Pinkster, Harm (1942-) [1990], Latin Syntax and Semantics [info], xii, 320 p.: ill.; 24 cm. [word count] [Pinkster].
Previous Section

Next Section

5 Relators

By `relators' I understand formal means which serve to indicate the relation between constituents on the sentence level or the noun phrase level. Means to indicate explicitly the relation between sentences (e.g. `connectors' such as enim, itaque) are not taken into account (cf. chapter 12), nor are coordinators (cf. crosssection 7.1.1). For Latin the following relators can be distinguished:

- cases

- prepositions

- subordinators

- agreement in number and/or gender

In this chapter I treat cases, prepositions and subordinators ( crosssection 5.1.–5.4.); crosssection 5.5. deals briefly with agreement.

5.1. Cases, prepositions and subordinators

Cases, prepositions and subordinators share one characteristic: they can formally mark both arguments and satellites. I first give three examples of arguments characterized in this way, then three of satellites:

(1) nostrae laudi dignitatique favisti (`You have stimulated my fame and esteem', Cic. Fam. 1.7.8)

(2) versabor in re difficili (`I will engage in a difficult matter', Cic. Leg. 3.33)

(3) nihil mihi optatius cadere posse quam ut (`that nothing could be more desirable than that', Cic. Att. 3.1) [Note]

(4) cui flavam religas comam? (`For whom do you bind up your blond hair?', Hor. C. 1.5.4)

(5) in maximis meis doloribus excruciat me valetudo Tulliae nostrae (`In this period of greatest sadness the health of my Tullia troubles me', Cic. Fam. 14.19)

-- 40 --

(6) esse oportet ut vivas, non vivere ut edas (`One ought to eat in order to live, not live in order to eat', Rhet. Her. 4.39.4)

It will be argued below for each of these categories that it is of little relevance to speak of a proper semantic value of these relators within the nuclear predication, whereas it is relevant with regard to their use in the periphery. A special subsection will be devoted to the relationship between prepositions and cases. With some verbs, an obligatory constituent can occur both in the form of a prepositional phrase and in the form of a noun phrase exclusively characterized by a case:

(7) hoc me libera miserum metu (`Liberate me, unhappy one, from this fear', Ter. An. 351)

(8) multos … a summo discrimine mortis liberavit (`He liberated many from the greatest mortal danger', Larg. 70)

With satellites, too, in particular instances there is sometimes a choice between a prepositional phrase and a noun phrase.

5.2. Cases [1]

I do not intend to describe in this section all uses of each individual case, but, rather, to describe the system as a whole. Neither does the following systematic description pretend to offer an explanation for every instance in our texts. I assume that the corpus of texts which has been handed down to us contains – apart from mistakes – a number of isolated, fossilized instances that cannot be captured by any system. This section mainly concerns the use of cases as a characteristic of noun phrases which are not governed by a preposition. [2]

First I present a survey of the distribution and principal functions of the cases. Then I try to point out a number of characteristics of the system as a whole. Finally I discuss the problems which arise from my description of the system.

5.2.1 Distribution of the cases; main characteristics of the system

Table 5.1 indicates the use of the so-called oblique cases on the various levels (sentence; noun phrase and adjective phrase). [3] The nominative has been left out of account. Its relative frequency can be seen in table 5.2, which indicates the use of cases and prepositions in Cic. de Orat. 1.1–73. I have also left out of consideration the vocative (since lexemes characterised by it do not form part of the sentence – Vairel-Carron 1981b) and the so-called accusativus exclamationis. Table 5.1 only illustrates noun phrases not governed by a preposition.

-- 41 --

Table 5.1 Frequency of oblique cases (in %)
Level
CaseNuclear predication FunctionPeripheryNoun and Adjective phrase
Subj. AcIObj. 2pl.aObj. 3pl.Compl. 2pl.Compl. 3pl.Ind. Obj.Pass.
Gen. (20.4)0.2b0.1b0c20.1
Dat. (12.2)2.54.80.5d3.1e1.2
Abl. (21. 7)1.21.80.915.7f2.1
Acc. (45.6)4.824.49.44.91.01g0h
Total number of instances 11,303
AcI: Accusative-and-Infinitive construction.
a In the survey (Bolkestein et al. 1978: 294) the number of cases of `obj./compl. 2-pl.' for a number of authors is puzzlingly low. In the survey fewer cases occur than is stated below. In Pinkster (1980: 114) `OBJ. Aanv.' means Object Complement (see p. 8). In table 5.1 this has been combined with Compl. 3pl.
b 2pl.: oblivisci, but also esse + so-called genitivus pretii.
3pl.: gen. pretii with aestimare, etc.; gen. criminis with absolvere, etc.; commonefacere, etc. Lemaire (1983: 310) also considers alicuius in levare luctum alicuius a genitive argument. To my mind this is an Attribute with luctum.
c The corpus does not contain cases like Germanicus Aegyptum proficiscitur cognoscendae antiquitatis (`Germanicus set out for Egypt to view its antiquities', Tac. A. 2.59; cf. K.–St. I.741).
d Dativus auctoris.
e Includes: (i) various constructions of dative constituents with esse (cf. Bolkestein et al. 1976: 365–8); (ii) the so-called dativus finalis; (iii) the dativus commodi, ethicus, etc. As dativus finalis were reckoned both cases like ne aut paupertas mihi oneri sit (`Lest poverty be a burden to me', Sen. Ep. 17.1; cf. Bolkestein et al. 1976: 368–9 `predicative dative') and cases like auxilio mittere. It is debatable whether the predicative dative can be considered omissible. To my mind it is obligatory (cf. also Scherer 1975: 140).
f Includes the ablative absolute (402 out of a total of 1773 cases).
g Includes: (i) accusative of respect (artem callebat (`He was a master of the art … ', Tac. A. 13.3)); (ii) accusativus spatii; (iii) accusative of direction; (iv) so-called adverbial accusative of extent. Especially in this last group it is defensible to speak of non-productive expressions (e.g. multum te ista fefellit opinio (`That notion had played you very false', Cic. Ver. 1.88)).
h In reality there are two cases, viz. (i) accusative of respect: os umerosque deo similis (`Godlike in face and shoulders', Verg. A. 1.589); (ii) accusativus spatii.

In interpreting table 5.1 the following points should be kept in mind. First, this general survey of total numbers fails to show the extent to which the use of different cases differs according to author or genre. Second, the choice of the corpus partly determines the conclusions that may be drawn from the survey. Various factors play a part in this. First, the subject matter involved: the frequent occurrence of the accusativus spatii in, for example, Caesar is related to the fact that the campaigns described by him frequently involve the covering of certain distances. Second, the literary genre: it is, for instance, well-known that poetry contains a relatively high number of nouns (many `content words'). As a result, (i) there are more nouns and the like, and (ii)

-- 42 --

Table 5.2 Use of cases and prepositions in Cic. de Orat. 1.1–73 (absolute numbers and %)
Marker
LevelCases (without prep.)Prepositions
NominativeGenitiveDativeAccusativeAblative
Subject277 22.1%--61 (AcI) 4.8%--
Subject Compl.53 4.2%--17 (AcI) 1.4%--
Compl. 2pl.--32 2.6%19 1.5%11 0.9%88 7%
Compl. 3pl.--33 2.6%3 0.2%11 0.9%48 3.8%
Obj. 2pl.---86 6.9%--
Obj. 3pl.---52 4.2%--
Satellite--2 0.2%1 0.1%85 6.7%119 9.5%
Noun phrase-170 13.6%2 0.2%--6 (obl.) 0.5%
Adjective phrase-22 1.8%12 1%-13 1%29 (om.) 2.3%
Total number of instances: 1252
Source: P. Masereeuw, research assistant's report, 1980.
AcI: Accusative and Infinitive construction.
Table 5.3 Numbers of noun phrases with and without preposition in Caes. Gal. 1.3–10 and Verg. A. 1.1–156
Number of noun phrases (without preposition)Number of preposition phrasesTotal
Caesar (1034 words)307 (76%)97 (24%)404
Virgil (1095 words)411 (89%)53 (11%)464
prepositions are used less often to mark the function of the nouns etc. in the sentence. This appears from table 5.3. [4] Although it is impossible to determine the exact influence of these (and other) factors, it is not inconceivable that the use of cases was more free in poetry than elsewhere; the Roman reader was willing (and able!) to take this into account. [5]

From table 5.1 the following general conclusions may be drawn with regard to the function of the oblique cases:

(i) the genitive is pre-eminently the case of the noun phrase level;

-- 43 --

(ii) the dative occurs relatively infrequently and is evenly distributed, predominantly marking arguments, i.e. constituents which form part of the nuclear predication;

(iii) the ablative is pre-eminently the case for the marking of satellites, i.e. constituents in the periphery;

(iv) the accusative is pre-eminently the case for the marking of constituents which form part of the nuclear predication.

In the following sections I will discuss `deviations' from these general rules. I first deal with the use of the cases to mark arguments ( crosssection 5.2.2), then the use of the cases to mark satellites ( crosssection 5.2.3.). crosssection 5.2.4 treats a number of particular problems. The distinction between sentence level and (noun and adjective) phrase level is discussed in crosssection 5.2.5. In crosssection 5.2.6 I discuss the differences between my approach and that in traditional grammars, and present some arguments in favour of my approach.

5.2.2 The use of cases in the nuclear predication

(a) Table 5.4 presents the data of table 5.1 grouped in a slightly different way. Table 5.4 shows the following tendency for the occurrence of the cases for marking constituents which form part of the nuclear predication:

If there is one constituent, it is marked by the nominative; if there are two constituents, the second constituent is marked by a case other than the nominative (as a rule the accusative); if there are three constituents, the nominative and the accusative are reserved for the first and second constituents (passivization is always possible) and the third constituent is marked by a case other than nominative and accusative.

Table 5.4 Frequency (in %) of the cases as markers of obligatory constituents with one-, two- and three-place verbsa
Type of argument
Type of verb123
One-placenominative
Two-placenominativeaccusative 88.3
dative 7.6
ablative 3.6
genitive 0.5
Three-placenominativeaccusativedative 70.3
ablative 26.6
genitive 1.7
accusative 1.4
a Active, finite sentences have been counted; impersonal constructions such as taedet me laboris (`I am annoyed at the effort') have been considered as three-place predicates without a nominative. Prepositional phrases have, again, been excluded.

-- 44 --

This tendency shows the existence of a certain hierarchy of cases, and the choice of case seems largely a negative one: `be sure always to use a case other than that used already'. No such tendency holds for the use of cases in the periphery.

(b) Exceptions

A number of exceptions to the rule given above may be discerned. First of all, there are also `positive' factors which determine which specific case is used within the nuclear predication for arguments with the function complement. I return to this point on p. 51. Furthermore, the rule as formulated above is too strict. I first address the latter point.

(i) From table 5.4 it would appear that there are no three-place predicates whose second place is marked by a case other than the accusative. This is not actually the case. It is true that table 5.4 is correct with regard to the material on which it is based, but from the grammars we know that several exceptions exist: among others, the verb interdicere. With this verb the animate entity is marked by the dative and the inanimate entity by the ablative. [6]

(9) cum … interdictum (nobis) externis bellis (`When we were forbidden to wage war against other nations', Liv. 30.44.7)

(10) omnes quibus aqua et igni interdictum est exsules appellentur (`All those to whom water and fire are denied (= who are exiled) are called exiles', Rhet. Her. 2.45)

>The deviant construction can be interpreted as a merger of the constructions of verbs of communication (e.g. dicere, `to say') and verbs of removing (e.g. arcere, `to ward off'). Something similar is the case with invidere + dat. + abl. (`to envy someone for something'), immolare (`to sacrifice') and supplicare (`to entreat') (cf. Sz. 90).

(ii) Verbs governing two accusatives. An exception to the rule that within the nuclear predication cases merely have a distinctive function is formed by verbs with a so-called double accusative. We will see that this is an exception that can be accounted for within the system. The verbs which govern a double accusative can be divided into three different groups; an example of each group follows:

(11) malitiam sapientiam iudicant (`They regard malice as wisdom', Cic. Off. 2.10)

(12) quid? nunc te … litteras doceam? (`What? Do I have to teach you how to read and write?', Cic. Pis. 73)

(13) equitum magnam partem flumen traiecit (`He brought a large part of the cavalry across the river', Caes. Civ. 1.55.1)

-- 45 --

In example (11) we have a construction with an Object constituent and an Object complement (cf. p. 22). Sapientiam predicates something of malitiam, as in the case of the copula esse, of fieri and of other `kopula-artige Verben' the constituent with the function Subject complement predicates something of the Subject constituent. The rule in Latin is that such constituents agree in case and, if possible, also in number and gender. Other examples may be found in K.–St. I.292-7. An example of a `kopula-artige' construction is (14):

(14) (Gyges) repente anuli beneficio rex exortus est (`Gyges has suddenly, thanks to the ring, become king', Cic. Off. 3.38)

In the passive version of verbs with an Object + Object complement construction both the Object constituent and the Object complement constituent are marked by the nominative. Cf. (15) and (16):

(15) me … universa civitas … consulem declaravit (`The entire state declared me consul', Cic. Pis. 3)

(16) consules declarantur M. Tullius et C. Antonius (`M.T. and C.A. are declared consuls', Sal. Cat. 24.1)

This group of verbs seems to be `irregular' from the point of view of the distinctive function of the cases within the nuclear predication, but is `regular' from the point of view of the rule of agreement.

Example (12) is an example of the second group of verbs with which a double accusative occurs. This group consists of verbs meaning `to teach', `to ask', `to demand', `to inform', `to hide'. Other examples may be found in K.–St. I.297–305, e.g.:

(17) te hoc beneficium rogo (`I ask you this favour', Ant. in Cic. Att. 14.13A.3)

(18) Racilius … me primum sententiam rogavit (`R. first asked my opinion', Cic. Q. fr. 2.1.3)

(19) non … te celavi sermonem T. Ampii (`I have not kept the conversation with T. A. from you', Cic. Fam. 2.16.3)

Unlike the group of verbs treated above, this group does not exhibit agreement between the two constituents. In the passive, the finite verb agrees, if possible, with the constituent that indicates the animate entity, cf. example (20): [7]

(20) nosne hoc celatos tam diu! (`That we have been kept ignorant of this for such a long time!', Ter. Hec. 645)

With this group of verbs we sometimes also find a construction which does fit into the system:

(21) Socraten fidibus docuit nobilissimus fidicen (`S. was taught to play the lyre by a very famous lyre-player', Cic. Fam. 9.22.3)

-- 46 --

(22) litteris … docta (`well-read', Sal. Cat. 25.2)

In Petronius (46.7) we even find docere with a third argument marked by the genitive. Note that these verbs often have alternative prepositional constructions.

(23) nunc a te illud primum rogabo (`Now I shall ask you this first', Cic. Fam. 13.1.2) [8]

(24) cum … ex eo de me percontaretur (`When he asked him about me', Cic. Att. 11.10.1)

This group of verbs is characterized by the fact that apart from the three-place frame we also find a two-place frame. At any rate a two-place construction occurs with the inanimate entity as the Object. Often, but not always, the animate entity can be the Object in a two-place frame. The case form of the third argument can be explained on the basis of the occurrence with these verbs of two-place constructions in which the second argument is marked by the accusative. With semantically related verbs, however, which have a third constituent marked by the ablative, there are no two-place constructions in which the second constituent is marked by the accusative. Next to erudire/instituere aliquem aliqua re (`to teach someone something') we do not find erudire/instituere aliquid. [9] To the connection between two-place and three-place constructions we will return below (p. 51). [10] Detailed study of the verbs with `thing and person' Objects is necessary to examine whether also in this group the `irregularity' may be explained on the basis of another `regularity'.

Example (13) is an example of the third group of verbs with a `double accusative'. This group consists of compound verbs with a prepositional prefix. The prefix is related to a preposition which governs the accusative. [11] In the passive the constituent connected with the prepositional prefix remains unchanged:

(25) ne maior multitudo Germanorum Rhenum traducatur (`Lest a greater number of Germans be brought across the Rhine', Caes. Gal. 1.31.16)

The occurrence of the accusative with this – limited – group of verbs is usually explained as a relic. There is no synchronic explanation. [12] More examples may be found in K.–St. I. 305.

In conclusion, it may be said about the three groups of verbs that govern a `double accusative' that:

(a) it can be explained for one group;

(b) in another group there are systematic similarities with other constructions of the same verbs and there are `ungrammatical' exceptions which fit into the case system;

(c) there is a limited group of compound verbs with which the `double accusative' can at least be explained diachronically.

-- 47 --

In view of the limited number of exceptions and the remarks under (a)–(c) above, there is no reason to reconsider the rule formulated on p. 43, which claims that within the nuclear predication the cases have a distinctive function.

5.2.3 The use of cases in the periphery

Table 5.5 indicates the frequency of the occurrence of the cases in the periphery. Unlike constituents in the nuclear predication, constituents in the periphery do not each have their own case marking. The number of peripheral constituents is not determined by the predicate. On the other hand, a number of peripheral semantic functions can only occur with certain types of nuclear predication. As a rule, the semantic functions `Beneficiary', `Instrument' and Table 5.5 Frequency (in %) of the cases marking peripheral constituents (excluding prepositional phrases)
Abl.79
Dat.15.8
Acc.5.2
Gen.0
to some extent `Manner', for instance, can only be used if the nuclear predication contains a combination of a noun phrase referring to an animate entity and a predicate that refers to a state of affairs which can be controlled by an animate entity. Duration Adjuncts, likewise, occur with nuclear predications which describe a non-dynamic state of affairs (cf. crosssection 3.1.). The case form of a peripheral constituent is exclusively determined by the semantic function which this constituent fulfils. From table 5.5 the ablative appears to be the most frequent case. Satellites marked by the ablative are subject to few restrictions with regard to (a) the types of lexemes which may occur (in the case of the accusative this is a highly limited group), and (b) the content of the remainder of the sentence in which they occur.

There is a very close relation between the meaning of the lexeme which occurs in the periphery and the semantic function it fulfils (cf. crosssection 3.4.). Many attempts have been made in the past to determine for all uses of the ablative a limited number of `meanings', a number as small as possible. This has often resulted in rather forced classification of divergent uses under the same heading (see below p. 63). It is precisely because of the close relation between lexical meaning and semantic function that one case may be used to mark a large number of semantic functions, all of which have in common merely that they specify more precisely the content of the nuclear predication. Thus, we arrive at the paradoxical conclusion that the semantic relations within a sentence are revealed by the cases only to a very limited extent, because:

-- 48 --


- within the nuclear predication the predicate determines the possibility of lexemes to occur as arguments with that predicate; the number and nature of the semantic functions are fixed for each verb;

- outside the nuclear predication the lexical meaning itself determines to a high degree whether a lexeme may be used with a given semantic function.

Above I have argued that in nuclear predication and periphery the cases have different functions. This view is not in accordance with common practice in Latin grammars. In general, in these grammars no attention is devoted to the fact that a specific case occurs both in the nuclear predication and in the periphery, but an attempt is made to assign to that case one single value – or, in the case of the ablative, the smallest possible number of values (see p. 63). In itself, the use of one and the same case in the nuclear predication and in the periphery does not present a problem. Compare, for instance, the equally double function of colour oppositions in geographical atlases: on the one hand, colours are used to mark and identify certain characteristics (e.g. various shades of brown to indicate altitude, green to indicate lowland); on the other, colours are used arbitrarily to distinguish, for example, countries or states. The use of different cases in the periphery has an identifying function; the use of different cases in the nuclear predication has a discriminating function.

5.2.4 Problems with regard to the postulated case system on the sentence level

In the preceding sections I have attempted to determine the main characteristics of the Latin case system, leaving out of account statistically negligible exceptions. In this section I address a number of questions which are left unanswered by this approach. In answering some of these questions I attempt to show that the problems are not serious enough to lead us to question the correctness of the main characteristics outlined above; with respect to other questions it will appear that the regularities indicated above are supplemented by other regularities. The problems to be treated are:

(a) How can we explain the apparent opposition between two different cases, e.g. metuere aliquem/alicui?

(b) What is the difference between the two constructions with three-place verbs such as donare?

(c) Is there a regularity in the use of the same case to mark the Complement of two- and three-place verbs which seem to be semantically related (e.g. the ablative in the case of cedere/movere)?

(d) Is there a semantic explanation for the occurrence of a specific (non-accusative) case in the nuclear predication, e.g. the ablative with uti?

(e) Is there a relation between the use of a case in the nuclear predication and in the periphery?

-- 49 --

(f) The so-called ablative of price and value (ablativus pretii) almost exclusively marks words which mean `price' in some way or other; as Complement of `Price' of a (three-place) verb like `to buy', however, we may find lexemes with widely diverging meanings. Is there a relation between lexical meaning, semantic function and case form?

(g) How does the Accusative and Infinitive construction (AcI) fit into the system?

(a) The `opposition' dative/accusative with metuere, cupere, etc.

As to the occurrence of specific cases to mark the second (non-Subject) constituent, two-place verbs can be divided into three groups: (a) There are verbs which allow only one case; this is the largest group. With, for example, laudare we find only the accusative, with favere only the dative. Borrowing a term from phonology, we may speak of `complementary distribution'. (b) Then there is a case like potiri + genitive or ablative: `free variation', without a difference in meaning. (c) A third group is formed by those verbs which have both a construction with the dative and a construction with the accusative (or a prepositional phrase), described in K.–St. I. 336–41. Here it is commonly assumed that there is a semantic difference that correlates with the difference in construction. We might call this an example of `opposition'. Examples of the last type are:

(26a) quem metuunt oderunt (`Whom they fear, they hate', Enn. Sc. 402)

(26b) etiamsi nos nobis non timeremus, tu tibi metuere debueres (`Even if we did not worry about ourselves, you ought to worry about yourself', Sen. Contr. 1.2.3)

(26c) nullam maiorem pupillo metuunt calamitatem (`They do not fear a greater disaster for their pupil', Cic. Ver. 1.141)

(27a) se … cupere pacem (`That he longs for peace', Cic. Att. 14.20.4)

(27b) te … ipsum cupio (`I long to see you', Cic. Fam. 1.9.9)

(27c) quod ipsi cupio Glycerio (`That I want the best for G.', Ter. And. 905)

Instances such as (26a–b) and (27b–c) are sometimes adduced to prove that the difference in content of the sentences is caused by the difference in case form. I have added (26c) and (27a) in order to show that there may be more at stake. In the case of metuere, the dative constituent need not necessarily be considered obligatory: (26c) shows that metuere may occur with both an accusative and a dative constituent. Here the dative constituent may be seen as a satellite with the semantic function Beneficiary (so-called dativus (in) commodi). Thus we could see (26b) in relation with (26c) rather than in opposition with (26a). In (26c), then, the object of the fear is specified, whereas in (26b) it is more general, or rather, unspecified. Cupere, however, presents a

-- 50 --

slightly different picture. From the data given in the TLL it appears that its occurrence in the meaning favere is restricted to animate Complements. It might be attractive, therefore, to assume that cupere may have two different meanings, each with its own construction. This seems even more attractive with regard to other verbs belonging to this group, such as consulere alicui (`to take care of'), consulere aliquem (de aliqua re) (`to consult') and consulere aliquid (`to deliberate about'). [13] Since in the case of metuere and cupere the difference in meaning corresponds to restrictions on the types of lexeme that may be used, we may conclude that the difference in meaning is not caused by the difference in case form, and that, therefore, there is no need to assume an `opposition' of the cases. Yet, the difference in case form is always linked to a difference in meaning, and is thus `positively motivated'. The question remains why in one instance one case is chosen and in the other instance another case. In (d) below I will argue that for the time being this question cannot yet be answered in a satisfactory way. [14]

(b) Three-place verbs with two constructions: donare

In active sentences the dative occurs within the nuclear predication to mark the obligatory third constituent with three-place verbs involving `transfer' or `communication'. In those cases, the dative constituent always has the semantic function Recipient (verbs of transfer) or Addressee (verbs of communication). [15] Some of the verbs which govern acc. + dat. also have an abl. + acc. construction. An example is donare:

(28) anulus aureus quo tu istum in contione donavisti (`The gold ring with which you endowed him during the meeting', Cic. Ver. 3.185)

(29) Gabinii, cui regna omnia Syrorum … donaras (`G., to whom you had given all the kingdoms of the Syrians', Cic. Dom. 124)

Both constructions are possible in many circumstances. Yet, this need not mean that they are interchangeable. K.–St. (I.334) claim that the ablative is necessary when `von solchen Gaben die Rede ist, bei welchen die subjektive Tätigkeit (die Annahme) der Person wesentlich ist' (`when such gifts are concerned that require the active participation of the recipient involved'). For many verbs, however, there is little evidence to substantiate this claim. For the verb donare at least one subtle difference seems to exist between the two constructions: the acc. + abl. construction is not used when e.g. a god or king is concerned; then the dat. + acc. construction is obligatory. Apparently, gods and kings cannot easily occur as Patient (in the acc. + abl. construction), but more easily as Recipient (in the dat. + acc. construction). There seems to be a semantic difference given the existence of selection restrictions. In the case of other verbs, however, only the construction of the immediately preceding sentence seems to be relevant. This means that a pragmatic difference is involved. [16]

-- 51 --

(c) Correspondence between constructions of two-and three-place verbs

A number of regularities may be mentioned with regard to the use of cases in the nuclear predication which modify the hypothesis that the cases in the nuclear predication are primarily negatively motivated. In brief, I mention the fact that for antonyms often the same case is used (dare (`to give'): adimere (`to take away') + acc. + dat.) and that many compound verbs govern the dative. In the latter case it is not clear whether the verbs are related semantically. On p. 46 it was suggested that the occurrence of the so-called double accusative with verbs such as docere might be explained on the basis of the fact that there is a corresponding two-place construction, in which the constituent expressing the `subject matter' is marked by the accusative as well. There are more indications for this type of correspondence. As is known, the ablative occurs with a number of groups of verbs, both two-place and three-place, which are related semantically. Examples are:
Two-placeThree-place
`to leave, remove oneself'`to hold back, remove'
`to lack', `to be free'`to deprive'
`to be full'`to fill', `to provide with'
The two-place verbs may be seen as the non-causative counterparts of the three-place verbs. Note that the semantically related adjectives (may) also govern the ablative. Something similar holds for a number of occurrences of the genitive in the nuclear predication. The ablative with the groups of two-place verbs mentioned above is often called `instrumenti' and `separativus'. In (e) below I will argue that arguments for this are lacking. Within the system the ablative could satisfactorily be explained as the case which – in view of the positive motivation for nominative, accusative and dative – with three-place verbs is the most obvious case for a number of semantic functions in the nuclear predication. This, in turn, may explain its occurrence with semantically related two-place verbs.

(d) The case form of the Complement with two-place verbs

We have seen above that with certain verbs, among which are certain groups of semantically related verbs, the cases marking the Complements seem to some extent to be semantically motivated. For most two-place verbs, however, it is not possible to indicate certain semantic characteristics on the basis of which they govern a case other than the accusative. It is even more difficult to determine why they govern the specific non-accusative case which makes them deviate from the `normal' pattern. Heilig (1978) has examined a large number of Latin verbs to see whether there is a connection between the case forms used for the obligatory constituents on the one hand and certain semantic

-- 52 --

properties of these verbs on the other. He arrives at a positive conclusion with regard to the opposition between verbs which govern the accusative and verbs which do not. Roughly, his conclusion is that most verbs `der ausser-subjektiven Relation' (`of non-subjective relation') govern the accusative, while most verbs `der subjektiven Befindlichkeit' (`of subjective experience') govern another case. [17] This division is a more refined version of that proposed by Chafe (1970) and is in some ways similar to the classification of states of affairs given in chapter 2. In Chafe's terminology, Heilig would say that the opposition accusative: non-accusative corresponds to the opposition between `Actions' and `Action-Processes' on the one hand and `States' and `Processes' on the other. Examples of Heilig's classification are given in table 5.6. In terms of the classification of states of affairs given in chapter 2 the Table 5.6 Examples of Heilig's (1978) classification
Actions and Action-ProcessesStates and Processes
(1) habeo(1) sum
contineodeficio
coerceoiaceo
sentio
(2) reprehendo(2) discedo
fateorexcedo
tractolabor
(3) facio(3) maereo
fingomaestus sum
tollohaereo
(4) maestus fio
adhaeresco
succedo
+ ACCUSATIVE- ACCUSATIVE
non-accusative instances should be non-controlled, the accusative instances controlled. This can be shown to be incorrect. [18] But even if Heilig's conclusions were correct, they would not explain why a specific verb governs a specific non-accusative case.

The traditional explanations, too, do not indicate the reasons why certain verbs govern certain cases, at least not in terms of common lexical characteristics of these verbs. The explanation is usually based on the alleged more or less autonomous semantic value of the case. As often, K.–St. are most explicit (and, as a consequence, in this case also most easily open to criticism). For the problem under discussion K.–St. I.253, 307 and 346–7 are the most interesting. On the last-mentioned page K.–St. first give as a common characteristic of dative and ablative the fact that they constitute a further specification of the sentence as a whole and mark something involved in the

-- 53 --

action. According to K.–St., the two cases differ in that `der Dativ ein persönliches oder persönlich gedachtes, also mit Willenskraft begabtes, der Tätigkeit des Subjektes tätig entgegentretendes, mit dem Subjekte in tätiger Wechselbeziehung stehendes, der Ablativ hingegin ein sächliches oder als Sache gedachtes, also willenloses Objekt bezeichnet'. [19] It is undoubtedly true that most of the semantic functions which can be marked by the dative are fulfilled by lexemes which indicate an animate being, and that in the semantic functions marked by ablative few lexemes occur which indicate animate beings. Problems arise with this explanation, however, in the case of the two-place verbs that govern a dative. K.–St. offer the following solution (I.307): `So werden z.B. in legibus paremus die Gesetze ebenso persönlich aufgefasst, wie in regi paremus der König.' [20] However, no independent proof is given for this statement, which, therefore, is not at all convincing.

I conclude that there is no satisfactory synchronic explanation for the specific case governed by two-place verbs (both those verbs which involve ostensible opposition – my group (a) on p. 49 – and others). There are some indications that the Romans, too, considered it an isolated, idiomatic phenomenon that a verb such as uti (`to use') governed the ablative. There are, in fact, `regular' examples of uti + acc. (not in Cicero or Caesar). Attention may also be called to the so-called personal gerundive (utendus, `having to be used'), which the `irregular' verbs share with the regular ones. [21]

(e) Relation between the use of a case in the nuclear predication and in the periphery

The current grammars, striving to assign to each case the smallest possible number (preferably one) of semantic values, equate the use of the ablative with uti and that to mark a peripheral constituent with the semantic function Instrument. Hence, in the case of uti the ablative is called `ablativus instrument;'. Several linguists have already pointed out that the semantic function fulfilled by a lexeme in a certain sentence need not necessarily be attributed to one single, constant semantic value of the case by which this lexeme is marked, since the context in which such a lexeme occurs also contains information (and the lexical meaning plays a role, too). A more upto-date formulation would be: `the ablative is used to mark both obligatory and peripheral constituents with the semantic function Instrument'. Applying this statement to sentences (30) and (31), one could say that eorum beneficio and officio et fide differ syntactically (Complement in the nuclear predication and Adjunct in the periphery, respectively), but have the same semantic function.

(30) multi eorum beneficio perverse uterentur (`Many people profit by their benefactions in a perverse way', Cic. N.D. 3.70)

(31) amici … officio et fide pariuntur (`Friends are made through devotion and faith', Sal. Jug. 10.4)

-- 54 --

This approach is actually defended by some linguists. According to the same line of reasoning, a semantic function Instrument is assumed also with complere (`to fill'). [22] Some semantic functions do, indeed, occur both in the nuclear predication and in the periphery: [23] cf., for instance, (32) and (33) (cf. p. 27), which contain an obligatory ((32)) and a peripheral ((33)) constituent with the function Direction:

(32) quo me miser conferam (Gracchus in Cic. de Orat. 3.214)

(33) quo ambulas tu? (Pl. Am. 341)

It remains to be seen whether eorum beneficio in (30) also has the semantic function Instrument. This is not the case. The ablative constituents in examples (30) and (31) differ from one another according to a number of tests. See also the behaviour of the ablative constituent with potiri (`to get possession of') in (34); here there seem to be even fewer arguments to distinguish a semantic function Instrument:

(34) (Augustus) Alexandrea … brevi potitus est (`A. took A. in a short time', Suet. Aug. 17.3)

(i) Even intuitively it seems strange to assume in (30) and (34) that the italicized constituents have the semantic function Instrument. This also appears from the impossibility to paraphrase the sentences in such a way as to give more prominence to the element of the instrument: cf. the application of the paraphrase with `to make use of' in (30'), (31') and (34'):

(30') * Many people make use of their benefactions in order to use … (?) [24]

(31') People make use of proofs of devotion and faith in order to acquire friends [25]

(34') * Augustus made use of Alexandria in order to conquer …

(ii) Alexandrea in (34) and similar nouns cannot very easily be used as (peripheral) Instrument Adjuncts. Cf. the occurrence of personal names as Complements with uti; this, too, would be impossible in the case of (peripheral) Instrument Adjuncts. [26]

(35) his Fabriciis semper est usus Oppianicus familiarissime (`With these Fabricii Oppianicus has always had a very friendly contact', Cic. Clu. 46)

In other words, there is a difference in the restrictions on the meaning of lexemes which may occur as Complements with this group of verbs, when compared to the restrictions on the group of lexemes which may occur as Adjuncts with the semantic function Instrument.

In (i) and (ii) I have listed some considerations which provide evidence against the assumption of an instrument relation between the Complement and the

-- 55 --

two-place verbs of the group uti, potiri. It is incorrect to explain the ablative with these verbs from the fact that the ablative is used to mark (peripheral) Instrument Adjuncts. Similar objections may be raised against the idea that the dative with verbs like favere should be explained by assuming that the Complement constituents have the semantic function `Beneficiary'. [27]

(f) Restrictions on the meaning of arguments and satellites in relation to the meaning of the predicate

In crosssection 3.2. I pointed out that arguments and satellites can have the same semantic function. I have also noted ( crosssection 3.4.) that in the periphery there is a relation between the semantic function and the lexical meaning of a constituent. Thus, as satellites with the semantic function `Price' (in the grammars, roughly speaking, the ablativus pretii) we will find those words which mean `purchase price'. In this section I will discuss in more detail the relation between lexical meaning and semantic function and the extent to which this relation is determined by the meaning of the predicate. In concrete terms: are there fewer or more restrictions on the meaning of arguments with a semantic function such as Price, Direction, etc. than on the meaning of satellites with this function? I will answer this question with the aid of what the current grammars say about the ablative of value and price. [28] K.-St. (I. 389-90) cite (36)-(39) as examples of the ablative of value and price:

(36) emi virginem triginta minis (`I have bought a woman for 30 minae', Pl. Curc. 344)

(37) (Caelius) conduxit in Palatio non magno domum (`On the Palatine C. rented a house for little money', Cic. Cael. 18)

(38) cum mercede doceret (`Although he taught for money', Cic. de Orat. 1.126)

(39) locavit … omnem (agrum) frumento (`He leased out all the land in exchange for corn', Liv. 27.3.1)

Most verbs with which the ablativus pretii occurs are classified under `to buy', `to sell', `to rent', `to contract out', `to bid' (liceri), but the ablative pretii also occurs with other verbs (as in (38)). With this group of verbs the Price Adjunct can be regarded as omissible. [29]

Also non-omissible constituents, e.g. the Complement with verbs meaning `to cost', are classified under the heading `ablative pretii'. Examples are (40) and (41):

(40) multo sanguine ac volneribus ea Poenis victoria stetit (`That victory cost the P. a lot of blood and injuries', Liv. 23.30.2) [30]

(41) Caesar … edocet, quanto detrimento et quot virorum fortium morte necesse sit constare victoriam (`C. sets forth how much damage and how many deaths of brave men the victory should cost', Caes. Gal. 7.19.4)

-- 56 --

The satellites in (36)-(38) indicate a `value/price' on the basis of the meaning of the nouns; also, with a verb like docere (example (38)) there can be no misunderstanding thanks to the unambiguous meaning of mercede. [31] But within the group of examples given under (36)-(39) some differences may also be pointed out. In example (39) frumento does not indicate unambiguously the `Price' by virtue of its lexical meaning. It is nevertheless interpreted as such, as a result of the meaning of the `commercial' term locare (`to lease out'). [32] Thus also the Adjuncts with emere (`to buy') enjoy a relative freedom from restrictions, though statistically the group of nouns meaning `price' predominates. [33]

(42) quid enim erat quod discessu nostro emendum putaremus? (`What was it I thought had to be bought with my departure?', Cic. Q. fr. 3.8.1)

With a verb such as docere, which does not have a `commercial' meaning, a satellite can only be interpreted as `Price' if it is expressed by a lexeme which indicates a price. Schematically, this can be illustrated as in table 5.7. With Table 5.7 Possible combinations of predication and satellite such that the interpretation `Price' is possiblea
PredicationSatellite
`commercial' noun, etc.`non-commercial' noun, etc.
`Commercial' predicate++
`Non-commercial' predicate+-
+ / -: interpretation `Price' possible/impossible
a The table is organized as a matrix of verb and noun, but a similar representation can be made for a nuclear predication and a peripheral constituent.
verbs meaning `to cost' there are no such restrictions on the meaning of the (obligatory) Complement.

From the discussion of the ablative pretii and table 5.7 we may draw the following conclusions:

- within the nuclear predication the semantic functions are determined by the predicate; therefore, relatively less severe restrictions are necessary on the meaning of lexemes which occur with the predicate in a specific semantic function;

- outside the nuclear predication highly specific lexical characteristics are required in order for a lexeme to be able to fulfil a specific semantic function. We see that for peripheral functions more cases are used than only the ablative (and prepositions). This is perhaps so because on the basis of the

-- 57 --

lexical characteristics a number of lexemes might be able to fulfil more than one semantic function in the periphery.

(g) Accusative and Infinitive (AcI)

By AcI I understand the AcI in the strict sense of the term (cf. crosssection 7.4.2).

(43) quis ignorabat Q. Pompeium fecisse foedus (`Who did not know that Q. P. had concluded a treaty', Cic. Rep. 3.28)

(44) iam tempus est me ipsum a me amari (`It is time for me to love myself', Cic. Att. 4.5.3)

crosssection 7.6.2 will be devoted to the historical explanation of the accusative as mark of the Subject of the AcI. What we are concerned with here is how the fact that the Subject constituent of the AcI is marked by the accusative fits into the synchronic case system. An essential distinguishing feature of the AcI as compared to other embedded predications and main sentences is the lack of explicit marking of the Subject: Predicate relation (agreement in Person and in most cases also Gender and Number). Also, the number of possibilities for the verb form to express tense and mood is limited. The AcI lacks formal markers such as conjunctions, which in the case of `normal', finite subordinate clauses clearly indicate the subordinate position of the clause; yet, there is a clear formal distinction between the AcI and the `main predicate'. Similarly, the clear recognizability of the Subject of the governing verb is guaranteed, since the nominative is not used to mark the Subject of the AcI.

Theoretically speaking, to mark the Subject of the AcI by a case other than the nominative, Latin could use the accusative, the dative, the genitive or the ablative. In reality, however, Latin uses the accusative. From a number of points of view this fits into the case system described (i-iii), but on the other hand, it creates some problems (iv-v). Yet, in the light of the following considerations, the accusative fits into the system better than the other cases.

(i) After the nominative the accusative is the most common case form in the nuclear predication.

(ii) The genitive occupies a clear position in the system as the marker of constituents on the noun phrase level. This distinction is of great importance (see p. 61); the ablative, too, plays a clear role in the system as the marker of satellites; note that it holds for both genitive and ablative that constituents marked by them may occur in the main sentence next to the AcI. Finally, the main function of the dative on the sentence level is to mark the third argument with a group of verbs which is easily identifiable semantically, among which are the communication verbs. It is precisely with these verbs that the AcI is rather frequent.

(iii) The AcI often fulfils the Object function in a predication; this function could also be fulfilled by a noun phrase, which would then be marked by

-- 58 --

the accusative. In many cases, therefore, the accusative is, as it were, left unused.

(iv) The remarks under (iii) do not, of course, hold for verbs such as docere, which govern a double accusative, nor for impersonal verbs, copula + adjective constructions etc., i.e. predicates with which the AcI fulfils the Subject function.

(v) A further problem is posed by the fact that within the embedded predication the formal distinction between Subject and Object (Complement) constituent disappears. How serious this problem is could only be determined on the basis of prose texts written in a not too literary Latin. In K.-St. (I.699-700) the problem is considered relatively inconsequential. 33a Tables 5.1 and 5.2 (pp. 41 and 42) show the number of instances in which the accusative occurs to mark the Subject constituent of an AcI. In only a small minority of instances will this concern a predicate with second or third arguments in the accusative. Exact data are lacking. K.-St. point out that, if the Subject constituent precedes, the presence of two accusatives does not create confusion:

(45) eo se periculo posse liberare eos (`That he could liberate them from that danger', Liv. 23.2.8)

In direct speech, too, the Subject constituent in most cases precedes the Object/Complement constituent. Furthermore, K.-St. mention passivization as a means of avoiding ambiguity (does the passive predominate indeed in cases of possible ambiguity?), while the context may also serve to shed some light on the matter. This is, for instance, the case in example (46):

(46) totiens te senatum … adisse supplicem (`That the senate has so often turned to you as a suppliant', Cic. Ver. 5.21)

5.2.5 The use of cases on the noun and adjective phrase level

In table 5.1 on p. 41 it may be seen that the genitive is pre-eminently the case used to mark constituents on the noun and adjective phrase level. This holds especially for constituents which function as Attributes to another noun phrase (type domus patris, `the house of the/my father'). Exceptions as in example (47), with an accusative Attribute hanc rem, are statistically negligible:

(47) quid tibi hanc curatiost rem? (`What have you got to do with this matter?', Pl. Am. 519)

The Head noun and the noun phrase functioning as Attribute may be related semantically in many ways. In this connection, the grammars distinguish a large number of different `genitives' (gen. materiae, possessivus, partitivus, etc.). In reality these labels are just as superfluous as those we will see below in

-- 59 --

the case of the ablative. The semantic relation is not determined by the `type of genitive', but by the meaning of the lexemes involved. [34]

In adjective phrases the genitive as case marker of the constituent depending on the adjective is less predominant. As for obligatory constituents depending on adjectives, note that there is a certain regularity in the case marking of arguments of adjectives and of semantically related verbs. Examples are:

liber/liberare + abl.

(48) animus … omni est liber cura et angore (`A mind is free from every form of worry and fear', Cic. Fin. 1.49)

memor/memini + gen.

(49) se eorum facti memorem fore (`That he would remember what they had done', Caes. Civ. 1.13.5)

This is similar to the regularity mentioned under (c) on p. 51. It is not at all surprising to find a similarity between adjectives and verbs: adjectives, too, can combine with a copula to form a Predicate and the semantic functions are the same.

Optional constituents with adjectives often have the same case that is used to express a comparable semantic relation on the sentence level. An example is the accusative with adjectives which indicate a distance or dimension (for examples see K.-St. I.282):

(50) clausi lateribus pedem altis (`Closed off by sides a foot high', Sal. Hist. fr. 4.79)

(51) negat umquam se a te … pedem discessisse (`He says that he has never moved one foot from your side', Cic. Deiot. 42)

Compare also the so-called ablative of respect (or: ablativus limitationis), of which examples can be found in K.-St. I. 392:

(52) non … tota re, sed … temporibus errasti (`You did not make a mistake with regard to the case as a whole, but in the circumstances', Cic. Phil. 2.23)

(53) quidam … voce absoni (`Certain people with a terrible voice', Cic. de Orat. 1.115)

It is, of course, due to similarities of this type that authors of grammars have decided to treat adjectives and verbs together, not paying any attention to the difference between sentence level and adjective phrase level. The genitive-case marker of the noun phrase level par excellence-is in all these cases predictably found as a competing construction, especially in poetry: [35]

(54) libera fortunae mors est (`Death knows no fortune', Luc. 7.818)

(55) alta novem pedum (`Nine feet high', Col. 8.14.1)

-- 60 --

(56) celer nandi (`A rapid swimmer', Sil. 4.585)

The fact that the genitive also occurs as a competitor in cases where there was sufficient semantic support for other case forms proves how fundamental the opposition between noun phrase level and sentence level was within the system. See also the following section and the reference to recent studies by Plank (1979a; 1979b).

5.2.6 The relation between case system and sentence structure

The question as to the function of the cases has been answered within the framework of the sentence model set forth in chapter 1. This sentence model rests on independent arguments: for instance, through such a sentence model it is explained why, on the one hand, with certain predicates certain types of constituents occur and, on the other, certain types of constituents cannot occur with certain predicates. This approach is based, on the one hand, on the assumption that the content of the sentence is determined on the basis of the lexical meaning(s) of the predicate and the arguments occurring with it (in certain semantic functions), and, on the other hand, on the basis of the meaning of the satellites. Case, word order and the like serve formally to structure this content, naturally according to certain rules. [36] The function of the cases to mark arguments is more important, as a certain predicate can govern more arguments of the same semantic class: for the correct interpretation of (57) it is less important that the Subject is marked by the nominative and the Object by the accusative than in (58):

(57) is illius laudare infit formam virginis (`He begins to praise the beauty of that girl', Pl. Rud. 51)

(58) laudat Africanum Panaetius, quod … (`P. praises A., because …', Cic. Off. 2.76)

After all, forma virginis is unable to laudare a person. [37] In (58), on the other hand, in principle both A. and P. could praise or be praised. I have systematically distinguished between sentence level and noun phrase level above. On the noun phrase level the relation between noun phrases is unequivocally marked by the genitive.

It is exactly by treating the function of the cases in connection with the syntactic and semantic structure within which they function that we are able to create an overall picture of the entire system which is both more balanced and more transparent than that provided by the current grammars (see p. 63). By assigning an important role to sentence structure and lexical meaning, we may also understand more easily why with a small number of cases Latin is nevertheless able to mark a multitude of semantic relations. This approach is supported by, among other things, (a) research on the possible number of ambiguities in coherent texts and (b) studies of diachronic developments.

-- 61 --

(a) Research on possible ambiguities (which requires further elaboration) was done in the following way: in a number of texts [38] all noun phrases were examined as if no information concerning the case form was available; however, the information concerning gender and number was taken to be known. [39] The information at the reader's disposal either on the basis of the preceding context or on the basis of his or her general knowledge was left out of account, as was any information available on the basis of the word order and the occurrence of a case form in prepositional phrases.

The results described below do need some further clarification, but the outcomes are interesting enough. In by far the greatest number of instances the information conveyed by the case form does not appear to be strictly necessary, something which may first of all be explained on the basis of the fact that the meaning of the predicate greatly restricts the choice of the arguments. A second factor is the fact that in a large number of sentences the very expression of number (singular and plural) on both the noun phrases and the (main) verb (agreement) identifies the Subject and, consequently, also the Object. [40] In periphrastic verb forms the expression of gender does, of course, play a role. Furthermore, confusion is avoided because there is a small number of satellites per sentence or clause, something which has to do with the fact that satellites very often contain new or salient information. Apparently, collocation of a number of such constituents in one sentence is avoided. Finally, nominal lexemes are more or less fit for certain functions on the basis of their meaning.

Case marking is indispensable under the following circumstances:

(i) Confusion as to the semantic and syntactic function of a constituent is possible when a predicate allows the occurrence of two or more semantically similar noun phrases, e.g. two noun phrases referring to (in) animate beings:

(59) nunc iam nulla viro iuranti femina credat (`Let from now on no woman believe a man who swears an oath', Catul. 64.143)

(60) sancte puer, curis hominum qui gaudia misces (`Noble young man, you who mix the worries of men with joy', Catul. 64.95)

(61) ulmisque adiungere vites (`To wed vines to elms', Verg. G. 1.2.)

In Catullus this phenomenon occurs more often than in Caesar, something undoubtedly connected with the subject matter.

(ii) Especially in the more complex sentences of Caesar the lack of case marking would cause the distinction between word group level and sentence level to disappear:

(62) trium mensum molita cibaria sibi quemque domo efferre iubent (`They ordered everyone to take from home with them corn for three months', Caes. Gal. 1.5.3)

-- 62 --

(63) viderunt … mortales oculis … Nymphas (`Mortals saw with their eyes the Nymphs', Catul. 64.16)

(64) alius … funda … verberat amnem (`Some lash the stream with a casting-net', Verg. G. 1.141)

(iii) Since in Latin the Subject, if known from context or situation, is often not expressed explicitly, the absence of case marking would cause the difference between ablative absolute and the so-called predicative or adverbial participle (`participium coniunctum') to disappear (another shift from one level to another, therefore), so that, as a result, it is no longer clear which constituent is modified by the participle:

(65) Helvetii ea spe deiecti … operis munitione et militum concursu repulsi hoc conatu destiterunt (`The Helvetians, having lost hope for this, were warded off by the fortifications and the rush of soldiers and the spears thrown at them, and desisted from their attempt', Caes. Gal. 1.8.4)

In fact, however, the wider context excludes any ambiguity.

(iv) In the AcI, where non-periphrastic verb forms show no agreement in number, confusion may arise.

In summary, it may be said at this point that the marking of the syntactic and semantic function by means of a case is really necessary in less that 5–10% of instances. From this it cannot, however, be concluded that case marking is an insignificant element in language use. It is a well-known phenomenon that linguistic structures contain more information than is strictly necessary (in technical terms, we may speak of `redundancy'). Among other things, this serves to guarantee successful communication in case of impaired intelligibility, reduced attention, etc., or – in the case of written communication – to compensate for the lack of non-verbal means of communication. [41] We can, however, deduce from this conclusion that case marking is merely one of the building blocks, one much less important for successful communication than e.g. the meaning of the lexemes with which the sentences are built. [42]

(b) A second element to support the approach chosen here is to be found in the diachronic developments. In the development from classical Latin to Old French the systematic distinction between sentence level and word group level is apparently of great importance, as appears from the fact that the genitive continues to exist for a long time. Furthermore, the distinction between nominative and accusative is very persistent, especially for lexemes indicating animate beings. For these are able to occur in a multitude of semantic relations with regard to predicates, e.g. as Agent and Patient. This is in accordance with the mainly distinctive function of cases within the nuclear predication, as discussed above.

-- 63 --

The approach chosen here differs considerably from the usual descriptions. In the grammars the cases are assigned a much more `directive' role than I have done above. Consequently, such grammars successively treat the various cases in all their uses, which are provided, therefore, with semantic labels. With the exception of the genitive, no systematic distinction is made according to the level (sentence level/noun phrase level; nuclear predication/periphery) on which a certain use is found. Generally, for all the various uses of one case a common semantic denominator is determined, sometimes (especially in the case of the ablative) a number of common denominators. These common denominators are either seen as historical predecessors from which the various uses have developed, or as a synchronically relevant common semantic value (in German grammars the term `Grundbedeutung' is used). See as an illustration of K.–St.'s approach table 5.8; this type of treatment is also found in many school grammars. Such an approach does not take into account the distinction between levels, as appears from the enumeration of examples of the ablativus causae in table 5.9. Table 5.8 The ablative in K.–St. (I.346–412)
Grundbedeutung `basic semantic value'Use (ablativus …)
Vertreter des Lokativsloci
`representing the locative'temporis
als eigentlicherseparativus
Ablativoriginis
`ablative proper'comparationis
Vertreter desinstrumenti
Instrumentalispretii
`representing the instrumental'limitationis materiae causae mensurae sociativus modi qualitatis
Table 5.9 Ablativus causae in K.–St. (I.394-401)
lacrumo gaudio (`I am crying with joy')omissible; sentence level
aetate hoc corpus putret (`Owing to old age this body is rotting')omissible; sentence level
delectari imperio (`To enjoy power')obligatory; sentence level
desiderio maestus (`Sad with desire')omissible; adjective phrase level
humanitate mihi praestat (`He is superior to me in education')obligatory; sentence level
summa laude dignus (`Deserving of the greatest praise')obligatory; adjective phrase level

-- 64 --

From table 5.8 it may be deduced that the labels given to the various uses (and thus also the distinction between the uses) are often superfluous:
ablativus originis:parentibus natus humilibus (`Born from simple parents'): the `origin' is inherent in the lexical meaning of natus.
ablativus materiae:eadem materia fiunt (`They are made of the same material'): the `material' is inherent in the lexical meaning of the noun materia.
Cf. also above ( crosssection 5.2.4(f)) about the `ablativus pretii'. [43]

The main objections against K.–St.'s approach, which I briefly described above, are that the common labels assigned to the uses of the cases for constituents in the nuclear predication and in the periphery at times seem highly artificial, and that no arguments are given for assuming one and the same semantic value in the nuclear predication and in the periphery. This is exemplified by the treatment of the so-called ablativus instrumenti, with which I have already dealt above ( crosssection 5.2.4(e)). The intriguing question is, of course, why the same case form occurs both in the nuclear predication and in the periphery. There are at least two explanations: (i) diachronic developments; (ii) economy: the system avoids using more case form distinctions than it needs. There are also other, more important, instruments which serve to preserve coherence between constituents, e.g. the meaning of verbs and nouns.

In the traditional description the matter is – to put it bluntly – presented upside down. The semantic functions of the lexemes are seen as dependent on the case forms and the lexical meaning of the nominal lexemes and of the predicate are considered less important. An explanation for this may be sought in, among other things, the following five points:

- implicitly or explicitly the principle `one form, one meaning' (and vice versa) was taken for granted, a principle which does no justice to reality; [44]

- morphological characteristics were examined in isolation, without taking into account syntax and semantics;

- a need was felt for a systematic approach of the various developments of the Indo-European languages. This in its turn resulted in the assumption of very vague common semantic values; [45]

- most people raised in the context of a modern Western European language have some trouble finding their way in a case system. The extensive treatment of the cases in the grammars is, therefore, also based on didactic considerations;

- the scope of classical linguists was limited; they had little knowledge of non-Indo-European languages with sometimes much more complicated case systems.

-- 65 --

5.2.7 Conclusion

We may conclude from the preceding discussion that the cases ought to be assigned a considerably less essential function than is sometimes suggested in scholarly grammars and often in didactic grammars. The meanings of the verb and nouns involved play a much more important role in the sentence. This relative lack of importance explains in part why the structures of the Romance languages could develop (practically) without cases, without communication problems of any magnitude arising. What changed was the system of formal marking of the semantic structure, which itself was largely based on the meanings of predicates and noun phrases. The case system has given way to another system of marking the syntactic and semantic relations, a system in which particularly word order and prepositions play an important role.

5.3. Prepositions

In this section I deal briefly with the following points: the distribution of prepositions ( crosssection 5.3.1), the (historical) relation between prepositions and cases ( crosssection 5.3.2), the idiomatic use of prepositions ( crosssection 5.3.3.) and the internal structure of prepositional phrases ( crosssection 5.3.4).

5.3.1 Distribution of prepositions

From table 5.2 on p. 42 we may deduce the following relation in the underlying corpus between the use of prepositions (and the case form they govern) to mark noun phrases and the use of oblique cases:
- Prep.+ Prep.
Obj./Compl. two-place verbs14888
Compl. three-place verbs4748
Satellites88119
On the noun phrase level nominal constituents are, of course, much less often marked by a preposition than by the genitive, but still slightly more frequently than by dative and ablative taken together. In prose prepositions are used more often than in poetry (table 5.3 on p. 42).

To illustrate the distribution of prepositions indicated above I give some examples of the preposition de on each of the various levels.

(a) de as a marker of satellites

(66) de vehiculo … dicebat (`He spoke from a cart', Nepos Tim. 4.2)

-- 66 --

(67) non bonust somnus de prandio (`It is not healthy to sleep immediately after a meal', Pl. Mos. 697)

(68) (Regulus) de captivis commutandis Romam missus esset (`R. had been sent to Rome in order to speak about the exchange of prisoners', Cic. Off. 1.39)

(69) de talento nulla causa est quin feras (`As for this talent, there is no reason why you should not take it', Pl. Rud. 1397)

(b) de as a marker of Complement constituents

(70) qui … de veneficiis accusabant (`Those who made accusations of poisoning', Cic. S. Rosc. 90)

(71) de Hortensio te certo scio dolere (`I know for sure that you are sad about H.', Cic. Att. 6.6.2)

(72) cum … senatus de annona haberetur (`When there were meetings of the senate about the corn supply', Cic. Att. 4.1.6) [46]

(73) ut de istius facto dubium esse nemini possit (`In order that nobody may have any doubt as to his deed', Cic. Ver. 4.91)

(74) his de rebus conscium esse Pisonem (`That P. was an accomplice in these deeds', Cic. Att. 2.24.3)

(c) de as a marker of constituents on the noun phrase level (governed by a noun phrase)

(75) horrifico lapsu de montibus adsunt (`With a frightening swoop from the mountains they are upon us', Verg. A. 3.225)

(76) nullus umquam de Sulla nuntius ad me (pervenit) (`No message concerning S. ever reached me', Cic. Sul. 14)

(77) ut aliquam partem de istius impudentia reticere possim (`That I can keep to myself some part of his impudence', Cic. Ver. 1.32)

(78) de collegio quis tandem adfuit? (`Who was there, then, of the board?', Cic. Dom. 117)

(79) de tribus enim quae proposui hoc extremum est (`For of the three things I have proposed this is the last one', Cic. Phil. 7.21)

5.3.2 The relation between cases and prepositions

It is commonly assumed that in Indo-European languages prepositions (and postpositions) have developed from adverbs which were added to specify the information given by the cases. [47] The development from adverbs to prepositions may be illustrated schematically as follows:

-- 67 --

In many languages examples may be found of the development of prepositions from adverbs. Assuming this development does not necessarily mean – as is often assumed – that at one stage there were no prepositions or postpositions and only cases. The main objection against this assumption is that among the languages known to us there are hardly any examples of languages without pre- or postpositions and with cases. Even languages with highly complex case systems – such as Hungarian and Tabassaran- have prepositions. In other words, the postulated structure is typologically unattractive. It is, on the other hand, not unlikely that in a preliminary stage of Latin cases were more important in marking the relation between constituents than were prepositions. [48]

For (Classical) Latin prepositions are often seen as a means of specifying the alleged semantic value of the cases. In support of this view linguists often refer to two phenomena, viz. (i) the existence of prepositions which govern two cases with a semantic difference, and (ii) the occurrence of verbs with which the preposition seems optional (liberare).

(i) With the prepositions in and sub we find two cases:

- venire in + acc. (urbem) indicates direction (`to come into the city')

- habitare in + abl. (urbe) indicates location (`to live in the city')

Many scholars explain this synchronic phenomenon by taking the preposition as a specification of a semantic difference whose essence is already inherent in the case forms involved. This explanation does not, however, hold for intra.

- venire intra + acc. (urbem) indicates direction (`to come within the city')

- habitare intra + acc. (urbem) indicates location (`to live within the city')

Cf. also ante murum and pro muro (`before the wall'). [49]

(ii) With a number of three-place verbs we find Complement constituents marked either by the ablative or by a preposition + ablative. [50] Cf. examples (80) and (81):

(80) te a quartana liberatum gaudeo (`I am glad that you have recovered from the fever, Cic. Att. 10.15.4)

(81) febri quartana liberatus est (`He has recovered from the fever, Plin. Nat. 7.166)

Instances of this type are quite often considered to offer proof for the general hypothesis that prepositions specify semantic relations already given in the

-- 68 --

case form itself. In the preceding section (pp. 53–4) I have already proposed that it is not very useful to say that the `separative' relation present in example (81) is localised in the case form, since the separative relation is inherent in the meaning of liberare. Nor is it useful to take a(b) in example (80) as a `specification' of the semantic value of the case. The two expressions, rather, seem to be synonymous, [51] the meaning of the preposition a(b) corresponding exactly to the semantic relation expressed at any rate by liberare. In this connection it is remarkable that we find hardly any other prepositions which indicate a slightly different kind of specification of the `separative' relation: the construction with ab has to a high degree become a set phrase. [52] Rather than drawing from examples (80) and (81) the general conclusion that prepositions are specifications of relations marked in themselves by the case, it is much more interesting to observe that verbs like liberare apparently allow the `separative relation' to be implicit or explicit. In both cases the noun phrase happens to be marked by the ablative. [53] Conversely, there are examples of preposition + case X competing with a different case Y. Cf.

(77) aliquam partem de istius impudentia

(82) ne residere in te ullam partem iracundiae suspicemur (`Lest we suspect the presence in your heart of some element of anger', Cic. Deiot. 8)

The genitive is, as we have seen, the pre-eminent case for the noun phrase level; the same formal marking allows a large number of semantic relations, according to the lexical meanings of Head and Attribute. The so-called `partitive' relation present in the examples cited here is made explicit by the preposition de (+ abl.). [54] Cf. also the competition on the sentence level for marking of the Addressee between the dative and ad + acc.: [55]

(83) tibi … praetor … palam dicit (`The praetor openly tells you', Cic. Quinct. 85)

(84) ad eos (deos) is deus … fatur (`To those gods this god says', Cic. Tim. 40; translated from the Greek)

Generally speaking, we may say that in a number of instances in which the semantic relation is sufficiently clear on the basis of the lexical meaning(s) there was a choice between a so-called `analytical' expression with the aid of a semantically related preposition + the case governed by this preposition and a so-called `synthetic' expression marked only by a case form. Sometimes, e.g. in the case of liberare, the case forms in the `analytical' and the `synthetical' expressions are identical. As is known, the analytical expression has been generalized in the development of the Romance languages. [56]

5.3.3 Idiomatic use of prepositions

At the end of the preceding section I have stressed that prepositions differ from cases in that they explicitly express the semantic relation. It is, however, not

-- 69 --

always easy to determine the exact semantic aspect of a preposition. Particularly in the case of prepositions used to mark arguments this is sometimes difficult or even impossible. Examples are bene mereri de (`to render a service to'), pertinere ad (`to pertain to'), communicare cum (`to share with'). This phenomenon is, of course, also common in other languages; cf. English John goes into the house and John goes into the problem. In the case of a so-called prepositional Object (with to go into) the meaning of into cannot be described very easily, and certainly cannot synchronically be related very easily to into in into the house. Note, also, that precisely the more frequent prepositions have a rather broad spectrum of diverging semantic aspects. To illustrate this I give the survey of the article on de in TLL:

conspectus materiae:

I de loco:

A proprie:

1 ab superiore loco p. 44.25

2 unde aliquid vel aliquis exit p. 46.37

B translate et in imag.:

1 de deductione p. 50.33

2 de origine p. 54.23

3 de parte deducta p. 56.7

4 de materia vel condicione mutata p. 59.49

5 de impensis p. 61.13

6 de poena sumpta p. 61.84

7 pro abl. instrum. p. 62.18

8 pro abl. comparationis p. 64.34

II de tempore p. 64.51

III de causa p. 65.47

IV de relatione:

A i.q. p. 67.25

B secundum p. 77.75

V adverbialia p. 78.84

VI varia et singularia p. 80.1

(from: TLL s.v. de, p. 44)

From this multitude of semantic aspects alone it can be seen that the characterization of prepositions as `specifications' of cases discussed in crosssection 5.3.2 is too simple. Some of the various semantic aspects may be characterized as `concrete' or `local', others may rather be called `abstract'. Sometimes there are direct relations between different semantic aspects (e.g. local `from' and temporal `immediately after'). Sometimes such relations are of a more artificial nature (e.g. between local de and the so-called causal de (cf. fessus de via `tired owing to the trip')). Sometimes one wonders whether it is at all useful to distinguish a meaning, e.g. because the meaning of the governed noun

-- 70 --

phrase is decisive (temporal de, for instance, only occurs with words indicating a `Time Position'). Consequently, the problems discussed in crosssection 5.2.6 with regard to the multitude of case uses to some extent also occur with regard to prepositions.

5.3.4 The internal structure of prepositional phrases

Prepositional phrases (PP) consist of a preposition and as a rule a noun phrase. [57] With regard to each other they fulfil the functions Relator and Prepositional Complement, respectively (see figure 5.1). [58]

Figure 5.1

In terms of their internal structure prepositional phrases differ from noun phrases, e.g. in that preposition and `preposition Complement' are mutually dependent. On the one hand, in an example like (66) de cannot be omitted without causing the sentence to become ungrammatical.

(66) de vehiculo … dicebat

This rule does not hold for verbs like liberare, where an argument slot may be filled either by a noun phrase or by a prepositional phrase (see p. 67). On the other hand, as a rule a preposition cannot independently replace an entire prepositional phrase, a phenomenon which we will see with regard to so-called substantivised adjectives (p. 88). A prepositional Complement can only be omitted in cases such as (85):

(85) et in corpore et extra esse quaedam bona (`Some bodily and external things are good', Cic. Fin. 2.68) (see K.–St. I. 579)

Extra does not only occur as a preposition, but also as an adverb. Its normal use, however, is as exemplified in (86):

(86) in urbe et extra eam (`Within the city and outside it')

When two prepositional phrases governing different cases are coordinated, both prepositions require an explicit Complement. [59]

-- 71 --

5.3.5 Conclusion

Cases and prepositions (+ cases) are complementary systems which each serve to mark the relations of arguments and satellites or parts of noun phrases. Prepositions generally (but not always) have a semantic aspect of their own; only in certain structures, if at all, can cases be assigned a general semantic value.

5.4. Subordinators

In the introduction to this chapter we saw that subordinators occur both in clauses which are arguments of the (main) predicate and in clauses which are satellites. A number of subordinators are found in both types of clause, cf. examples (3) and (6), cited above:

(3) nihil mihi optatius cadere posse quam ut (`that nothing could be more desirable than that', Cic. Att. 3.1) note

(6) esse oportet ut vivas, non vivere ut edas (`One ought to eat in order to live, not live in order to eat', Rhet. Her. 4.39.4)

As we have also seen with reference to cases and prepositions, it is difficult to assign a specific semantic aspect to a subordinator in an argument clause. This appears, among other things, from the fact that it is impossible to change the meaning by substituting another subordinator. In (3) the use of ut could, for instance, be compared with that of  quod in (87), with a subtle difference in meaning, but where for quod it would be difficult to determine a specific semantic aspect. [60]

(87) nihil mihi … gratius cecidisse quam quod Tulliam meam suavissime … coluisti (`that there is nothing I have valued more than your tender attention to my Tullia', Cic. Att. 10.8.9)

60a It is, on the other hand, possible to contrast ut in satellite clauses with another subordinator, e.g. quod (`because'), cf. (54) on p. 118. It is difficult to indicate the relation between the uses of one and the same subordinator in argument and satellite clauses. Two separate points will be dealt with later: in crosssection 7.3.3 I will argue that the categories of subordinators and relative adverbs are not easily distinguishable. In crosssection 7.6.1 I will address the fact that in some cases there is a historical relation between subordinators on the one hand and connectors (e.g. enim) and adverbs (e.g. ideo) on the other.

5.5. Agreement (within noun phrases)

In Latin nouns and pronouns are marked not only for case but also for number (Numerus). The number marking in principle corresponds to the number of entities referred to (e.g. unus filius/duo filii); exceptions are, for

-- 72 --

instance: `pluralia tantum': e.g. tenebrae `darkness', and `singularia tantum': e.g. lac `milk'. [61] Number is, therefore, a semantic category.

Generally, nouns are not marked for the category of (grammatical) gender (Genus). Exceptions are certain nouns which indicate animate beings, such as filius/filia; productive suffixes are e.g.-tor/-trix. There is extensive literature on the nature and origin of the grammatical gender of nouns. Most linguists agree that, in spite of the existence of cases like -tor/-trix and many rules in Latin of the type `names of trees are feminine', there is no relation between natural gender (sex) and grammatical gender, [62] and, more generally, that there is no semantic basis for the gender of nouns. [63] Much more remarkable is, for instance, the relation between declension type and gender (e.g. `3rd-declension words ending in -s are generally feminine').

Agreement of adjectives, pronouns, etc. consists in the visibility of explicit (case, number) and implicit characteristics of the Head nouns in the form of these adjectives, pronouns, etc. Agreement of adjectives is a grammatical phenomenon which serves to show the coherence within noun phrases, even if such noun phrases are discontinuous.

Bibliographical information

My views on the case system may be found in Pinkster (1972a; 1980; 1985b). Bibliographical surveys of the enormous literature are provided by Calboli (1972; 1983) and Serbat (1981). Statistical data are to be found in Bolkestein et al. (1976; 1978). For the discriminating and identifying function of cases see Dik (1978: 157–70) and Harris (1975: 186). For the possibility of using one case for different functions (both discriminating and identifying) see Ebeling (1957), de Groot (1956a; 1956b) and Kurylowicz (1949). For contextual factors which determine the choice of constructions with verbs such as donare see Bolkestein (1985) and Bolkestein & Risselada (1985). For an explanation of the use of the accusative as the Subject of the AcI see Baldi (1983). The most extensive description of the genitive may be found in de Groot (1956b). For the development of the Latin case system see Plank (1979a, 1979b). [add. 12.23: A structural approach to the semantic values of the cases can be found in Echarte (1991b). A recent survey of studies on the Latin case system is Viparelli (1993).]

For the relation between prepositions and cases see Pinkster (1972c: 145–52). The diachronic approach may be found in Sz. 214–15. For the multiple uses of prepositions see Kooij (1971: 57–115) and Vester (1983).

On the function of grammatical gender Gerlach Royen (1929) offers the most extensive information. See also Ibrahim (1973).

-- 73 --

Previous Section

Next Section


Pinkster, Harm (1942-) [1990], Latin Syntax and Semantics [info], xii, 320 p.: ill.; 24 cm. [word count] [Pinkster].
Powered by PhiloLogic