming after each other can be immediately followed by about ((52) and (53)), whereas with and between are mutually exclusive (54). For non-native speakers of German, two mutually exclusive syntactic readings for the lexeme Abkommen or agreement must be made where the PAs can be introduced with both with and about in the first reading whereas, in the second reading, between and about introduce the PAs. An appropriate explanation of this distinction should be included in mono- and bilingual dictionaries under the lemma Abkommen or agreement. This is, however, not the case in German.

7. The internal structure of a PA-construction is characterised both by the subcategorising frame of the predecessor-noun<PA> and by the selectional restrictions because ceratin constituents for the postposition-NP<PA> are excluded. A detailed analysis of the lexeme Attentat (attack) will demonstrate this point.

An Attentat / assassination is a "politisch od. ideologisch motivierter [Mord]anschlag auf eine im öffentlichen Leben stehende Persönlichkeit"13 (a politically or ideologically motivated attempt on the life of a public personality). The following phrases are given as examples to show how the term is used: "ein Attentat [auf jmdn.] begehen, verüben" (to commit, carry out an assassination [of some one], "einem Attentat zum Opfer fallen" (to be the victim of an assassination), "ein Attentat [auf jmdn.] vorhaben" (to plan an assassination [of someone].14 According to the German dictionaries, an assassination can be carried out only with regard to human beings or other living beings. The following phrases, however, can be constructed.

A court can be an "öffentliche Institution, die vom Staat mit der Rechtsprechung betraut ist, Verstöße gegen Gesetze bestraft und Streitigkeiten schlichtet" (public institution commissioned by the state to carry out the law, with powers to punish offences against the law and to settle disputes), a "Richterkollegium" (assembly of judges) or a "Gerichtsgebäude" (law-court building).15 An assassination attempt can be carried out on the assembly or the collegium which in this context means a "Gruppe von Personen mit gleichem Amt od. Beruf" (group of people with the same office or profession)16 ; it is, however, necessary to read (Verfassungs)gericht (constitutional court) in this fashion to justify the selectional restrictions of Attentat auf (assassination attempt on). The unusual interpretation is acceptable for special contexts and cotexts as in the case when some one has a special relationship with a tom-cat and there is also an element of political or ideological motivation so that it is felt to be an assassination attempt when someone deliberately tries to murder the cat in question.

In the text corpus a total of 96 examples have been found for Attentat auf (assassination attempt on). Of these, 79 refer to human beings, ten to buildings (Cafehaus, Supermarkt, Hotel, Botschaft (a café building, supermarket, hotel, embassy)), four to machines (Flugzeug, Schnellzug (aeroplane, express train)), two to abstract entities (Pressefreiheit, Freiheit (press freedom, freedom)), and one example for an animal referring to a book title (Attentat auf Heilbutt (Assassination of halibut)). Even if the corpus analyses considerably extend the scope of the limits of available choice, the semantics in the postposition-NP<PA> of the predecessor-noun<PA> are still quite coherent. In the case of buildings and machines, the assassination attempts had to apply more to the people living in them rather than the buildings themselves. Even the abstract examples in the postposition-NP<PA> refer indirectly to human beings; this is because it concerns their freedom.

4. Future Research

The above survey still gives only an incomplete picture of the internal structure of the PA-construction. The tests, definitions, and features outlined in this paper show that there is no test which can be applied to all PA-constructions and that each preposition must be analysed on an individual basis as the criteria are to be applied differently to each preposition. All the different readings of the predecessor-noun<PA> together with both the possibilities of the diverse meanings of the predecessor-nouns<PA> and of the postposition-NP's<PA> must be conveyed if the relationships in the PA-constructions are to be fully understood. At the same time, a kind of prototypical PA-construction does, in fact, exist and in most cases for a native speaker, it is possible to decide even with a superficial knowledge of the grammar whether it is a case of a PA and an AAP. It is, however, quite clear that the lexical proportion is greater than the grammatical input so that information on the PA-constructions ought to be included in dictionaries rather than in grammars.

5. Literature

Barnett/Endres/Latniak/Lehmann/Schülzke/Storrer/Zoeppritz (1989): Kriterien für die Kodierung von Lexikoneinträgen (unveröffentlichtes Manuskript). Heidelberg.

Droop, Helmut Günter (1977): Das präpositionale Attribut (= Forschungsberichte des Instituts für deutsche Sprache Bd. 34). Tübingen.
Duden: Grammatik der deutschen Gegenwartssprache (1984): 4., völlig neu bearbeitete und erweiterte Auflage. Hrsg. und bearb. von Günter (sic!) Drosdowski in Zusammenarbeit mit Gerhard Augst, Hermann Gelhaus, Helmut Gipper, Max Mangold, Horst Sitta, Hans Wellmann und Christian Winkler. Mannheim/Wien/Zürich 1984.

Duden. Deutsches Universalwörterbuch 1989: Hrsg. und bearb. vom Wissenschaftlichen Rat und den Mitarbeitern der Dudenredaktion unter der Leitung von Günther Drosdowski. 2., völlig neu bearbeitete und stark erweiterte Auflage. Mannheim/Wien/Zürich.

Duden. Deutsches Universalwörterbuch A-Z (1994): CD-ROM-Version. Mannheim/Leipzig/ Wien/Zürich.

Eisenberg, Peter (1989): Grundriß der deutschen Grammatik. 2., überarbeitete und erweiterte Aufl. Stuttgart.

Engel, Ulrich/Schumacher, Helmut u.a. (1978): Kleines Valenzlexikon deutscher Verben. Forschungsberichte des Instituts für deutsche Sprache. Mannheim 31. Tübingen.

Eroms, Hans-Werner (1981): Valenz Kasus und Präposition. Untersuchung zur Syntax und Semantik präpositionaler Konstruktionen in der deutschen Gegenwartssprache. Heidelberg.

F.A.Z. (1994): Die F.A.Z. auf CD-ROM. Jahrgang 1993. Frankfurt a.M.

Haumann, Dagmar (1993): Referential Arguments (= Arbeiten des Sonderforschungsbereichs 282: Theorie des Lexikons Nr. 44). Wuppertal.

Helbig, Gerhard/Buscha, Joachim (1988): Deutsche Grammatik. 11. Aufl. Leipzig.

Hellwig, Peter (1988): Bausteine des Deutschen. Daten für das Programmsystem PLAIN (unveröffentlichtes Manuskript). Heidelberg.

Lauterbach, Stefan (1993): Genitiv, Komposition und Präpositionalattribut - zum System nominaler Relationen im Deutschen. München.

Rauh, Gisa (1992): Präpositionale Formen im Lexikon (= Arbeitspapiere des Sonderforschungsbereiches 282: Theorie des Lexikons Nr. 12). Düsseldorf.

Rauh, Gisa (1993): Präpositionen und Rollen (= Arbeitspapiere des Sonderforschungsbereiches 282: Theorie des Lexikons Nr. 46). Düsseldorf.

Sandberg, Bengt (1979): Zur Repräsentation, Besetzung und Funktion einiger zentraler Leerstellen bei Substantiven. Göteborg.

Schierholz, Stefan J. (1992): Syntaktische Merkmale deutscher Substantive (= IWBS-Report 214). IBM-Deutschland.

Schierholz, Stefan J. (1995a): "Präpositionalattribute: Wieviel Syntax - wieviel Semantik? Wieviel Grammatik - wieviel Lexik?" In: GAL-Bulletin 22, 19-26.

SCHIERHOLZ 1996 Schierholz, Stefan J. (1996): "Grammatische Informationen zu Substantiven in einsprachigen deutschen Wörterbüchern". In: Wiegand, Herbert E. (Hrsg.): Wörterbücher in der Diskussion II (= Lexicographica. Series Maior 70). Tübingen, 140-203.

Sommerfeldt, Karl-Ernst/Schreiber, Herbert (1983): Wörterbuch zur Valenz und Distribution der Substantive. 3. Aufl. Tübingen.

Tarvainen, Kalevi (1981): Einführung in die Dependenzgrammatik (= Reihe Germanistische Linguistik 35). Tübingen.

Teubert, Wolfgang (1979): Valenz des Substantivs. Attributive Ergänzungen und Angaben (= Sprache der Gegenwart 49). Düsseldorf.

Wahrig, Gerhard (1986): Deutsches Wörterbuch. Herausgegeben in Zusammenarbeit mit zahlreichen Wissenschaftlern und anderen Fachleuten. Völlig überarbeitete Neuausgabe. München.


1 Cf. Eroms 1981, Haumann 1993, Hellwig 1980, Rauh 1992.

2 Cf. Engel/Schuhmacher 1978, Sommerfeldt/Schreiber 1983, Teubert 1979.

3 Cf. Sandberg 1979, Teubert 1979.

4 Cf. Droop 1977, Eisenberg 1989, Lauterbach 1993.

5 Cf. Duden 1984, Helbig/Buscha 1988, Rauh 1992, Schierholz 1992.

6 FAZ 1994. The corpus contains about 500 MB and includes the 1993 edition. Further references and inquiries are taken from FAZ text corpus insofar as no separate points are made.

7 Cf. Helbig/Buscha 1988: 265.

8 On account of the large amount of documentary evidence (almost 200 items), only a part of the evidence is assessed.

9 There is not the place to pursue the problem of judging whether the text corpus is sufficiently representative; this is because it is obvious that one text corpus alone is insufficient to formulate definite propositions concerning the frequency of a phrase.

10 Cf. Rauh 1992: 20f.

11 Cf. Rauh 1992:15ff.

12 Cf. Barnett et al 1989: 53; cf. Eisenberg 1989: 272f.

13 Cf. Duden 1989: 149.

14 Cf. Duden 1989:149; cf. Wahrig 1986:192.

15 Cf. Duden 1989:592.

16 Duden 1989:860. 14