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

Next Section


1 Introduction

Note (end) return to page 1 For a definition of the notion `sentence' see Matthews (1981: 26ff). See also note 8.

Note (end) return to page 2 For the notion `lexeme' see Matthews (1981: 51).

Note (end) return to page 3 From now on I will avoid expressions like `constituents functioning as Predicate' and briefly call them `predicates'. Notice that the notion `predicate' is used in this book in a restrictive sense if compared to what is common in British and American Latin grammars. It does, for example, not cover the italicized items in the following sentences:
- John is a soldier/brilliant (for these I will use the term `Subject Complement', following Quirk et al. 1985: 55)
- He came home sadder and wiser (I will call such constituents `Praedicativum', discussed in chapter 8).

Note (end) return to page 4 I consider rex esse as a whole, primarily because it is comparable to the verb regnare. The copula esse itself does not contribute to the content of the sentence and is best regarded as a formal means to express tense, mood and number (by the ending of the verb). It has to be distinguished from:
(a) the two-place verb esse = `to be somewhere', as in ut … vos istic commodissime sperem esse (`that you will be extremely comfortable there', Cic. Fam. 14.7.2)
(b) the auxiliary esse, as in laudatus est (`he has been praised')
(c) identifying esse, as in (cum) Pylades Orestem se esse diceret (`when Pylades said he was Orestes', Cic. Amic. 24).
[added 12-08: There is still another type of esse, namely 'existential' esse. Marouzeau (1910) is still indispensable.]
It is not easy to distinguish these types of esse formally. Further research is required to establish the difference between copulative and identifying esse.

Note (end) return to page 5 `Predicate' (with a capital P) refers to the syntactic function, `predicate' refers to a number of lexical categories (nouns, verbs, adjectives, in particular). In general, labels starting with a capital will denote semantic, syntactic and pragmatic functions (thus Subject).

Note (end) return to page 6 The notion `omissibility' will be defined more precisely in crosssection 2.1.

Note (end) return to page 7 Unlike Happ (1976) and many linguists who regard a verb like `to buy' as four-place (someone, something, from someone, at a certain price) I do not recognize the existence of four-place predicates in Latin (see Bolkestein 1977a). This may, however, be feasible for certain other languages (see Bossong 1980: 360). I do not accept five-place verbs either. For zero-place verbs like pluit (`it rains') see. p. 23.

Note (end) return to page 7a [added 12-08:
Nomina Agentis derived from verbal stems correspond with controllable States of Affairs and usually refer to human beings (Torrego 1993). She regards lexemes like amator and osor as Action nouns and also obsessor in the ex. below as humoristic play on the Action meaning. Her statements are in contrast to those of Fruyt (1990: 61) who states that verbal nouns in -tor denote the first argument of the verb whatever its semantic subclass, referring to the same sets of examples.
- Sed cur sedebas in foro ... Tu solus ...? ...
# Hoc ego fui hodie olus obsessor fori (Pl. Ps. 800-7).]

Note (end) return to page 8 The information contained in figure 1.1, which represents the structure of the predication, is much less than that contained in the actual sentence (ex. 7). The figure does not show, for example, that the sentence is an interrogative sentence

-- 260 --

(instead of a declarative sentence), nor does it contain information about tense and other elements. In this book the notion `predication' will be reserved for describing the structural relationships between predicates, arguments and satellites in sentences and (main and subordinate) clauses.

Note (end) return to page 9 For focalizing particles see Panhuis (1982: 52; 89–92; he calls them `rhematizers').

Note (end) return to page 10 Other sections of Latin syntax where pragmatic information is highly relevant are the nominative and infinitive construction (see p. 130), the active/passive distinction (see chapter 12) and the expression of Subject pronouns (see Pinkster 1986b; 1987b).

Note (end) return to page 11 For the different meanings attached to the notion `Topic' see Chafe (1976) and Dooley (1982). Dik (1989: ch. 2) has still another pragmatic function, viz. that of `Tail' for cases like `He is a nice chap, your brother'. I omit these constituents from the present account.

2 Nuclear predication

Note (end) return to page 1 The Thesaurus Linguae Latinae is the most complete dictionary of Latin. With the exception of the letter N the work has proceeded far into the P. The Packard Humanities Institute now offers the complete texts of most classical authors as well as other texts on CD Rom.

Note (end) return to page 2 Context types are discussed in Happ (1976: 239–61). Critical comments can be found in Bolkestein (1977a). In many of Happ's examples, like (2) and (3), it is not so much the context that explains the absence of arguments as the difference between `specific&rsquo and `non-specific&rsquo, etc. (for which see below, p. 9).

Note (end) return to page 3 One might analyse the construction in a different way, taking infesta reddere as a causative correlate to infesta esse (itinera), which would mean taking it as a two-place predicate (see. p. 2).

Note (end) return to page 4 I assume that with the meaning `to speak' a Manner expression is obligatory. For (obligatory) arguments with the semantic function `Manner' see also crosssection 3.2. on p. 27.

Note (end) return to page 5 Notice that in (11) and (13) no Addressee can be added.

Note (end) return to page 6 The examples come from Stein (1979: 126–9).

Note (end) return to page 7 Lehmann (1988a: 42) describes the cognate object in the context of `exteriorization of participants'. He draws attention to the restrictions that hold in many languages, for example a certain restriction on passivization (which is lacking in Latin, however).
[added 12-08: Poets frequently use basically intransitive verbs with a direct Object instead of a construction with a satellite. Examples from Vergil can be found in Görler (1982; 1985: 266), for example
- currere aequor (Verg. A. 3.191)]

Note (end) return to page 7a [added 12-08: The use of facere as a proverb is briefly mentioned in López Moreda (1987: 76-77).]

Note (end) return to page 8 However, see Keenan (1976) for the confusion that exists even with regard to the question of what constitutes a Subject.

Note (end) return to page 9 See Heberlein (1989: 70) on the `indifference' of the gerundive to voice (active/passive).

Note (end) return to page 10 The notion `control' will be discussed in more detail in the next section.

Note (end) return to page 11 In (42) and (43) the Agent is implied in the verb form. The fact that the Agent controls the action follows from the imperative verb form (see. p. 17).

Note (end) return to page 12 On habitare see crosssection 3.2. on p. 28. I use Cause instead of Force in Dik (1978).

Note (end) return to page 13 A recent critical discussion of Dik's semantic functions can be found in Somers (1987: 94–108).

Note (end) return to page 14 Forces of nature resemble human beings in many respects, especially in poetry.

Note (end) return to page 15 Dik alos makes a distinction `± Experience' to differentiate `normal' activities etc. from `psychic ones' (e.g. believe). I leave this out of account here. It should be noted, however, that the categories used in this book can never explain all the variation that exists in natural language, let alone in reality: linguistic description implies simplification, unfortunately.

-- 261 --

Note (end) return to page 16 In expressions which are not exhortations or orders, properly speaking, but rather wishes, maledictions and curses, non-controllable predications may occur, such as sleep well, get to sleep, go to the devil, be damned. Examples and discussion in Bolkestein (1980a: ch. 6). Passive imperative verb forms are extremely rare in Latin, and quite understandably so (Bergh 1975).
[added 12-08: Bergh (1990) rejects the ungrammaticality of passive imperatives, referring to embedded orders as in
patere te vinci consilio (Liv. 6.23.8)
- nummum a me accipe: iube te piari de mea pecunia (Pl. Men. 290-291)
- in his ergo venite et curamini (Lucas 13.14)
Bergh also refers to Löfstedt (1966: 83ff) for instances of coordination (Verg. A. 8.39; Plin. Ep. 7.3.3. terrere; Paneg. 60.2 adire) and a few uncoordinated instances (Lucan. 9.982; Apul. Met. 11.29.5).
Torrego (1993) suggests that the derivation of nomina agentis is restricted to controllable SoA's, which would give a new criterion for establishing SoA.]

Note (end) return to page 17 `Nuclear predication' is used in a loose sense for `state of affairs, which is denoted by the nuclear predication'.

Note (end) return to page 18 For this function of the passive see Flobert (1975: 564), Comrie (1977), and Siewierska (1984) [added 12-08: Letoublon & Maurel (1985)]. Apart from inflexional means like active/passive Latin has `derivative' means as well to express different states of affairs like the `infix' -sc-: ardere (`to burn') vs. ardescere (`to take fire'). For details see Sz. (287–300). Furthermore, there are cases like: tradux bimus onerat vetustate (`cross shoots make too heavy a weight when they grow old', cf. Plin. Nat. 17.211). The normal use of onerare is as a two-place or three-place predicate. Compare expressions like English `This book reads easily' (for examples see Flobert 1975: 408–9; Feltenius 1977; Sánchez Salor 1981). Moreover, passive verb forms are also used in a so-called `mediopassive' way, where other languages use reflexive expressions (movetur: `he moves (himself)'). In fact, Latin too has a large number of reflexive expressions more or less synonymous with the mediopassive ones (e.g. se movere; further examples can be found in Flobert 1975: 384–6). A survey of `deagentization' devices in Latin can be found in Tesařova-Nováková (1988a; 1988b). See also Beneš (1971). [added 12-08: Valency alteration is quite frequent in poetry. Görler (1985: 272) notes especially the use of intransitive verbs instead of transitive reflexive expressions, as in venti posuere (Verg. A. 7.27).
In Latin 'argument incorporation' as in animadvertere is rare (Fugier 1994, Sz. 46)]

Note (end) return to page 19 In these examples I give only active forms of those predicates that may occur in the passive.

Note (end) return to page 19a [added 12-08: It is quite common from Plautus through Celsus to Ammianus Marcellinus not to express the second argument explicitly if it is continuous within its discourse. There are no signs in these authors of the increase of pronominal reference (by is, for example), as is obligatory in the Romance languages for second arguments (Mulder 1991). ]

Note (end) return to page 20 Exceptional in this respect is the construction with the so-called retained accusative (often called: Accusativus Graecus) like Verg. A. 3.81 sacra redimitus tempora lauro (`his brows bound with hallowed laurel', corresponding to active alicui tempora redimire with a so-called sympathetic dative). For examples see Flobert (1975: 485–94).
[added 12-08: The dativus sympatheticus is sometimes regarded as a synonymous variant of the genitivus possessivus, which has undergone `possessor ascension' and has become a distinct participant in the SoA. Löfstedt (1956) cites pairs like the following, which support this view:
- cui caput erigere si volueris (Mul. 316)
- cuius caput si erigere volueris (Veg. Mul. 2.88.1)
Herslund (1988) uses the label 'datif partitif', Schmid (1988) the label 'Pertinenzdativ'.]

Note (end) return to page 20a [added 12-08: A typology of impersonal constructions can be found in Rosén (H.B.) (1992a).
For the evolution of certain impersonal and intransitive constructions see Garcia Hernandez (1992) and Luque Moreno (1978).
For the functions of the impersonal passive see Letoublon (1991), Maurel (1991), Mortureux (1991), Pinkster (1992). For ancient terminology see Desbordes (1991). ]

Note (end) return to page 21 From antiquity onwards people have assumed some sort of implied Subject with impersonal expressions like pluit, for example Iuppiter. Flobert (1975: 557–9) suggests pluvia (`rain') as an underlying Subject. The religious aspects of these expressions are discussed in Le Bourdelles (1966). General discussion in Seefranz-Montag (1983).
[added 12-08: For weather verbs in Latin see also De Carvalho (1991). Weather verbs have been the subject of intense debate between various theoretical models. See Ruwet (1989) for discussion. A thorough analysis of impersonal verbs, including weather verbs, in German is Corrodi (1925). In the case of German regnen he makes a threefold distinction (1925: 28-29): (i) es regnet as a mere event, where regnen applies to the `situation'; (ii) die Wolken regnen, etc. with reference to some causee; (iii) Blut regnet, with reference to some material coming down in the form of rain. The latter two usages are under Latin and/or biblical influence. It makes no sense to derive the first type from the other ones.]

Note (end) return to page 22 The normal expression in the Accusative and Infinitive construction is lapidibus pluisse (lit.: `it rained with stones'). By way of exception we do find a Subject with pluere in two coordinated Accusative and Infinitive constructions: Liv. 28.27.16 lapides pluere et fulmina iaci de caelo … portenta esse putatis (`showers of stone and thunderbolts hurled from the sky you reck on portents') and 10.31.8 (terram `earth').

Note (end) return to page 23 The fact that human beings play such an important role in states of affairs, not only as first arguments (Subjects), but also as second arguments with two- and three-place predicates, has serious consequences for the way in which such constituents are marked formally. If two human beings are engaged in the same state of affairs, it is highly desirable to indicate formally what role each of them plays in that state of affairs. See ch. 5 and Plank (1979b).

3 Periphery 1: Adjuncts

Note (end) return to page 1 There are also obligatory Direction constituents, such as in the case of se conferre (see crosssection 3.2.). A special case is constituted by expressions such as fui Romam (`I have been to Rome') (see Garcia Hernandez 1983; Petersmann 1977: 104; Siegert 1959).

Note (end) return to page 2 There are also examples like: `This painting hangs here in order to be more conspicuous'. In such cases it must be assumed that someone has deliberately

-- 262 --

hung the painting in that particular spot.

Note (end) return to page 3 It might perhaps be better to distinguish a number of semantic functions `Manner', just as a number of functions `Place' are distinguished (`whereto', `whence', `where'). See, however, Vester (1983).

Note (end) return to page 4 Tres iam menses villam suam aedificabat, cum subito … (`He had been building his house for three years, when suddenly … ') and the like are possible. See chapter 11 on the imperfect.

Note (end) return to page 5 The assumption that consecutive clauses have a more independent status (unlike Purpose clauses, their occurrence is not restricted by the nuclear predication) may be supported by the fact that consecutive clauses are to a lesser degree subject to the so-called consequence of tenses.

Note (end) return to page 5a [added 12-08: Fugier (1993) rejects the idea that in habitare laxe voluit the presence of laxe compensates for the omission of a locative expression. According to her habitare in this meaning can be used both as a one-place and as a two-place verb.]

Note (end) return to page 6 See Dönnges & Happ (1977: 37). I will return to the ablative with these verbs and the undesirability of speaking of a semantic function Instrumental in crosssection 5.2.4(e).

Note (end) return to page 7 This division is based on Scherer (1975: 196). It is contestable on a number of points: in class (b) for example, `Time within which' and `Route along which' should perhaps also be considered more `central' with regard to the verb. Note that many of the semantic functions in (b) can also be expressed by (adverbial) subordinate clauses. For a similar division see also Hoberg (1981: 132–4).

Note (end) return to page 8 Some question words, however, correspond with more than one semantic function, e.g. quo modo? (`how'), the answer to which may concern Instrument as well as Manner.

Note (end) return to page 9 Time and Place Adjuncts constitute a deviation from the coordination rule: two Place Adjuncts or two Time Adjuncts may occur in the same sentence without being coordinated: - in Lysandri … statua … in capite corona subito exstitit (`On the statue of Lysander all of a sudden there appeared a crown on the head', Cic. Div. 1.75) This is only possible if the one Adjunct is more precise than the other. See Pinkster (1972c: 93–5). A second deviation occurs in the case of Purpose and Cause satellites and in the case of Instrument and Manner satellites, which can, in fact, be coordinated. They are, however, distinguished on different grounds.

Note (end) return to page 10 [added 12-08: The usefulness of the coordination test is rejected explicitly by Adrados (1992) with reference to Ottervick (1943). The mere fact, however, that this author offers so few cases of exceptional coordination is an indication of the validity of the coordination test. Rosén (1990, also 1989: 399) discusses examples of what she calls 'asymmetric coordination' (also called 'épitaxe') as exemplified by
- more hoc fit atque stulte mea sententia (Pl. St. 641)
- dicitur quidem istud ... a Cotta et vero saepius (Cic. Div. 1.8)]

4 Periphery 2: Disjuncts (sentence adverbials) and Theme constituents

Note (end) return to page 0 [added 12-08: In many languages modal adverbs derive from former sentential expressions. Latin examples are forsitan, scilicet, sis, dumtaxat. Related is the use of credo and other parenthetical verbs (Lindquist 1971 discusses this phenomenon, also for other lexical classes like adverbs, subordinators).]

Note (end) return to page 1 The so-called modal adverbs of K.-St. also include words which are currently treated as `Modalpartikel' in German studies, e.g. Franck (1980). An English example is `you surely do not mean that?'.

Note (end) return to page 2 Another argument for Disjunct status of a constituent could be the co-occurrence with an Adjunct of another constituent which could in itself also occur as an Adjunct with the same or a related semantic function, as e.g. stulte in stulte miles celeriter aufugit (see the coordination test on p. 30).

Note (end) return to page 3 Scherer (1975: 240) speaks of `umgangssprachlicher Ersparung einer Wendung wie hoc tibi dico'. See also K.-St. II. 233–4, Lodge, Lexicon Plautinum, s. v. ut, p. 930, 2nd col., and Vairel-Carron (1981a: 252). R. Bartsch (1972: 65–7) offers a similar analysis (`specification of an omitted verb of communication'); see also the notion `hyper-sentence'. Other examples from Scherer are: sed ad rem ut veniam (Cic. Att. 14.16.1); illaec tibi nutrix est, ne matrem censeas (Pl. Cist. 558). For other cases see also TLL s. v. atque 1078. 5 ff.; Merguet (1962) Phil. s. v. dico 677a; TLL s. v. dico 975. 82 ff.; Hey (1908); cases with ne in Lebreton (1901: 302–3); Vairel-Carron (1981a: 252–3). In K.-St. (II. 278), too, something like dico is supplemented to explain cases like: quod Salas te cum Clodio loqui vult, potes id mea voluntate facere (`As to the fact that Salas wants you to speak with Clodius, as far as I am concerned you can do so', Cic. Att. 12.30.1).

-- 263 --

Note (end) return to page 4 In Dutch the difference may also be seen in that in the case of pseudo-Purpose clauses there is no inversion of Subject and main verb, something which is obligatory in the case of real Purpose clauses. This also holds for the pseudo-conditional clauses discussed in crosssection 4.1.5. below.m

Note (end) return to page 4a [added 12-08: Torrego (1988) shows that in Purpose Adjuncts both quo and ut are allowed, whereas in pseudo-Purpose clauses quodoes not occur.]

Note (end) return to page 5 Example (13) is cited in van de Griend (1988:8) as an example of `interaction management'. Example (16) is also from her study.

Note (end) return to page 5a [added 12-08: On the various interpretations of siclauses, among them pseudo-conditional ones, see Bernal-Laveso (1992), esp. p. 11.]

Note (end) return to page 6 Pseudo-Cause satellites have been treated by Bolkestein (1991) and Fugier (1988).

Note (end) return to page 7 More instances of the so-called `dativus iudicantis' may be found in K.-St. (I. 322). Scherer (1975: 153–4) classifies heterogeneous instances under `Einschränkung der Gültigkeit', such as fortasse (`perhaps'), spero (`I hope'), quod sciam (`as far as I know'), nisi erro (`if I am not mistaken'), ut ita dicam (`so to speak') and the dativus iudicantis.

Note (end) return to page 8 It is not always possible to distinguish between Theme and Topic. Chausserie-Laprée (1969: ch. 1) points out that historians often place names of persons at the beginning of the sentence, as a kind of paragraph marker. In cases like (27), Havers (1926) speaks of `isoliert-emphatische Nominativ'. Emphasis is, however, often not relevant (see Boon 1981; Svennung 1936: 182). Examples may be found in K.-St. II. 586–7; Sz. 29, 401; TLL s. v. de 76.44 ff.; Bolkestein (1981a: 65–9). Further distinctions are made by Hoffmann (1989) and Somers (1989).

Note (end) return to page 8a [added 12-08: Another explicit way of 'thematizing' a constituent is by the use of a quod attinet ad ...' clause:
- ... quod quidem ad nos duas attinuit, praepotentes pulchre ... fuimus. (`... for our part we two gained a glorious victory.&rsquo, Pl. Poen. 1182-4)
A related phenomenon is the so-called proleptic use of the accusative as me in:
- Responde: quo leto censes me ut peream potissimum? ( `Answer - by what death dost think I had best breathe my last?&rsquo, Pl. Mer. 483)
In this example me refers to the Subject of the embedded clause (peream). For these expression types see Rosén (H.) (1992a).
Latin does not make frequent use of a well-known device of giving prominence to a constituent in the Romance languages, known as `cleft' in English grammars (Quirk et al. 1985: 1383 ff.). However, Löfstedt (1966: 259 f.) gives several examples, e.g.:
- ... neque tu eras tam excors tamque demens, ut nescires Clodium esse qui contra leges faceret, alios qui leges scribere solerent ... (`and yet you were not so senseless and so infatuated as not to know it was Clodius' part to act in defiance of the laws, and the business of others to formulate them&rsquo, Cic. Dom. 48)
More common are expressions like hic est qui and quis est qui. In all these cases parallel expressions of the type X est is qui/ are more frequent.
A related type of cleft is found in expressions like magis est ut (Lucr. 2.826 ff.), vix est ut (Aug. in psalm. 85.7), ante est ut (Cypr. Sent. episc. 55). See Löfstedt (1977: 273-275) and Norberg (1937: 112ff.).] 5 Relators

Note (end) return to page 0 [added 12-08: The English edition has an alternative example:
insperanti mihi … cecidit ut in istum sermonem … delaberemini (`Without my expecting this, it so happened that you ended up in this discussion', Cic. de Orat. 1.96)]

Note (end) return to page 1 crosssection 5.2. is partly based on Pinkster (1980).

Note (end) return to page 2 By noun phrases (NP) I understand here nouns (vir, `man'), pronouns (tu, `you'), combinations of a noun and an adjective (vir bonus, `good man'), substantival adjectives (boni, `the good (citizens)'), etc. A prepositional phrase is defined as a noun phrase marked by a preposition. The third type of `phrase' to be distinguished throughout this chapter is the `adjective phrase'; this term refers to expressions consisting of an adjective and a nominal constituent governed by or connected with that adjective, such as poena dignus (`worthy of punishment'), tres pedes altus (`three feet high/deep'). For the sake of convenience I sometimes use `noun phrase' to refer to a `prepositionless noun phrase'.

Note (end) return to page 3 The corpus consists of a total of 250 pages of text, viz.: Caes. Gal. 1.1–40; Catul. 1–14b, 69–102; Cic. Mil. 1–64, N. D. 2.1–32, Fin. 1.32–9, Rep. 6.1–21; Hor. C. 1.1–34; Liv. 21.1–25; Ov. Met. 2.1.–339, 11.410–748; Sal. Jug. 5.4–16, Cat. 5–22; Sen. Ep. 1–2, 5–7, 10, 12, 17–18, 23, 26; Tac. Ann. 13.1–43; Verg. A. 1. Details may be found in Bolkestein et al. (1976; 1978). For some relativizing remarks on parts of table 5.1 see Pinkster (1980: 113).

Note (end) return to page 4 On the infrequent use of prepositions in Virgil see Görler (1982: 72–3). A general survey of the frequency of prepositions in Latin texts is presented by Denooz (1988). Another factor underlying this difference may be the fact that Virgil and Caesar do not contain the same (relative) number of satellites (A. M. Bolkestein, personal communication).

Note (end) return to page 5 Stankiewicz (1960: 75–7); Sanders (1973: 31).

Note (end) return to page 6 We do, incidentally, also find the `normal' use in cases such as the following (see K.-St. I.336 `vereinzelt und unklassisch'):
- quotiens hoc tibi … interdixi, meam ne sic volgo pollicitere operam (`How often have I forbidden you this, viz. to promise my support in public like that', Pl. Mil. 1056)
- interdixit histrionibus scaenam (`He denied the actors access to the stage', Suet. Domit. 7.1). For the construction acc. + abl. see TLL s. v. 2175, 62 ff. TLL s. v. 2174, 79 also

-- 264 --

mentions Liv. 9.43.24 (difficult to interpret) and 34.7.3 (conjecture). The person who is forbidden or denied something can become Subject in the passive:
- philosophi … eiecti atque urbe et Italia interdicti sunt (`the philosophers have equally been expelled and denied access to Rome and Italy', Gel. 15.11.4)
For constare + abl. see note 30.

Note (end) return to page 7 An exception is:
- cur haec … celata me sunt? (`Why have these things been hidden from me?', Pl. Ps. 490)
Examples of this group of verbs may be found in Flobert (1975: 400–1).

Note (end) return to page 8 K.-St. I.379 incorrectly interpret the ablative fidibus in example (21) as an ellipsis of fidibus canere. In example (23) most editors delete a as `unciceronian'.

Note (end) return to page 9 On the analogy of docere there is an occasional occurrence of erudire aliquem aliquid (K.-St. I.299 A. 5):
- (Daedalus) damnosas erudit artes (natum) (`D. teaches his son the harmful art (of flying)', Ov. Met. 8.215) TLL s. v. erudire 830.7 ff.; instituere 1990.46 ff. For teaching verbs see also Dik et al. (1981).

Note (end) return to page 10 The `double accusative' construction is also occasionally found with donare and condonare. In Classical Latin these verbs have two constructions, viz.
- Object animate acc. + Compl. inanimate abl.
- Indirect Object animate dat. + Object inanimate acc.
In Old Latin we find e.g.:
- argentum quod habes condonamus te (`The money you have we give to you', Ter. Phorm. 947)
(cf. K.-St. I.302–3). Apparently, as the two existing constructions allow both the animate entity and the inanimate entity to be the Object, we also find a mixed construction. Examples, mostly from later Latin, may be found in Svennung (1936: 226 ff.).

Note (end) return to page 11 Note that also compound verbs with prepositional prefixes such as de-and e-have Complements marked by the case governed by the related prepositions:
- decurrere monte (`They ran from the mountain', Hor. C. 4.2.5)
- vox horrenda edita templo (`A terrible voice coming from the temple', Liv. 6.33.5)
This usage is mainly poetical. See K.-St. I.362 f.; Lehmann (1983).

Note (end) return to page 12 For the double accusative one might consider a regularity transire flumen (non-causative): transducere flumen (causative), see. p. 51 about `to lack', `to deprive'. For examples see Flobert (1975: 404–8).

Note (end) return to page 13 With consulere alicui/aliquem the same lexemes are found. The difference in meaning appears from, among other things, the fact that only to the construction with the accusative a de-satellite can be added.

Note (end) return to page 14 See for cupere TLL s. v. 1435, 33 ff.; cupere + dat. is synonymous with favere, which always governs the dative. Coincidence? This kind of ostensible opposition is only found between arguments in the dative and arguments in the accusative.

Note (end) return to page 15 I distinguish a separate semantic function Addressee with communication verbs. Dik (1978: 70) remarks that Recipient is a rather central semantic function (in some languages constituents with this semantic function can be assigned the syntactic function Subject: cf. English `John was given the book'). Note that in passive sentences the so-called dativus auctoris and the ablativus causae differ in terms of restrictions on the constituents that may be used: + animate vs. - animate, respectively (Bolkestein 1980a: 138–9). For Latin I see no semantic similarity between Recipient arguments and Beneficiary satellites (so-called dativus (in)commodi). I assume that Recipient and Beneficiary constituents might co-occur in one sentence - as they may in other languages. The alternative prepositional expressions and the historical development are quite different:

-- 265 --

Recipient > ad + acc., Beneficiary > pro + abl. For a discussion of this problem for Vedic - with a different outcome - see Hettrich (1988).

Note (end) return to page 16 Data on donare were supplied by R. Risselada (personal communication). For pragmatic factors as determining the use of cases with these verbs see Bolkestein (1985) and Bolkestein & Risselada (1985). See also p. 255. An extensive syntactic and semantic classification of this group of verbs is to be found in Lemaire (1983).

Note (end) return to page 17 See especially Heilig (1978: 53, 119, 125). He bases his conclusion on research on 800 sentences in Cicero's philosophical works. Outside this corpus, the strict separation of + accusative and - accusative verbs cannot always be applied. Heilig's research covers more ground than I can deal with here. Ebeling (1957: 135–6) proposes a semantic explanation for the 'normal' accusative: according to him, transitivity may be defined semantically. He does not give any further arguments for this.

Note (end) return to page 18 The semantic classes distinguished by Heilig should be defined more precisely; to my mind, Chafe's (1970) criteria are insufficient. A more elaborate treatment of the problems dealt with in this section is Pinkster (1989).

Note (end) return to page 19 `the dative indicates an entity which is animate (human) or considered as such, and is therefore endowed with free will, is actively involved in the action performed by the Subject and actively interacts with the Subject, whereas the ablative indicates an entity which is inanimate or considered as such, and therefore without free will'.

Note (end) return to page 20 `Thus, for instance, in legibus paremus (`We obey the laws') the laws are presented as personal in the same way as the king in regi paremus (`We obey the king')'.

Note (end) return to page 21 See K.-St. I.383; Väänänen (1981: 112) for Pompeian inscriptions and Harris (1975: 184–6) for the development in the Romance languages.

Note (end) return to page 22 E.g. Dönnges & Happ (1977: 37). In my opinion this can be proved to be wrong. See also Touratier (1978: 114–15); Dönnges & Happ also assume a semantic function Separativus within the nuclear predication, equally incorrectly to my mind.

Note (end) return to page 23 Pinkster (1972c: 78–82).

Note (end) return to page 24 I add the example with potiri ((34)) because the meaning of uti (`to use') is misleading.

Note (end) return to page 25 The paraphrase is somewhat odd because we are dealing with abstract notions.

Note (end) return to page 26 An exception is the case of an army commander and his solidiers.

Note (end) return to page 27 Other tests which might show the differences are: what interrogative pronoun corresponds with the constituent under discussion? Are there restrictions on the semantic function of the Subject constituent? Note that strictly speaking the dativus commodi and the ablativus instrumenti can only occur with a nuclear predication which has an Agent. Favere and frui both occur with inanimate Subjects.

Note (end) return to page 28 The genitive of value and price is not taken into consideration.

Note (end) return to page 29 Certain scholars consider Price Adjuncts obligatory, e.g. Fillmore (1972: 9).

Note (end) return to page 30 If the dative constituent is obligatory, we have here the exceptional pattern nom. - dat.-abl. See p. 44, (i).

Note (end) return to page 31 Incorrectly, the ablative Complement constituent with (com)mutare is sometimes classified as ablative pretii:
- nemo nisi victor pace bellum mutavit (`No one exchanges war for peace unless he has won', Sal. Cat. 58.15)

Note (end) return to page 32 Example (39) may possibly be explained with the aid of the context:
- Capuae interim Flaccus dum bonis principum vendendis agro qui publicatus erat locando - locavit autem omnem frumento - tempus terit (`At Capua meantime, while F. was spending his time in selling the property of leading men, in leasing lands that had been confiscated - he leased it all in exchange for corn')
The aspect of transaction is sufficiently clear in the context.

-- 266 --

Note (end) return to page 33 I leave aside the issue of whether we might not just as well speak of an Instrument Adjunct with emere.

Note (end) return to page 33a [added 12-08: In Antiquity instructions can be found to avoid this type of ambiguity, as in Quint. I.O. 8.2.16. See Roberts (1912). I owe the reference to Heine (1990:8).]

Note (end) return to page 34 Fugier (1973) offers a classification of the uses of the genitive on the basis of permutation and transformation tests.

Note (end) return to page 35 For the extension of the use of the genitive see K.-St. I.44 ff. The acc. of respect, almost certainly due to Greek influence, is limited to poetry and Tacitus (examples in K.-St. I.285 ff.).

Note (end) return to page 36 These rules are so-called `expression rules' in the sense of Dik (1978: ch. 7).

Note (end) return to page 37 For interesting observations on the relative lack of importance of word order for the understanding of sentences such as (58) by English-speaking children see Clark & Clark (1977: 501–2).

Note (end) return to page 38 Passages from Caes. Gal., Catul. 64 and Verg. G. 1.

Note (end) return to page 39 An obvious objection against this type of approach is that research is done on a language as if the case system were a marginal phenomenon, whereas in reality it may be assumed that the case system on the one hand imposes certain limitations, but creates certain possibilities on the other.

Note (end) return to page 40 It is striking that in practically all instances of the same case form (filium/filios) singular and plural are formally distinguished, but that in many instances of the same number different case forms do not formally differ (e.g. hortis (dat. pl.)/hortis (abl. pl.)); an exception is res/res.

Note (end) return to page 41 R. Pfister (personal communication) points out that in Latin the relatively late position of the predicate in the sentence enhances the necessity of case marking.

Note (end) return to page 42 About redundancy see Calboli (1965: 72–81). My percentage of 5–10% is close to that calculated for Old French by Hupka (1982) (personal communication of G. Bossong).

Note (end) return to page 43 Thus there are hardly any examples of the ablativus temporis in K.-St. which do not inherently indicate a period or moment (e.g. dies) or a function (e.g. consulatus). Exceptions are formed by a number of action nouns, such as servili tumultu (`during the uprising of the slaves') and proscriptione (`during the proscriptions') (Suet. Jul. 11). I have not found examples comparable to English during/after the soup.

Note (end) return to page 44 For the problems involved with the postulation of a one-to-one relation between form and meaning (the so-called universal principle of Von Humboldt) see van Marle & Koefoed (1980). This principle is, incidentally, also the basis of many structuralist approaches of language (among others de Groot (1956a; 1956b).

Note (end) return to page 45 A number of elements of the traditional approach to the case system is discussed in crosssection 5.3.

Note (end) return to page 46 Example (72) may alos be analysed differently: de annona on noun phrase level with senatus.

Note (end) return to page 47 Cf. Sz. 214–15; Baldi (1979); Calboli (1965: 74); Dressler (1971); Lehmann (1982: 90–4); Luraghi (1988). Further literature in Pinkster (1972c: 148). Calboli sees the addition of prepositions as a manifestation of the need for redundancy (see note 42). [add. 12-08: For a treatment of prepositions as specifiers of the semantic value of the cases see Echarte (1991a).]
Prepositions derive from various sources. Adverbs are commonly seen as the source for prepositions (especially the `original' monosyllabic ones, but there are deadjectival, denominal and deverbal prepositions as well. Examples of the latter category are secundus, versus, secus(?), trans, iuxta. For this kind of development see Kortmann & König (1992).

Note (end) return to page 48 Other arguments which are adduced to support the `specification hypothesis' involve tmesis and inversion.
- per te, ere, obsecro, deos immortales (`I beg you, lord, in the name of the immortal gods', Pl. Bac. 905–6)
- nec … demimus … hilum tempore de mortis (`We take away nothing from the duration of our death', Lucr. 3.1087)
Both phenomena are highly stylistic and non-productive. Also, it is by no means always possible to determine whether tmesis is really involved (Löfstedt 1961: 276–93). Furthermore, words like extra are adduced, which are both preposition and adverb. This can never be an argument to prove that one of the categories did not exist at some point. Moreover, Dressler (1971:9) remarks that the fact that it cannot be determined whether Indo-European had pre- or postpositions is an

-- 267 --

argument to assume that it had neither. Examples of languages without pre- or postpositions but with cases are Dyirbal (Australia), Mojavi and Diegueño (Yuma) and possibly Hittite (personal communication, Chr. Lehmann). Sometimes it is difficult to determine whether we are dealing with cases or with other morphological devices. Just as in Latin the noun causa developed into a preposition, so in other languages there are developments resulting in prepositions. A reverse development, leading to the adverbialization of former prepositions, also occurs (Löfstedt loc. cit.). For the relation between cases and prepositions see also Kilby (1981) and Lehmann (1982: 74–106). Dik (1983) tries to explain why the system with cases was followed by a system with pre positions (rather than postpositions). For the usefulness of typological considerations of this kind see Dressler (1971) and Strunk (1977).

Note (end) return to page 49 Cf. Pfister (1988: 71).

Note (end) return to page 50 Note that the phenomenon is limited to three-place verbs and to the ablative.

Note (end) return to page 51 If the two constructions are synonymous, this need not necessarily mean that they are used with the same frequency. On the basis of the data in the TLL s. v. 1309.23 ff. the following tendency may be formulated:
noun phrase + humannoun phrase - human
liberare + abl.-+
liberare + ab + abl.++
For this phenomenon in Ancient Greek see Moreux (1976). It is not surprising that in the case of `people' there is a greater need for the semantic function to be expressed explicitly than in the case of non-animate entities, since `people' may fulfil a multitude of semantic functions. Cf. also Théoret (1982), who sees a relation between concrete and preposition on the one hand and abstract and a mere case form on the other (cf. above p. 27).

Note (end) return to page 52 For liberare ex cf. Cic. Ver. 5.23 ex his incommodis pecunia se liberasse (`That he freed himself from these inconveniences with the aid of money'). Here the substitution by ab and perhaps of the mere case form might be possible. In the only other extant Classical instance, Vitr. 7.3.6, this is certainly not the case. Liberare de does not occur in Classical Latin. TLL s. v. liberare 1309.23 ff.

Note (end) return to page 53 `Happens to be', in the sense that there is no synchronic explanation.

Note (end) return to page 54 Note that, if in the synchronic grammar of Latin prepositions were specifications of semantic relations which are in reality already expressed by the case form, one would strictly speaking also expect de + gen. etc. For de cf. the development in the Romance languages.

Note (end) return to page 55 Cf. the Romance languages (French dire à) and K.-St. I.519 [add 12-08: for early instances of prepositional expressions with ad that are more or less corresponding to a dative. See also Pinkster (1990) and Heine (1990:3-4) with justified criticism on the Dutch and German editions.
Grassi (1964) discusses several constructions where prepositions seem to be reserved for particular lexical classes such as the use of the preposition per for human instruments as opposed to the bare ablative case for inanimate instruments. His explanation is that the ablative is often identical to the dative, the latter case being typically a 'human' related case. Per, therefore, avoids ambiguity between instrument and beneficiary. (See also Löfstedt 1942: 274). A similar disambiguating function may be seen in the pairs per Latium and Latio.]

Note (end) return to page 56 To give an impression: the French translation of the passage from Caesar which forms the basis of table 5.3 on p. 42 contains (i) c. 20% more words, and (ii) c. 40% more prepositions. An excellent study on the increasing use of prepositions to express an Instrument relation is Beckmann (1963).
[add.: From ancient times onwards prepositions are regarded as devices to explicitate semantic and syntactic relationships. Cf:
- Praecipuamque curam duxit (scil. Augustus) sensum animi quam apertissime exprimere. Quod quo facilius efficeret aut necubi lectorem vel auditorem obturbaret, neque praepositiones urbibus addere neque coniunctiones saepius iterare dubitavit, quae detractae afferunt aliquid obscuritatis, etsi gratiam augent (`He made it his chief aim to express his thought as clearly as possible. With this end in view, to avoid confusing and checking his reader or hearer at any point, he did not hesitate to use prepositions with names of cities, nor to repeat conjunctions several times, the omission of which causes some obscurity, though it adds grace.', Suet. Aug. 86)]

Note (end) return to page 57 For an exception see ch. 6, n. 35 (p. 270).

Note (end) return to page 58 The term `Prepositional Complement' has been adopted following Quirk et al. (1985: 657). The term `Relator' derives from tagmemic grammar. For `relator' in Functional Grammar see van Limburg (1985).

Note (end) return to page 59 Cf. K.-St. I.579; Pinkster (1972c: 146–7) and Mallinson & Blake (1981: 202).

Note (end) return to page 59a [added 12-08: The English edition has another example, viz. insperanti mihi &hellip cecidit ut in istum sermonem … delaberemini (`Without my expecting this, it so happened that you ended up in this discussion', Cic. de Orat. 1.96)]

Note (end) return to page 60 [modified 12-08: For semantic and pragmatic differences between the ut- and quod-constructions, especially with verbs of happening, see K.-St. II.275; Ehlers (1971), Heine (1990: 4), and Rosén (1988)]. Serbat (1988b: ch. 4) stresses the point that subordinators in

-- 268 --

argument clauses have no semantic value of their own.

Note (end) return to page 60a [added 12-08: Both ex. (87) and the introductory text differ from the English edition, in which an example from Cic. de Orat. 2.15 is used.]

Note (end) return to page 61 Late exceptions in TLL s. v. 816.6 ff.

Note (end) return to page 62 Cf. Matthews (1974: 45–7). An exception is de Groot (1948: 451; 460–1; 463), who offers the following classification (: 460–1):
without lifelessnesswith lifelessness

Note (end) return to page 63 Distinctions like bonus/bona est (`He/she is good') are, of course, semantic.

6 The (internal) structure of noun phrases

Note (end) return to page 1 For the notion `noun phrase' as used in this book see ch. 5, n. 2.

Note (end) return to page 2 For the notion `omissibility' see p. 11 ff.

Note (end) return to page 3 Another term for Attribute is `Modifier'.

Note (end) return to page 4 See p. 70 for the term `Relator'.

Note (end) return to page 5 Unless cum irraso capite is to be interpreted as Praedicativum (see ch. 8). Similar cases are given in Fraenkel (1968: 67).

Note (end) return to page 6 For Attributes in the dative see K.-St. I.260; Sz. 91, 95, 99. For the accusative see K.-St. I.260; Sz. 34. For the ablative see Sz. 128; for the ablativus qualitatis see K.-St. I.454–7; Sz. 117–19. For cases such as (6) see also Flobert (1975: 551) and Rosén (1981: 41 ff.; 1983: 194 ff.). For the so-called genitivus qualitatis see K.-St. I.454–7; Sz. 68–71. See also Sz. 34 for nouns in -tor.

Note (end) return to page 7 Additional examples of prepositional phrases as Attributes may be found in Sz. 428, Scherer (1975: 206–7), Nägelsbach (1905: 306 ff.).

Note (end) return to page 8 Additional examples in Sz. 171; Adams (1976b: 82–3); Scherer (1975: 208).

Note (end) return to page 9 Woytek (ad locum) gives a Greek translation.

Note (end) return to page 10 [modified 12-08: quod-clauses are especially common if introduced by a pronoun, as here by illam.]

Note (end) return to page 10a [added 12-08: The English edition has another example:
- di eam potestatem dabunt ut beneficium bene merenti nostro merito muneres (`Heaven will give you the ability to do a deserved kindness to a man that has been so kind to us', Pl. Cap. 934–5)]

Note (end) return to page 11 Examples of such complex (or `periphrastic') predicates can be found in Rosén (1981).
[added 12-08: For criticism of the earlier edition see heine (1990: 4-6).]

Note (end) return to page 12 So-called `relative connection'; German: `relativer Anschluss'.

Note (end) return to page 13 For the sake of the argument I do not consider the fact that the two sentences would normally differ in that in (33) a capital letter would be used (`the Queen').

Note (end) return to page 14 Some observations from antiquity about pause phenomena are extant, e.g. Quintilian's remarks (11.3.35 ff.) about the pause assumed by him in Verg. A. 1.1–2 arma virumque cano, Troiae qui primus ab oris Italiam fato profugus Laviniaque venit, but remarks of this type are not very helpful. C.J. Ruijgh (personal communication) notes that rather than `pause' we may also find `terminal deceleration'.

Note (end) return to page 15 Mallinson & Blake (1981: 359–67) point out that in non-restrictive relative clauses in English the relative pronoun cannot be omitted. In non-restrictive relative clauses in Greek the relative pronoun cannot undergo attraction (see (iii), p. 82). See Rijksbaron (1981: 240 + bibliography). See also Quirk et al. (1985: 1208) and Touratier (1980a: 272–4). Kleiber (1987) is highly sceptical about the distinction.

Note (end) return to page 16 See Vester (1977: 257). Touratier (1980a: 278) remarks that in cases where a proper name may refer to more than one person the pronoun ille is added to indicate that the author/speaker is referring to a specific person, identifiable for the reader/hearer.

-- 269 --

Note (end) return to page 17 See also Touratier (1980a: 284).

Note (end) return to page 18 K.-St. II.320 treat this example as an instance of a connecting relative. See (ii) below.

Note (end) return to page 18a [added 12-08: Meyers (1992), discussing Lavency (1981)'s examples and Touratier (1980), rejects the idea that relative clauses can be distinguished in determinative and appositive ones, corresponding to a difference in mood (indicative and subjunctive, respectively). He follows up a suggestion made by Benveniste (1966) and regards the relative pronoun as deriving from an original article-like marker, which functions in the same way as the article in a NP with a substantivized adjective, which can be reinforced by is (`déterminatif solidaire'). The mood distinction correlates with the distinction `specific' (indicactive) : `unspecific' (subjunctive). The translation of the following Latin sentences [which in my eyes resist such an analysis] has to be as indicated:
- o fortunate adulescens, qui tuae virtutis Homerum praeconem inveneris (`Fortunate youth, to have found in Homer an herald of thy valour!', Cic. Arch. 24): `Heureux jeune homme, toi, un homme quia trouvé Homère pour chanter ta valeur'.
- Bellovaci autem, defectione Haeduorum cognita, qui ante erant per se infideles, manus cogere atque aperte bellum parare coeperunt (`The Bellovaci were disloyal in themselves before the revolt of the Aedui, and when they heard thereof they began to collect companies and openly to prepare for war.', Caes. Gal. 7.59.2) : `Les Bellovaques, ce peuple qui'.
- (Galli) hac impulsi occasione, qui iam ante se populi Romani imperio subiectos dolerent, liberius atque audacius de bello consilia inire incipiunt (`Such an opportunity served as a stimulus to those who even before were chafing at their subjection to the sovereignty of Rome, and they began with greater freedom and audacity to make plans for a campaign.', Caes. Gal. 7.1.3): `Les Gaulois, un peuple qui'.
For a pragmatic analysis of relative clauses see Eckert (1992).For the Romance developments see Schafroth (1993: 59-80) ]

Note (end) return to page 19 For the frequency in Latin see Kurzová (1981: 47). For Greek see Ruijgh (1971: 611).
[added 12-08: The use of relative connection is a phenomenon that is relatively common in Caesar and Apuleius. The relative occurs relatively rarely as Subject of its clause and is then not interchangeable with is and hic (nor against Zero (0) and ille). In Caesar this device is particularly frequent in resuming ablative absolutes (quibus rebus nuntiatis) and as an Attribute (quarum adventu). See Bolkestein (1993).]

Note (end) return to page 20 Some languages do have relative clauses but do not have relative pronouns, cf. English `the friend I met today'. For types of `relative markers' see Mallinson & Blake (1981: ch. 5).

Note (end) return to page 21 See Comrie (1981a: 148 ff.); Lehmann (1984); Mallinson & Blake (1981: 346 ff.) for a typological survey. In English the relative pronoun that can only occur as Subject or Object. For Latin see Touratier (1980a: 390 ff.).

Note (end) return to page 22 According to Kurzová (1981: 48), `attractio' is much more frequent in Greek, whereas in Latin `attractio inversa' occurs more often. See also Comrie (1981a: 146–8).

Note (end) return to page 23 See Sz. 555; Kurzová (1981: 22–35), and especially Lehmann (1984: chapter VI) on various types of development. Lehmann (1984: 372–3) observes that the interrogative/indefinite origin of the Latin relative pronoun requires an original order `qui-clause: main clause'. In Classical Latin relative clauses usually follow the antecedent. He presents a hypothetical explanation for the change in word order.

Note (end) return to page 24 For an exception: animo consulem esse oportet, consilio, … (`A consul should be such in spirit, in discretion, …', Cic. Pis. 23); see Pinkster (1982: 252).

Note (end) return to page 25 See Untermann (1980: 343–5). I leave out of consideration the fact that many adjectives may also be modified by degree-modifiers (valde bonus). This is a different matter.

Note (end) return to page 26 I use the vague term `qualifying adjective' here for the sake of brevity in order to distinguish it from the other categories of adjectives.

Note (end) return to page 27 See K.-St. I.240–1 for the types of Attribute that may be nested, and also for multus. In Greek, too, we find πολλαι και βαρειαι λυπαι (C.J. Ruijgh, personal communication). A methodological problem is posed by the fact that in Latin asyndetic coordination cannot formally be distinguished from nesting. I ignore this issue here (but see Pinkster 1972c: 123–4). Very extensive treatments of nesting of adjectives are given by Fugier & Corbin (1977), Fugier (1983) and Risselada (1984).

Note (end) return to page 28 Cf. the survey in Fugier & Corbin (1977: 251), from which I deviate on some points. The formal distinction between `quantifying' and `qualifying' adjectives appears from the restrictions on the former group with regard to occurrence in exclamations and rhetorical questions (Vairel-Carron 1975: 80; Hoff 1983). See crosssection on p. 201.

Note (end) return to page 29 For the separate place in this survey of indefinite pronouns cf.
- ut hos dies aliquos sinas eum esse apud me (`That you allow him to be these few days with me', Pl. Truc. 872)
- At pol illi quoidam mulieri nulla opera gratuita est (`But, by God, for this woman there are no free services', Pl. Cist. 740; NB: quidam MSS)
See also crosssection 6.7, p. 93.

Note (end) return to page 30 Fugier & Corbin (1977) reckon under (a) also nonnulli, aliquot, plerique, uterque, neuter, alteruter. These words, however, can be combined with pronouns, cf. expressions such as Pl. Pseud. 283 aliquot hos dies (`These few days') (cf. TLL s. v. aliquot 1616.36 ff; TLL calls aliquot - rightly, in my opinion - a `numerale infinitum'). Cf. Risselada (1984).

Note (end) return to page 31 Examples for ille in TLL s. v. 360.12 ff.

Note (end) return to page 32 But see example (56) on p. 85.

Note (end) return to page 33 Another possibility is old, beautiful paintings. I leave this out of consideration. Ney (1983) discusses word order variation in English of the type illustrated by example

-- 270 --

(67). According to him, native speakers accept much more variation than is allowed by Quirk et al. (1985)

Note (end) return to page 34 See Risselada (1984). From a statistical point of view, there are many possibilities, e.g.
- novos legatos tres Liv. 22.61.7
- novas quattuor legions Liv. 22.36.2
- quattuor legiones novas Liv. 33.25.10
- legiones quattuor novas Liv. 42.31.2
- quinqueremes novas quinquaginta Liv. 35.24.8
(Data from R. Risselada: Research Report January 1982)

Note (end) return to page 35 For prepositional phrases and adverbs see Lebreton (1901: 90–1); Nägelsbach (1905: 306 ff.). For a genitive see K.-St. I.232; for the use of a prepositional phrase as a prepositional Complement (mainly in official contexts) see Sz. 274–5; 693, e.g. de in equis (`About those on horseback', Clem. ad Cor. 37), and Väänänen (1977).

Note (end) return to page 36 See Lehmann (1984: 293 ff.) on independent relative clauses in general and on their syntactic functions.

Note (end) return to page 37 K.-St. (II.282 A2) do point out that in many cases modification of a demonstrative is problematic. Pfister (1984: 167 n. 1) opposes the idea that an antecedent has been `omitted' in cases such as (78).
[added 12-08: Lavency (1991) describes is in is qui bene amat bene castigat as a cataphoric adjective, which is more or less obligatory according to the syntactic position of the relative clause in its sentence. Similarly Serbat (1991: 167) compares is to the use of the article in French NPs. See also Serbat (1988b).]
For related problems with regard to cum … tum, etc., see Pinkster (1972c: 171 ff.) and crosssection 7.3.3.

Note (end) return to page 38 The best survey of the many ways in which the genitive is used is offered by de Groot (1956a). In the grammars a total of c. 30 different uses are distinguished! See also den Hengst (1976) and Lavency (1981b). For the objective and subjective genitive see K.-St. I.416.

Note (end) return to page 39 Further research is required. Benveniste (1962) describes the occurrence of the genitive with verbal nouns such as amor as a kind of nominalization transformation. See also Fugier (1973a). Examples of non-genitive Attributes are given by Wistrand (1933: 67–74).

Note (end) return to page 40 K.-St. I.415 mention examples like Cic. Tusc. 1.23 quaestio animorum, which they translate as `Untersuchung über'. The Loeb edition, however, translates `this problem of the nature of the soul'. Another example mentioned by K.-St. is Cic. Tusc. 1.27 excessus vitae (parallels in TLL s. v. excessus 1229.44 ff.), which is less normal than excessus e vita (Cic. Fin. 3.60). One might suggest that the example in question has to be related to excedere vitam (found once in Itala II Macc. 6.27) and not to excedere e vita. Consider also the case of bene mereri de (`to render a service to'). TLL s. v. meritum 816.58 lists quite a lot of instances in which the person to whom a service was rendered is marked by a preposition, but only one instance with the genitive: Sen. Ben. 6.41.2 merita amicorum (`Services rendered to friends'). The Loeb edition, however, translates amicorum as a subjective genitive: `the services of (rendered by) friends'. In Greek we find more or less the same situation. Further research is required.

Note (end) return to page 41 TLL s. v. deditio mentions as an example of deditio + dative Liv. 31.18.6 qui … eorum deditionem vivorum hosti fecissent (`Who had surrendered them, while they were still alive, to the enemy&rsquo), but here deditionem fecissent can also be interpreted as one complex expression, with which hosti is construed. For a limited number of examples of a third argument in the genitive see Pinkster (1985b). There is at least one instance of an Addressee in the dative (Cic. de Orat. 3.207 sibi ipsi responsio (`Answering one's own question&rsquo); see K.-St. I.317).

Note (end) return to page 42 An extensive treatment may be found in Maurel (1985); see also Lavency (1981b: 7). Diachronic developments are certainly discernible (Sz. 68–71).

Note (end) return to page 43 In English (and in Greek) the article also serves other purposes, e.g. its use in generic expressions such as The tiger can be dangerous (see Quirk et al. 1985:

-- 271 --

265 ff.). I do not deal with such matters here. For the uses of the definite article in English see Hawkins (1978: 106–49). For the recognizability of Latin NPs as `generic' or `non-generic' see Leal Carretero (1986).

Note (end) return to page 44 In Sz. 191 it is implied that already in Petronius there are instances of ille which could be explained as definite articles. This is incorrect; see Petersmann (1977: 135) and Stumpf (1976). It is true that ille may be used to make clear to the reader/hearer that he is considered capable of identifying an entity not present in the context (Pinkster forthcoming b). For recent suggestions about the development of articles see Raible (1985) and Selig (1989). See in general Givón (1981) and Lehmann (1982: 51–4). Wehr (1984: 39–46) distinguishes three uses of unus: (a) numeral, (b) `irgendein beliebiger' (`any(one)'; cf. my example (93)) and (c) unus used in order to introduce a new entity, often with verbs meaning `to exist' or `to appear'.

Note (end) return to page 45 For the notion `associative anaphora' see Hawkins (1978: 123 ff.); see also Vester (1973: 50–7) and p. 248. Languages may, therefore, differ as to the degree to which definiteness is explicitly marked. In Greek, words belonging to group (i) may occur without the definite article (e.g. ηλιος).

Note (end) return to page 46 For the distinction deixis : anaphora cf. Ehlich (1982).

Note (end) return to page 47 Krenn (1984) argues that in Latin indefinite NPs are placed post-verbally at the end of the sentence (as is the case in Russian). This is incorrect. Neither is it correct that the preposition de is used as an indefiniteness marker (Krenn 1984: 56). It has been suggested above that there is a clear-cut distinction between `definiteness' and `indefiniteness'. See, however, Comrie (1981a: 128–9) for the notion that there is a hierarchy of definiteness and that languages may differ as to the degree to which they explicitly express definiteness. See also Dik (1978: 61–2).

Note (end) return to page 48 See Hawkins (1978: 203–14) about English, Rijksbaron (1981: 250–1) for Greek, and for Latin, Orlandini (1983) and Serbat (1986).

Note (end) return to page 48a [added 12-08: Serbat (1984) objects to traditional descriptions of quidam in which it is said or implied that quidam is used in order to be vague about the identity of the Head constitutent of the NP. In fact, as Serbat observes, the very use of quidam with Proper names is in conflict with this idea. A striking example is:
- erat Pipa quaedam, uxor Aeschrionis Syracusani, de qua muliere plurimi versus ... tota Sicilia percelebrantur (`There was a certain Pipa, the wife of Aeschrio of Syracuse, a woman about whom a number of lampoons were in circulation throughout Sicily', Cic. Ver. 5.81-82)
Cicero eleborates on the identity of this Pipa. Another example is:
- voluisti enim in suo genere unum quemque nostrum quasi quendam esse Roscium(`For you would have every man of us be a kind of Roscius in his own line.', Cic. de Orat. 1.258)
According to S. the value of quidam is the following:
(a)`un être déterminé'
(b)`appartenant à un ensemble d'analogues. C'est donc une sorte de quantificateur ...'.
However, the aspect `ensemble' is sometimes almost absent, which, especially in Lucretius, brings quidam close to aliquis en alius. The use of quidam in expressions like:
- incredibili quodam studio (`with a really incredible enthusiasm', Cic. de Orat. 1.14)
must be explained along the same lines: `L'ardeur est qualifiée d'incroyable ... C'est un ardeur que "se distingue parmi d'autres" conçues comme ayant la même intensité'.
The use of quidam decreases after Lucretius. For Apuleius' use of quidam see Van Mal -Maeder (1994). ]

Note (end) return to page 49 For the definition of apposition see Quirk et al. (1985: 1300 ff.); Lambertz (1982: 239–45; 257–8). In K.-St. (I.206) and Sz. (462) `Apposition' is described as a kind of Attribute, viz. a noun in the function Attribute; Sz. also sees a historical relation between the two phenomena. For general remarks about apposition in Latin see further Fugier (1973b).

Note (end) return to page 49a [added 12-08: In cases of `restrictive' apposition as in urbs Satricum and Satricum urbs the two constituents do not depend on each other nor do they behave as one unit. Only semantically the former may be said to be more closely coherent, because the general noun urbs is specified by the Proper name. See Heberlein (1993). On the relationship between the two constituents in Apposition see also Longrée (1990). He also observes that the relationship between the two constituents is only semantic. He gives tests for distinguishing restrictive and non-restrictive Apposition, the latter allowing the paraphrase is est X. ]

Note (end) return to page 50 Cf. Fugier (1973b: 99).

Note (end) return to page 50a [added 12-08: For the use of relative clauses as Apposition see Longrée (1991).]

7 Complex sentences (embedded predications on the sentence level)

Note (end) return to page 1 In the case of the Praedicativum, too, we are dealing with an embedded predication (see p. 148 ff.). A Praedicativum is not obligatory, i.e. is not required by the (main) predicate, but does itself contain a predication of one of the nominal constituents of the main predication. Indirect questions, which function as arguments, will not be dealt with separately. On these see Bodelot (1987).

Note (end) return to page 2 Parataxis is actually a wider notion, comprising also sequences of independent sentences, either with or without a so-called `connector' (such as autem (`however'), itaque (`therefore')) indicating the semantic relation between those sentences. See chapter 12.

Note (end) return to page 3 In the case of est + fama we are dealing with an instance of existential esse (`to exist'), see chapter 1, n. 4. In the following example I take the embedded predication as an Attribute with famam, like the AcI with nuntium in example (4) above (p. 99).
- fuisse famam venenum sua sponte sumpsisse (`That there was the rumour that (Themistocles) had taken poison of his own accord', Nep. Them. 10.4)

Note (end) return to page 4 This is the reading found in the manuscripts. In cases of this kind editors tend to change the text (constaret + AcI). See TLL s. v. consto 535.49 ff.; K.-St. I.706. See also crosssection 7.5.1. on p. 135.

-- 272 --

Note (end) return to page 5 It is not completely certain that ut valeant is to be interpreted as an argument of refert. Refert may also refer to the preceding instruction to take good care of the animals. Considerations in favour of the treatment as argument are the word order and the fact that Purpose Adjuncts do not as a rule contain non-controllable predications. For the related expression interest K.-St. (I.461) mention more examples with ut.

Note (end) return to page 6 The examples and the table do not include constructions occurring on the noun phrase level, such as difficilis + inf. pass. (TLL s. v. 1090.68).

Note (end) return to page 7 It is difficult to render this construction, unusual also in Latin, in English.

Note (end) return to page 8 For other types of interrogative sentences see TLL s. v. 2113.30 ff.

Note (end) return to page 9 I do not discuss instances like
- fore tempus cum is tribunus plebis … tuum monumentum … disturbaret (`That there would be a time when this tribune was to destroy your monument', Cic. Dom. 114)
It is not beyond doubt whether the cum-clause is a relative clause in the function Attribute with tempus (Pinkster 1972c: 177).

Note (end) return to page 10 I limit myself to the example given by K.-St. (II. 273).

Note (end) return to page 11 Examples of two-place constructions with the AcI of verbs such as imperare and iubere (`to order') are, for the sake of simplicity, given in crosssection 7.2.3. on three-place predicates.

Note (end) return to page 12 Normally the AcP only occurs in the case of direct perception, which is here feigned by the speaker. For more common examples see crosssection 7.4.6.

Note (end) return to page 13 For the perfect infinitive with verbs meaning `want', `can' see K.-St. I.133 ff.; with cupere Justin. 5.4.15. For the occurrence of the perfect infinitive with verbs meaning `have to' see Bolkestein (1980a: 161, n. 16). The occurrence of the perfect infinitive with the latter group of verbs is subject to a restriction which does not hold for the former two: the verb meaning `have to' must be part of an irrealis. See p. 236.

Note (end) return to page 14 In the passive perfect infinitive esse is often omitted, e.g. illam (patriam) extinctam cupit (`He looks for the ruin of this country', Cic. Fin 4.66). See K.-St. I.714, A4 and Orlandini (1986). The distinction between such instances where the copula is absent and Praedicativa (see chapter 8) is not always clear.

Note (end) return to page 15 K.-St. (II.424) also discusses cases like
- quod si assecutus sum gaudeo (`If I have attained that goal, I am glad', Cic. Fam. 7.1.6)
According to K.-St. si is used when the content of the subordinate clause is not a real fact. Cases of this type may also be explained by interpreting gaudeo as intransitive. (Examples in TLL s. v. 1705.47 ff.) This may possibly also hold for (30b) and (30e).

Note (end) return to page 16 curare has a two-place as well as a three-place frame, as is the case here.

Note (end) return to page 17 facere + infinitive is disputed in this text; Lambinus already proposed [faciendum]. See K.-St. II.235; TLL s. v. 104.53 ff.

Note (end) return to page 18 There are few other cases of facere + AcI. See K-St. I.694; Petersmann (1977: 213–15); TLL s. v. 115.37 ff.

Note (end) return to page 19 See p. 128.

Note (end) return to page 20 See p. 128.

Note (end) return to page 21 For active persuadere + infinitive see Nep. Di. 3.3 (K.-St. I.682).

Note (end) return to page 22 For the interpretation of gerund(ive) constructions as Manner and Instrument Adjuncts see Vester (1983). K.-St. explain (46) with ` = quos gessi honores petens', but it is also debatable whether gerundive and present participle can be put on a par, especially in Classical Latin.
{added 12-08: The use of the ablative of the gerund as an alternative of the present participle increases from Livy onwards. An interesting example can be found in Amm. 21.5.3, where the gerund is accompanied by an agreeing Praedicativum (see p. 154)
stando immobilis (&rsquostanding firm&rsquo).]

Note (end) return to page 23 K.-St. (I.751–2) distinguish the so-called `instrumental' and `modal' uses of the

-- 273 --

gerund(ive), subdividing `instrumental' and `modal' uses of the gerund(ive), subdividing `instrumental' into `Bezeichnung des Grundes' (`indication of the cause') and `Bezeichnung des Mittels' (`indication of the means'). Under the latter heading they also classify instances of what I would be inclined to call arguments with the semantic function Cause (for this semantic function see p. 29), e.g. - sum defessus quaeritando (`I am tired due to the searching', Pl. Am. 1014).

Note (end) return to page 24 These also occur in correlation with Purpose clauses. See (ix).

Note (end) return to page 25 See Lakoff (1968: 196 ff.).

Note (end) return to page 26 Negation of the predication of the gerund(ive) construction is not possible (so: *ad non …). See K.-St. I.751.

Note (end) return to page 27 The collection of K.-St. contains instances of infinitives both as argument (all instances in (c) and as satellite (all instances in (a). The instances in (b), verbs meaning `to give', `to allow', `to deprive' (including example (62)), ought perhaps also to be treated as arguments, like (c). For we also find the - normal - construction dare ut (= `to allow'; TLL s. v. do 1690.57). In TLL s. v. do 1688.59 a limited number of instances is given where do is found with an Object constituent as well as an infinitive. (In some cases it is not certain whether the accusative constituent is dependent on do or on the infinitive.) The most unequivocal instance is:
- quod iussi ei dari bibere … date (`What I have ordered to be given to her to drink, give it', Ter. An. 484)
For the Purpose infinitive see Perrochat (1932) and Pinkster (1987a).

Note (end) return to page 28 K.-St. (I.422, A.8) say that as a rule causa indicates Purpose, but often also Motive. The examples they give concern causa + noun, e.g. morbi causa (`on account of illness', Cic. Att. 12.13.2), timoris causa (`on account of fear', Caes. Civ. 1.33.1). Further research is necessary. [added 12-08: A rare example of gratia + Dominant participle is Liv. 40.56.3 palam facti parricidii gratia. I owe the reference to Heine (1990: 9).]

Note (end) return to page 28a [added 12-08: The English edition has the following example:
tribunus et quadringenti ad moriendum proficiscuntur (`The tribune and his 400 troops are leaving in order to die', Cato Orig. 83 P (= Gellius 3.7.11)) ]

Note (end) return to page 29 Sz. (642) calls this type of subordinate clause `rhetorische Pseudofinalsätze', just like those discussed in crosssection 4.1.4. A number of cases is discussed by Nisbet (1923), who points out that sometimes an (ironic) undercurrent is present, referring to, among others, Gerber & Greef (Lex. Tac. 1725a). They say `indicatur voluntas fati'. Cf. also the term `fatal' ut.

Note (end) return to page 30 In TLL s. v. ad (consecutivum) 545.77 we find a practically exceptional instance of ad + gerund(ive) used as a Result Adjunct:
- is est beatus, cui nihil humanarum rerum … intolerabile ad demittendum animum … videri potest (`He is happy, who cannot consider any of the things that regard mankind as unbearable so as to make him lose heart', Cic. Tusc. 4.37).

Note (end) return to page 31 R. Pfister (personal communication) has some doubt as to the Adjunct status of conditions, and suggests that they might be regarded as Disjuncts. The indeterminate status of si-clauses is alluded to by myself as well (Pinkster 1972c: 177–8), but will require further investigation. See also p. 122.

Note (end) return to page 32 For the term see K.-St. II.327.

Note (end) return to page 33 The editions incorrectly read ad pacem (there are three unequivocal cases of ad in Cicero).

Note (end) return to page 34 For juxtaposition of an argument clause (with quod) and a satellite clause (quoniam) see above (54) on p. 118. See also the ut-clauses in Cic. Cael. 8 (init.) and Austin's hesitant comments.

Note (end) return to page 35 Sometimes such an Agent is implicit, as in
- scripsi … ad librarios ut fieret … potestas (`I have written to the booksellers that there should be an opportunity', Cic. Att. 13.21A.1)
Somewhat more complicated is the following example:
- ita suasi seni atque hanc habui orationem, ut cum rediisses ne tibi eius copia

-- 274 --

esset (`I have persuaded the old man and made a speech to the following effect, that after your return you would have no access to her', Pl. Epid. 355).

Note (end) return to page 36 See Bolkestein (1980a: 10–11). See Sznajder (1987) for examples in Caesar.

Note (end) return to page 37 There are two other tests which often prove useful, viz. the substitution test and the passivization test (left out of account here). For the possibility of substituting an infinitive for ut-arguments with two- and three-place predicates see crosssection 7.4.4.

Note (end) return to page 38 See crosssection 7.2. for examples of the AcI functioning as Subject.

Note (end) return to page 39 The historical relation postulated by scholars who use the term in this way is discussed in crosssection 7.6.

Note (end) return to page 40 Cf. passive illud … te esse admonitum volo (`I want you to be informed of this', Cic. Cael. 8).

Note (end) return to page 41 An impossible sentence is e.g. *te admonui laudari. As for the notion `controllable', we sometimes find predications which I would call `non-controllable' in the sense adopted above (p. 17) embedded with verbs such as admonere. Thus, for instance, ianua patet is non-controllable, since the ianua cannot influence its being open or closed. People, however, can influence the door's being open or closed. This explains why we do find instances like te admonui ut ianua clauderetur; te admonui ut ianua pateat. See Bolkestein (1976a: 287).

Note (end) return to page 42 Other predicates which allow both constructions without a difference in meaning are: imperare, iubere (`to order'), hortari (`to exhort'), mandare (`to order'), poscere (`to demand'), etc. (Bolkestein 1976a: 164). [added 12-08: With many verbs the infinitive expression becomes frequent from the augustan poets onwards (K.-St. I.680ff.).]

Note (end) return to page 43 Note that the Complement illud is explained in the subordinate clause.

Note (end) return to page 44 The question-word test and the deletion test are less useful. The material of admonere happens to contain no instances of interrogative sentences which inquire about the content of the warning (quid me admones? :: venire). The infinitive with dicere certainly cannot be referred to in a question with quid. As for the omissibility of the infinitive with dicere and admonere, *dicunt me with dicere in the sense of `to say' seems to me completely ungrammatical, whereas instances like admonent me occur under certain circumstances (Bolkestein 1976b: 272–3). In cases such as me cum dico, te, Brute, dico (`When I say `I', I mean `you', Brutus', Cic. Orat. 110)dicere means `to mention', `to mean', in which case it is two-place (contra Sánchez Salor 1983: 33).

Note (end) return to page 45 For this group of verbs see Bolkestein (1976b: 287–91).

Note (end) return to page 46 Cf. cases such as Caes. Gal. 2.33.1 portas claudi militesque … exire iussit (`He ordered the doors to be closed and the soldiers to leave'), with coordination of the types exemplified by (94) and (95).

Note (end) return to page 47 There does not seem to be an exact parallel; cf. (38e) sedentibus valere dicebat (cf. vale). Better perhaps: si hoc iudici praescriptum lege aut officio putatis, testibus credere … (`If you think that a judge is bound to this by law or by his position, viz. to believe witnesses … ', Cic. Font. 22).

Note (end) return to page 48 See for these restrictions Bolkestein (1976a: 293–5; 1980a: 73–6).

Note (end) return to page 49 See for the verba sentiendi and their similarities and differences in comparison with the verba dicendi Bolkestein (1976a: 164; 1977d).

Note (end) return to page 50 The extent to which the two constructions are used with the same verb naturally also depends on the frequency of the verb itself. Thus, oblivisci (`to forget') is (happens to be?) found with the prolative infinitive only.

Note (end) return to page 51 For this instance see Calboli (1962: 6–155).

Note (end) return to page 52 For the popular opinion that the two constructions are historically related see crosssection 7.6.

-- 275 --

Note (end) return to page 53 Two-place iubere does allow a personal passive (see example (42e)).

Note (end) return to page 54 For the description of the construction see Bolkestein (1979; 1981a), in reaction to Pepicello (1977); Pillinger (1980) and Comrie's reaction (1981b); Wales (1982). Saltarelli (1976) does not help very much. Torrego (1987) offers a historical explanation: tu diceris ire (`you are said to go') has been formed owing to the influence of tu diceris bonus (esse) (`you are called good/said to be good'). This parallel structure, however, exists for only part of the verbs that are found with the NcI.

Note (end) return to page 55 Cf. Austin (1964) ad locum `a mere Grecism'. See Coleman (1975: 139–40). Eklund (1970: 78) argues that in the following instance tuentes may be replaced by an infinitive:
- excipiunt plausu pavidos gaudentque tuentes (`The Dardanians greet the bashful boys with cheers and rejoice as they gaze', Verg. A. 5.575 (translation H. R. Fairclough, Loeb edition))
In reality, this is a case of a Praedicativum (see chapter 8).

Note (end) return to page 56 For a late instance, due to Greek influence, with a verb of `knowing' see Eklund (1970: 77–8):
- sciebant … Deum suum semper tentantem eos (`They knew that their God was always leading them into temptation', Iren. 1.27.3).

Note (end) return to page 57 Sz. incidentally also mentions late instances such as audierat … eam aegrotantem (Vitae Patrum 3.33), where eam cannot be considered as being perceived.

Note (end) return to page 58 Cognoscere has the general meaning `to perceive', not necessarily `to see', cf. non auditu cognoscenda, sed oculis spectanda (`not to be perceived through hearing, but to be seen with the eyes', Tac. Dial. 8). See TLL s. v. 1501.59 ff. `sensibus percipere'.

Note (end) return to page 59 Cf. also the AcP with audire in TLL s. v. 1269.32 ff.

Note (end) return to page 59a [added 12-08: The preceding text is modified. The following text in the English edition was removed:
Instances such as (106) are incorrectly treated together with the AcP (K.–St. I. 704–5; Sz. 388):

(106) (Xenophon) facit … Socratem disputantem formam dei quaeri non oportere (`X. describes S. while arguing that one ought not to enquire after the form of God', Cic. N.D. 1.31) No perception is involved here. Replacement by an infinitive construction is impossible. Also, contrary to instances of the `true' AcP, the participle and the constituents governed by it are not omissible. Rather, this is an instance of three-place facere + Object + Object Complement (see p. 22).
The reasons for changing the text can be found on p. 22, note a and in the note added to the inserted example (56a) on p. 151. ]

Note (end) return to page 60 The term `dominant participle' was introduced by A. G. de Man, in his school grammar Accipe ut reddas (1965). A more common term is `ab urbe condita' construction. Recent descriptions in Bolkestein (1980b; 1980c; 1981b; 1983b). See also Lambertz (1982: 568–86).

Note (end) return to page 60a [added 12-08: For its position among other satellites (some resemblance to cum- clauses) and its distinctness from the so-called ablativus qualitatis see Lavency (1986). ]

Note (end) return to page 61 The complete passage is: ante conditam … condendamve urbem. For the difference between participle and gerundive see Bolkestein (1980b: 93). Constructions with prepositions are very frequent throughout Latin. A recent survey may be found in Helttula (1985).

Note (end) return to page 62 In many studies (among others Sz. 137; further literature in Bolkestein 1980b: 81) it is argued that the ablative absolute can be regarded as an extension of the `normal' uses of the ablative to mark noun phrases. To support this view, examples are adduced such as invocat deos … capite operto (`He invokes the gods with his head covered', Pl. Am. 1093). The very fact that the participle cannot be omitted constitutes an objection against this approach. I treat expressions like capite operto as Praedicativa ( crosssection 8.1.6.). See p. 146, Pfister (1988: 115 f.), Pinkster (1972b: 53; 1982) and chapter 8, note 2.

Note (end) return to page 63 For the notion `factivity' see Kiparsky & Kiparsky (1971). Vairel-Carron (1975: 84) uses the terms `actuel' : `inactuel'. A further subdivision of factive predicates is given by Vendler (1980). On the noun phrase level, too, the dominant participle construction practically always differs from the gerundive construction in terms of factivity. See p. 79 and example (55) on p. 118 (Blümel 1979: 102; Bolkestein 1980b: 90–4; Lambertz 1982: 568). For the differences between the alternative constructions see Bolkestein (1980b: 86–8).

Note (end) return to page 64 For a future participle see example (111). For the absence of restrictions on the construction see Bolkestein (1980b: 84).

Note (end) return to page 65 See Stockwell et al. (1973: 557 ff.) and for Latin the literature mentioned in note 54.

Note (end) return to page 66 I use the term Object constituent for the sake of brevity. In the examples I do not

-- 276 --

find an instance of an Agent of the embedded predicate (in casu invenire).

Note (end) return to page 67 Often it may be doubted whether a word should be described as a supine, e.g. in the instance tergis vulpium … quae tactu mollia … sunt (`Hides of foxes which are soft to (the) touch', Sen. Ep. 90.16), which is treated as a supine by TLL s. v. mollis 1380.67–8. It seems to me that it would be better to treat tactu as an ablative (limitationis) of the noun tactus (for metaphorical mollis see Verg. A. 12.25 haec haud mollia fatu (`These things that are not easy to say')). For the distinction see Kroon (1987; 1989b).

Note (end) return to page 68 See TLL s. v. dicere 969.58 ff.

Note (end) return to page 69 Noun phrase level!

Note (end) return to page 70 See Kroon (1989b) and Lambertz (1982: 542–4). [added 12-08: Lambertz (1993) describes the use of the second supine as in virtus difficilis inventu est in terms of dependency relations: semantically virtus is governed by invenire.]

Note (end) return to page 71 Debere means both `must' = `be obliged' (the so-called deontic interpretation) and `must' = `undoubtedly', `probably', cf. the professor must be ill (`they say that he is ill', `I conclude that he is ill') (the so-called inferential interpretation). Here, I limit myself to the deontic interpretation, see Bolkestein (1980a: passim).

Note (end) return to page 72 For arguments in favour of this analysis see Bolkestein (1980a: 120–33). Inadequate criticism can be found in Tesarova-Nováková (1988b: 203).

Note (end) return to page 73 See Oxford Latin Dictionary s. v. 5 `expressing a contingency or possibility'; for the time being I assume for posse a two-place frame.

Note (end) return to page 73a [added 12-08: For a different approach see Núñez (1991) - with too much critcism of Bolkestein (1980a).]

Note (end) return to page 74 Literature: Sz. 526; Scherer (1975: 235); Mallinson & Blake (1981: 432); Pinkster (1972c: 167–9); Pinkster (ed.) (1983: IX-X).

Note (end) return to page 75 Connectors are words such as autem, igitur, enim (see chapter 12). Intonation is one of the means by which hypotaxis may be indicated.

Note (end) return to page 76 See on ut and cum Meillet (1948: 159–74). For the difficulty of deriving conditional si-clauses directly from paratactic sentence combinations see Hettrich (1986). Christian Lehmann (1988b) rightly points out that there may have been non-finite embedded predications, while no clauses marked by subordinators existed. The views of Haudry (1973: 147 f.) and Pfister (1988: 112 ff.) are similar to mine.
[added 12-08: Hamp (1982) observes that most scholars assume that the conjunction ut is to be derived from one single ancestor. However the syntactic behaviour of final clauses on the one hand and consecutive clauses on the other in respect of negation is such that two ancestors are just as likely.]

Note (end) return to page 77 Scherer (1975: 235, note) does hold that Indo-European had subordinate clauses. For a possible derivation of satellite Purpose clauses from independent interrogative sentences with ut see Leumann (1940). Hettrich (1987b), however, argues for a derivation from relative clauses. See also Handford (1946: 24–7) for examples of hypotactic constructions and independent sentences with a parenthetical verb: habeas, licet/habeas licet (Pl. Ep. 471); scribas, vide, plane et probe/scribas vide plane et probe (Pl. As. 755). See also crosssection on p. 203.

Note (end) return to page 78 I give Blatt's examples (1952: 252–3).

Note (end) return to page 79 Sz. (344; 353) considers the infinitive as having originated from a Purpose Adjunct. Then also, however, it is not clear how it is that there is an identity relationship between the person that is being accused and the person that has accepted money.

Note (end) return to page 80 Other explanations of the AcI start from extension with an infinitive of the construction videre + accusative or vetare + accusative. Hahn (1950: 12) assumes that one of the origins of the AcI was a reanalysis of dico te bonum (`I call you good') dico te bonum esse (`I say that you are good'), see Torrego (1987: 79) and Calboli (1978: 209). Hettrich (1987a) shows that in Greek and Latin formally more or less identical expressions have resulted from different developments. See also Coleman (1985).

Note (end) return to page 80a [added 12-08: Recent discussions about the Accusative and Infinitive and its replacement in later periods by finite subordinate clauses are Cuzzolin (1991), Frédouille (1992), Haverling (1988: 240 ff.), Herman (1989), Scivoletto (1962).]

8 Praedicativum

Note (end) return to page 1 A large number of different terms is used to indicate what I call Praedicativum: in German `Zustandsattribut', `Koprädikativ' (Plank 1985); in French `apposition

-- 277 --

circonstancielle' (Väänänen 1951) and `complément propositionnel' (Longrée 1987); in English `subject/object adjunct', `appositional adjective' (Adams 1976b: 70–1), `secondary predication', `apposition' and `adverbial apposition'. For quantifying adjectives (omnes, etc.) the term Quantifier Floating is used. See Pinkster (1983a). Moreover, in Latin grammars many of the expressions discussed in this chapter are treated separately, under different headings. In the course of this chapter this will be pointed out. For the terminological chaos in German grammars see Heine (1972).

Note (end) return to page 2 In chapter 2 I mentioned the `do so test' as a means to determine whether constituents are omissible or obligatory. An interesting instance of a Praedicativum in this connection is:
- summa voce versus multos uno spiritu pronuntiare consuescebat; neque is consistens in loco sed inambulans atque ascensu ingrediens arduo (`He used to recite many lines with a loud voice in one breath, and this walking around and climbing steep slopes rather than staying in one place', Cic. de Orat. 1.261)
I consider the ablative absolute as an embedded predication which as a whole functions as Adjunct. Contrary to what is said here about the instances of Praedicativum, in the ablative absolute construction the participle, adjective or noun may not be omitted ([ERROR: no link cross:]section 7.3.7.). I also do not describe as Praedicativa instances such as primo in primo vere and medio in in medio horto (differently Happ 1977:49). A survey of this `partitive' use of adjectives is given in Vaughan (1942). On the potential ambiguity of such expressions see Amacker (1986). [added 12-08: Romero-Sangüesa (1993) gives examples of partitive adjectives with anaphoric and relative pronouns. ]

Note (end) return to page 3 For a possible attributive relation between Galli and laeti see crosssection 8.5.1. Rura colentes could possibly be Subject.

Note (end) return to page 4 E. Vester (1988) discusses relative clauses which may be interpreted as Praedicativum. See also Longrée (1987: 173, 183).

Note (end) return to page 5 The difficulty may arise from the contrast domino : servus. Bennett (1914: II. 7) distinguishes for this type of instance `the appositive in a temporal or adversative relation', cf. Pl. Capt. 630; 631; Mil. 966.

Note (end) return to page 6 For literature see the `Bibliographical information' at the end of chapter 6. For German see Neubauer (1977).

Note (end) return to page 7 Multus from Sallust onwards (e.g. multus atque ferox instare (`he attacked persistently and boldly', Jug. 84.1) and TLL s. v. multus 1608.83 ff.). see also Lundström (1982: 50–1). For nullus instead of non see Sz. 205, K.-St. I.236; e.g. is nullus venit (`he does not come', Pl. As. 408). See Pinkster (1983a) on quantifying adjectives.

Note (end) return to page 8 Examples of several uncoordinated Praedicativa in one sentence are e.g.:
- libens … navita Bosporum temptabo (`As a sailor, I will happily put the Bosporus to the test', Hor. C. 3.4.29–31)
- ubi sunt isti scortatores, qui soli inviti cubant (`Where are these whore-hoppers, who do not like to sleep alone', Pl. Am. 287)
- solane perpetua maerens carpere iuventa (`Will you pine away all your youth, lonely and sad?', Verg. A. 4.32)
- is … amens animi … dicitur … supplex orasse (`Distraught in mind he is said to have prayed as a suppliant', Verg. A. 4.203–5)
See also example (29). These instances are to be distinguished from instances of so-called zero coordination (asyndeton), such as
- ne valeat id, quod imperitus adulescens … ignarus invitus, sine collegis, sine libris, sine auctore, sine fictore, furtim, mente ac lingua titubante fecisse dicatur (`In order that no value be attached to what an unexperienced youngster is said to have done, in ignorance, unwillingly, without colleagues, without books,

-- 278 --

without a supporter, without a shaper, secretly, with faltering mind and tongue', Cic. Dom. 139)
Cf. Pl. Bac. 613; Hor. Ars 121.

Note (end) return to page 9 A neat instance of a material adjective is aeneus in:
- latus ut in circo spatiere et aeneus ut stes, … nudus nummis, insane, paternis (` … to stroll all along the circus and to stand as if made of bronze, robbed of your father's money, idiot', Hor. S. 2.3.183–4).

Note (end) return to page 10 With terminative (resultative) verbs the participle indicates the state of affairs resulting from the preceding action or process; often anteriority is not relevant. Example (20) can be interpreted in this way (see further chapter 11).

Note (end) return to page 11 Note the parallel construction of adjective, so-called ablative of description (see crosssection 8.1.7. below) and preposition phrase. Cf. also
- si … superior aut aequa manu discessisset (`If he had been victorious or had held his own', Sal. Cat. 39.4).

Note (end) return to page 12 Also fusi is a Praedicativum. Further examples of preposition phrases in the function Praedicativum may be found in Happ (1976: 302, note 590), TLL s. v. cum 1351.20 ff.; Merguet Lex. Reden Cic. s. v. sine. Neat examples in Bulhart (1956). For expressions with in and pro see Väänänen (1951), e.g. in larvam intravi (`I entered as a ghost', Petr. 62. 10).

Note (end) return to page 13 See Fugier (1978: 130).

Note (end) return to page 14 For the ablative of description see Pinkster (1982). The grammars do not distinguish clearly between the ablatives of manner and description. Possibly some would be inclined to describe the examples given here as ablatives of manner. I do not do so, because (a) there are instances of coordination (e.g. (31) and (32)), and (b) the expression may be paraphrased with a Subject Complement. These examples are also treated as instances of the ablative absolute. This, too, is incorrect in my opinion, because an ablative absolute refers to a state of affairs which obtains (or can obtain) independently of the state of affairs of the `main sentence'. Lexemes such as mens and caput, however, cannot occur independently, since there is always an implicit possessor. Strikingly enough, Sz. (70) treats instances such as (34) and (35) as `adnominal', but `ohne Beziehungswort' (without a Head constituent); following this view, we would presumably have to add a noun in the context, as the Head constituent of the genitive constituent.

Note (end) return to page 15 See TLL s. v. ire 637.10 ff.

Note (end) return to page 16 I have not found any examples of a Praedicativum with a satellite in the ablative (see also note 2). For further examples of a Praedicativum with an Attribute see:
- quorum omnium testimoniis de hac … pecunia confirmatum est (`By the testimonies of all of whom confirmation was given with regard to this money', Cic. Ver. 2.23)
- eorum vivorum (`Of them, while they were alive', Liv. 31.18.7)
Examples of a Praedicativum with an Agent in the passive:
- qui illum a se adulescente … auditum esse dicebat (`Who said that he had heard him when he was a young man', Cic. de Orat. 3.68)
- a quo uno (`By whom alone … ', Cic. Sest. 132)
See also Cic. Ver. 3.43.
For a Praedicativum with a Beneficiary see Verg. A. 6.764 longaevo; Pl. Am. 459.
For a Praedicativum with a preposition Complement see:
- qui de te oratore sic praedicaveram + AcI (`I, who made the following statement about you as an orator' (NB: the preposition phrase is a satellite), Cic. de Orat. 2.296)
- qui cum sciret me ex Mustio vivo audisse (`Although he knew that I had heard it

-- 279 --

from Mutius when he was alive' (satellite?), Cic. Ver. 1.139)
- civitatem, in qua ipsa florentissima multum omnibus gloria praestitisset (` … a state, in which, at the height of its prosperity, he had amply surpassed all others in fame', Cic. de Orat. 3.8)
- a quo vivo nec … rex Deiotarus quicquam aequi boni impetravit (`From whom, when he was alive, King D. has not obtained anyting good', Cic. Phil. 2.94)
- et quid ego de homine nato diutius (`And what else about someone born as a man …', Apul. Apol. 8)
See also Pl. Am. 524; Capt. 779; Mart. 11.5.11.
For more examples of Praedicativa with non-Subject constituents see K.-St. I.312, Bennett (1914: 5), Laughton (1964: 31), Lundström (1982: 28–9), Touratier (1980a: 400–8) and Pinkster (1983a). For the distribution of Praedicativa in Russian see Nichols (1982).

Note (end) return to page 17 On the noun phrase level preposition phrases can, of course, also function as Attribute, see crosssection 6.2.2.

Note (end) return to page 18 The so-called ablative and genitive of description can occur as Attribute on the noun phrase level, see crosssection 6.2.1.

Note (end) return to page 19 See Happ (1976: 294–5) and Heilmann (1973: 52 ff.). Happ, following Heilmann, goes to great lengths to provide all those adjectives which cannot occur as Subject Complement but can occur as Praedicativum with - sometimes highly artificial and rather unconvincing - paraphrases. See Pinkster (1983a).

Note (end) return to page 20 See TLL s. v. invitus 233.20 ff.; libens 1327.6 ff. There is one strange exception:
- (viri) qui semper malo muliebri sunt libentes (v. l. iubentes) (`Men that are always glad about female evil', Pl. Cist. 681).

Note (end) return to page 21 Examples are:
- intensifying adjectives: a clear failure, a true scholar, mere repetition, total destruction, a slight effort
- restrictive adjectives: a certain person, the same student, the only occasion
- `adverbial' adjectives: my former friend, a possible friend, an old friend
- denominal adjectives: a criminal lawyer, a woollen dress
For Dutch see De Jong (1979); see also Happ (1976: 112). For German see Bicker (1984).

Note (end) return to page 22 As Praedicativum, referring to a human Subject, we also find many adjectives which only occur as Attribute with non-human entities (e.g. adclinis (`disposed to', `leaning on'), arduus (`difficult', `steep'), avius (`straying', `out-of-the-way')). Adjectives seem to shift from Object to Subject constituents (`enallage'), see Görler (1982: 81).

Note (end) return to page 23 For the preference for adjectives in poetry see Axelson (1945: 62 f.) and Håkanson (1986). Material adjectives as Praedicativum are predominantly found in Virgil, Ovid and Horace (Anthonissen & Kater 1986). Since Schäfler (1884: 58) this use of temporal adjectives is ascribed to Greek influence, a view which the corrupt (Watt 1962: 118 ff.) noctuabundus from Cic. Att. 12.1.2. was thought to support.

Note (end) return to page 24 It is, of course, often difficult to distinguish between participle and adjective. See Eklund (1970: 18–25) and Vester (1977: 272–9).

Note (end) return to page 25 See Laughton (1964: 5–10; 20–31) and Vester (1983) for the classes of verbs.

Note (end) return to page 25a [added 12-08: The use of the active present participle (p.p.a.) in complex constructions, with object constituents, and of a wide range of verbs is typical of the literary style, where it spread under Greek influence. The participial expressions become more and more semantically independent information units with often a specific position. Bible translation is a specific subtype with its frequency of the erat + p.p.a. expression and its use of the p.p.a. in an anterior sense. This is also under Greek influence. The replacement of the p.p.a. by the gerund can be explained by the advantage of this expresion type as unmarked for agreement (Piccoli 1972). ]

Note (end) return to page 25b [added 12-08: For facere, fingere, pingere, etc. + Acc. + Part. see K.–St. I.704. See also p. 22, note a. I now think that this use of facere should be distinguished from the causative use, as in table 23 on p. 22. See also Heine (1990: 6–7).]

Note (end) return to page 26 For the Purpose interpretation of the future participle see Sz. 390, K.-St. I.761.

Note (end) return to page 27 For the use of the perfect participle and the gerundive as Praedicativum in relation to habere and the consequences for the Latin tense system see Pinkster (1987a).

Note (end) return to page 28 See for Dutch van den Toorn (1969: 36–7); for German Bartsch (1972: 142–3).

Note (end) return to page 29 A current paraphrase is Cicero coniurationem Catilinae detexit cum consul esset (Happ 1976: 297). This is inadequate, because it does not sufficiently reflect the fact that the lexeme consul contains new information. A better paraphrase would be

-- 280 --

Cicero consul erat cum coniurationem Catilinae detegeret, unless the sentence is meant to be interpreted as `it was Cicero, rather than someone else, who … '. See crosssection 8.6. on p. 161.

Note (end) return to page 30 Older grammars state that the present participle sometimes expresses anteriority or posteriority with regard to the time of the main predication. The following examples are given, of anteriority and posteriority, respectively:
- si ita factum esset, ut ille Romam veniens magistratus et sentatum Romae offenderet (`If it had happened in this way, that on his arrival at Rome he had found the magistrates and the senate of Rome', Cic. Att. 7. 12.5)
- habuit proficiscens dilectum in Umbria (`Upon leaving he conducted a draft in Umbria', Cic. Mur. 42)
Laughton (1964: 38–41) rightly notes that the temporal reference of Praedicativum and main predication are the same; the tendency to assume a difference in temporal reference arises from the fact that the states of affairs referred to by the Praedicativa are momentaneous, whereas the states of affairs referred to by the main predications are durative. See crosssection on p. 234 for tense, aspect and `Aktionsart'.

Note (end) return to page 31 See for German Helbig & Buscha (1974: 493). For Dutch Sassen (1982: 41–4).

Note (end) return to page 32 Perhaps we are dealing here with a poetical extension of the construction with reddere (see crosssection 8.5.4.); moreover, sterilis might be attributive (in that case, the word order might be odd).

Note (end) return to page 33 I have not found good Latin parallels for the English examples. There are instances such as coloribus tuis in
- (Britanniam) quam pingam coloribus tuis penicillo meo (`That I shall paint in your colours with my pen', Cic. Q. fr. 2.15A.2)
These can be interpreted either as Complement or as Praedicativum. See for such `ablatives of result' Sz. 127. See for English Quirk et al. (1985: 1197 f.), for Dutch Honselaar (1980: 50–2); TLL s. v. ago 1371.70 ff.

Note (end) return to page 34 See Blüher (1967: 29).

Note (end) return to page 35 Further examples are:
- apparent rari nantes in gurgite vasto (`A few are seen swimming in the vast abyss', Verg. A. 1.118)
- serpens … sibilat ore, arduus insurgens (`The snake hisses with its mouth, towering aloft', Verg. A. 11.753–5)
- … exspirasse ferunt, alteram … maestam sedentem domi … gaudio nimio exanimatam (`They are said to have died, one sitting at home in sadness having succumbed to too great a joy', Liv. 22.7.13)
- infesto venienti (`To him, while he was arriving as an enemy', Liv. 22.6.4)
Similar to these are exceptional instances of a Praedicativum with an attributive participle, such as in
- vitis albae viridis tusae suco impetigines tolluntur (`Scabies is cured by the juice of the white vine if it is squeezed while it is green', Plin. Nat. 23. 4; see also 20.131)
See K.-St. I.237; Sz. 172 incorrectly assumes Greek influence.

Note (end) return to page 36 See Laughton (1964); Vester (1977: 270 ff.); K.-St. I.771 ff.

Note (end) return to page 37 Note that, conversely, adverbs are sometimes said to occur instead of adjectives and/or participles in the function Praedicativum. Thus, e.g. TLL s. v. modus (quomodo) (1288.67 ff.; 1293.24 ff.). A good example is [Quint.] Decl. 270, p. 108.21 nunc miser filiam quomodo perdidi, followed by nec virginem, nec nuptam (some of the other examples are debatable). For the so-called adverbial character of Praedicativa see Pinkster (1983a).

Note (end) return to page 38 `Die lateinische Ausdrucksweise ist in solchen Fällen ohne Frage lebendiger, energischer und anschauliger, indem der nähere Umstand einer Handlung

-- 281 --

zugleich in die Persönlichkeit des Handelnden aufgenommen wird, als: Socrates venenum laetus et libens hausit.'

Note (end) return to page 39 See Menge (1965: § 21).

Note (end) return to page 40 Unless turbidus is attributive. So Mozley in his Loeb edition. Smolenaars (personal communication) gave me parallels of predicative turbidus with verbs of motion, e.g.
- (Mars) turbidus aetherias currus urguebat ad arces (`Mars frantically drove his chariot to the heavenly fortress', Stat. Theb. 3.222).
For the role of metric see e.g. Bednara (1906), Engel (1914), Priess (1909).

Note (end) return to page 41 See Pinkster (1983a).

Note (end) return to page 42 The editors of the Oxford Classical Text read trepide, following a part of the manuscripts.

Note (end) return to page 43 TLL s. v. cupidus 1425.37 classifies this instance as: `praedicative pro adverbio'.
Editors object to the exceptional form invite in
- invite cepi Capuam (`I have taken Capua against my will', Cic. Att. 8.3.4).
Wistrand (1979: 206–11), however, argues at great length that invite here specifies the way in which cepi Capuam was realized, rather than Cicero's frame of mind during the capere. Other instances of coordination of Praedicativum and Adjunct are:
- impigre neque multus (Sal. Hist. fr. 4.41M)
- raro invitique (Cic. Off. 1.136)
- bene potus seroque (Cic. Fam. 7.22)
- effusi et contemptim (Liv. 2.30.12)
- immaturo et unde minime decuit (Sal. Jug. 14.22)
See Lundström (1982: 28) and H. Vester (1987: 349–57) for further examples.

Note (end) return to page 44 For the semantic difference between adjective and adverb see Adams (1976b: 70–1) and Pinkster (1983a).

Note (end) return to page 45 See TLL s. v. opportunus 775.74.

Note (end) return to page 46 Interpretations of this type are, incidentally, also possible in the case of adjectives used as Attribute. Fugier (1978: 127) illustrates the similarity between adjectives and adverbs which she assumes by means of the following paraphrase: propraetor hominem semivivum reliquit = quomodo propraetor hominem reliquit? :: semivivum. I know of no examples of answers consisting of adjectives and participles used as Praedicativum to questions introduced by quomodo? (or, for that matter, qualis?). But see note 37.

Note (end) return to page 47 Laughton (1964: 30) calls this use of the participle `equivalent to a clause of purpose'. In reality a paraphrase with a Purpose clause introduced by ut is not equivalent: a paraphrase with a dum-clause would be more plausible. For the various possible interpretations see Laughton (1964: 7–9; 24–31).

Note (end) return to page 48 Laeti could, of course, be used `substantively' (`the glad ones'), but in that case it is itself a potential Head, rather than an Attribute.

Note (end) return to page 49 As was stated above, in poetry we find more adjectives used as Praedicativum; we also find the reverse. An example is:
- securaque lora tenebam (literally `I held the secure reins', Stat. Theb. 7.356)
Cf. securusque lora tenebam. This is an instance of enallage. Note that the Praedicativum construction is metrically impossible.

Note (end) return to page 50 For tests to distinguish between one or two constituents see Blüher (1967: 24–5); Bolkestein (1980b: 83–5); Fugier (1978: 124–8); Happ (1976: 286–7).

Note (end) return to page 51 Of course, puer in Cicero puer can very well be Apposition, e.g. when Cicero is a person completely unknown to the hearer/reader: `Cicero, a boy, … '.
K.-St. (I.246–7) do not make the distinction Apposition : Praedicativum, but always speak of `Apposition', which according to them, refers to the Predicate. In

-- 282 --

the following examples, the italicised constituents do not express a property the validity of which is limited to the duration of the state of affairs. I am not, therefore, inclined to describe them as Praedicativa. It is not yet clear to me whether we should consider them instances of apposition, and, if so, what the relation would be between these cases and the `normal' instances of apposition exemplified above.
- ferox rapidusque in consiliis ac lingua immodicus primo inter paucos, dein propalam in volgus pro cunctatore segnem, pro cauto timidum, adfingens vicina virtutibus vitia compellabat premendoque superiorem … sese extollebat (`Violent and hasty in his opinions and of unbridled tongue, he spoke of him, at first in the hearing of a few, but after some time quite openly to everybody, not as deliberate but as slothful, not as cautious but as timid, inventing faults that neighboured on his virtues, and exalted himself by disparaging his superior', Liv. 22. 12. 12)
- tragoedias primus in lucem Aeschylus protulit, sublimis et gravis et grandiloquus saepe usque ad vitium, sed rudis in plerisque et incompositus (`Aeschylus was the first to bring tragedy into prominence, lofty, dignified, grandiloquent often to a fault, but frequently uncouth and inharmonious', Quint. Inst. 10.1.66)

Note (end) return to page 52 The terms have been taken from Quirk et al. (1985: 1171 f.). K.-St. I.15 also regard the passive forms of habere, putare, etc. (see p. 22) as `Kopula-artige Verben'.

Note (end) return to page 53 `die eigentliche Aussage des Satzes'; see also Scherer (1975: 191, note 13).

Note (end) return to page 53a added 12-08: Hintzen, while initially taking participle constructions as `adverbials', later in her work gives a good characterisation of the differences between ablative absulute, participium coniunctum, and finite subordinate expressions (1993: 184). ].

Note (end) return to page 53b [added 12-08: Heine (1990:12) rightly points out that this example could be regarded as a sixth type of ambiguity, auctoremextinctum being and AcI with esse omitted. ].

9 Word Order

Note (end) return to page 1 An extreme defence of the alleged `free' nature of Latin word order may be found in de Neubourg (1977).

Note (end) return to page 2 Exceptions with regard to Latin are Hinojo (1985), Metzeltin (1987), Panhuis (1982), Väänänen (1987). See also Perrot (1978). A critical discussion of the research summarized in Sz. 397–410 is given by Panhuis (1982: 22–9).

Note (end) return to page 3 Example from Quirk et al. (1985: 49).

Note (end) return to page 4 See for German Hoberg (1981: 42). For Latin Purpose Adjuncts see Panhuis (1982: 127 ff.). On AcI and ut-clauses see J. R. de Jong, Werkgroep Ingebedde Predikaties, University of Amsterdam, 1982. On the AcI see also Bolkestein (1981a: 105).

Note (end) return to page 5 The technical term for the phenomenon occurring in (7c) is `extraposition'. For Latin examples resembling the phenomenon discussed here see Charpin (1977: 404). P. Masereeuw and R. Risselada (in J. R. de Jong (see n. 4) have found that out of a total of 95 instances of relative clauses 84 preceded and only 11 followed the finite verb (corpus: Caes. Gal.; Liv. 1–5, 21–25; Verg. A. 1–6; Cic. Catil.; Sal. Cat.; Pl. Amph.; Tac. Agr.). This observation of Masereeuw and Risselada confirms the point of view of Sz. (692) that is mentioned in the text. Vincent (1976: 60–3) - incorrectly, as far as we can tell - argues that Latin does opt for extraposition as a strategy to avoid complex relative constructions.
R. Risselada has examined 100 instances in Cic. Phil. and Orat. (indices Merguet s. v. ut) of ut-clauses as Attribute with nouns such as consilium, mos, etc. In 5 cases the ut-clause precedes the Head, in 4 out of the remaining 95 the whole of the Head + ut-clause is framed by other constituents. Ninety-one cases involve either placement of the whole complex at the end of the sentence (20) or discontinuity (71), in the sense that the ut-clause is placed at the end of the sentence. Here, therefore, discontinuity does predominate.

Note (end) return to page 6 J. R. de Jong: see note 4.

Note (end) return to page 7 A very obvious explanation for the high percentage of verbs in end position is the

-- 283 --

fact that in subordinate clauses the initial position is not available. In main sentences the initial position is available.

Note (end) return to page 8 For the relatively unclear notion `particle' see Pinkster (1972c: 135, note 2) and Schenkeveld (1988).

Note (end) return to page 9 The term `emphasis' (German: `Ton') is not further defined.

Note (end) return to page 10 Recent literature: Baldi (1979: 49–53); Calboli (1983: 117–31); Mallinson & Blake (1981: 402–17 + literature); Hawkins (1979); Miller (1975); Panhuis (1982: 110–12); Strunk (1977: 16–22). The differences of opinion may be explained from the fact that the extant Indo-European languages have widely diverging word order rules. As a result, it is very difficult to find a common denominator which can explain the actual data. This also holds for the question of whether Indo-European had prepositions or postpositions. see Baldi (1979). In general Strunk (1977).

Note (end) return to page 11 Percentages taken form Linde (1923: 154–5). For the topographical digressions see Panhuis (1981). He - unconvincingly - assumes Greek influence. A priori it may be assumed that in different text types sentences have different pragmatic structures. I have not yet been able to do research to confirm this assumption.

Note (end) return to page 12 On the position of Praedicativa see Pinkster (1983a: 204).

Note (end) return to page 13 This is obligatory in the case of relative pronouns, see Pennell Ross (1987: 95 ff.).

Note (end) return to page 14 The first influential study on so-called clitic words was Wackernagel (1892), on which see Wanner (1987: 79–85). [added 01-09: Adams (1994a; 1994b) shows that the words involved often follow constituents fulfilling Focus function (wherever these constituents are positioned themselves). See also Janse (1993; 1994).]
For the position of connectors, subordinators, relative pronouns, etc. see Marouzeau (1949: 67–135). For subordinators see Pennell Ross (1987). She shows that this phenomenon occurs especially with subordinate clauses placed at the beginning of the sentence and occurs more frequently with some subordinators than with others. In her corpus the instances of cum-clauses include cases where more than one constituent precedes the subordinator, e.g.
- consules dilectum cum aegre conficerent (`When the consuls completed the recruitment with difficulty', Liv. 25.5.5)
This multiple displacement serves the cohesion of the narrative (Pennell Ross 1987: 100 ff.).

Note (end) return to page 15 According to Comrie (1981a: 211) it is a relic. According to Dik (1989: ch. 11) relatively `light' constituents are preferably placed early in the sentence. See also Harris (1978: 18–23). He adduces phonetic arguments to explain why at a late stage in the development of Latin the OV order was introduced for unaccented pronouns (in his opinion, the VO order was still common in the fifth century, for such pronouns as well as nouns and noun phrases).
[added 01-09:  In the Vindolanda documents personal pronouns are placed almost always immediately next to the finite verb. The same has been shown for ClaudiusTerentianus (73:8). In Pliny  the Younger pronouns are more often separated than juxtaposed (57:43%). See Bowman et al. (1990: 46). In Egyptian papyri the same tendency of juxtaposition seems to exist, especially with Object personal pronouns following the verb. Its position with respect to the verb is still variable (hence it is not a clitic) (Molinelli & Ricci 1991).]

Note (end) return to page 16 For the first position of the sentence see Dik (1978: 178–83).

Note (end) return to page 17 This in itself is a serious argument against assuming a kind of fixed word order (SOV or SVO). See Pinkster (forthcoming a).

Note (end) return to page 18 Figure 9.1 contains all sentences of Cic. Att. 1.5. The sentence constituents are distinguished by means of horizontal brackets (an interrupted line indicates discontinuous parts of a constituent). In the notes at the end of the text the constituent occupying the first position (in the main sentence) is discussed. Connectors have been left out of consideration.

Note (end) return to page 19 Shackleton-Bailey translates sentences 7 and 18 as thematic.

Note (end) return to page 20 C. Kroon has examined (Research Assistant's Report October 1983) in Cic. Att. 7–12 the placement of names of members of Cicero's family, since at the end of his letters Cicero often relates, in a rather stereotyped way, how the members of his family are. In 60 out of 75 instances the name of the family member occupies the first position; almost all deviant instances can be explained on account of the fact that a Focus constituent occupies the first position.

-- 284 --

Note (end) return to page 21 Quintus frater has been counted as one constituent.

Note (end) return to page 22 For Topic constituents at the beginning of the sentence see Fraenkel (1968: Anhang 7). For the relation between Topic and Subject cf. Keenan (1976). [ERROR: no link :]22a [added 01-09: For constituents in sentence initial position see Jones (1991/2) ].

Note (end) return to page 23 See also Dressler (1969: esp. 14–16); Kroll (1916).

Note (end) return to page 24 Krenn (1984) explains the order in (20) on the basis of the fact that the Subject is indefinite. See above p. 271, n. 47.

Note (end) return to page 25 The other deviant instances are 1.5.8 (Subject; the text is suspect, see note a to sentence 26 in figure 9.1); 1.6.2 (Complement, unless certiorem facere must be considered a complex Predicate). The remaining instances are Focus (1.3.3, 1.9.2 (Subject Complement); 1.6.1 (Price)).

Note (end) return to page 26 This passage has also generated comments of a different nature, e.g. with regard to cold-heartedness or subtle brevity on Cicero's part. Time Adjuncts in Cicero do not have a fixed position in the sentence (R. Risselada, personal communication).

Note (end) return to page 27 If the final position of the sentence were exclusively reserved for Focus constituents, one might expect focalizing particles like quidem in the final part of the sentence as well. Quidem is not found in the final position, as far as I know. Giannecchini (1982) has made an interesting investigation into the behaviour of expressions like mentionem facere, which consist of a relatively empty verbal part and a full nominal part. His conclusion is that the nominal part tends to come after the verb, as in: quarum (virtutum) modo feci mentionem (`which virtues I have just mentioned', Cic. Fin. 3.25). My assistants I. Anthonissen and M. Kater have studied more or less synonymous expressions in Cicero's philosophical works, without reaching similar conclusions.
H. B. Rosén (Paper Societas Linguistica Europaea Freiburg 1988) suggests that noun phrases with a focalized (hence preposed) Attribute should be expected in the final position of the sentence. I have not been able to find a correlation between NP internal Focus and sentence Focus.

Note (end) return to page 28 The total number of instances is too small for such a subdivision according to type to be useful. R. Risselada (personal communication) has done similar research on Cic. Att. 2 (also c. 50 `simple' sentences), with similar conclusions. There were few instances of discontinuity in the material. In such cases I have scored the first constituent of a discontinuous phrase, e.g.
- sed omnis in tua posita est humanitate mihi spes huius levandae molestiae (Cic. Att. 1.17.4): conn. [S]-[C]-Vf-C-sat. 1-S.
Imperative sentences have been left out of consideration.

Note (end) return to page 29 In the sentences immediately following that cited here, the Objects (nos and Acutilianam controversiam) are in the first position, contrasting with comitia in that they represent entities in Atticus' activities that are different from the comitia.
- obieris Quinti fratris comitia. nos longo intervallo viseris, Acutilianam controversiam transegeris (`You will be present at the election of my brother Q. You will come to see me after a long time, and settle the problem with A.').

Note (end) return to page 30 Here, too, I have left out of consideration the imperative sentences and copulative and existential sentences. The material contains 16 non-complex copulative sentences. The following `rule' can be formulated for the order of Subject, finite verb and Subject Complement: S - SC - Vf, with the variants SC - S - Vf and S - Vf - SC. For the word order of esse and adjectives, nouns and participles see Warner (1980), who - in my opinion on the basis of too little material - assumes that as early as in the time of Plautus a shift had begun from S - SC - Vf to S - Vf - SC.

Note (end) return to page 31 Luisa Collewijn (unpubl.) applied the same method as used in this section to Celsus' De Medicina (1–6). Taking only declarative sentences with explicit S and O, without embedding, with only nouns or proper names as S and O, no anaphoric

-- 285 --

elements, and with simple finite verb forms (in total no more than 87 suitable sentences), the following order patterns were observed: SOV 51 instances; OVS 15; VOS 7; OSV 6; SVO 4). Here too, the SOV order predominates in frequency. Similar results were obtained for Vegetius' Mulomedicina (4th century ad) and the Mulomedicina Chironis.

Note (end) return to page 32 Panhuis (1982: 126).

Note (end) return to page 33 Panhuis (1982: 119).

Note (end) return to page 34 Against Panhuis' arguments it might also be objected that the informative value of a constituent diminishes with every constituent that precedes (the predictability increases). See Lyons (1968: 81 ff.).

Note (end) return to page 35 For a detailed study of presentative sentences in Romance see Ulrich (1985).

Note (end) return to page 36 So already Schneider (1912).

Note (end) return to page 37 In reality the picture is more complicated. With vir (nominative singular) Attributes are, indeed, predominantly placed after the Head, but with the other (di-and trisyllabic) forms of the words Attributes are usually placed before the Head. With rex (nominative singular) Attributes predominantly precede the Head; this also holds for the polysyllabic forms of this lexeme. See also de Jong (forthcoming.).

Note (end) return to page 38 Sz. (406): `objektiv-bestimmend (intellektuell-sachlich)'; Marouzeau (1953: 1): `discriminatif'.

Note (end) return to page 39 Sz.: `affektisch'; Marouzeau (1953): `qualificatif' (see note 40).

Note (end) return to page 40 For the inadequacy of the distinction between qualifying and determining adjectives - made especially by Marouzeau - for the word order on the noun phrase level, see also Ramat (1980: 189). He agrees with Waugh (1977) that in Romance placement of the adjective after the Head characterizes it as `restrictive'. See also Perrot (1978: 20). Fugier (1973) adopts Marouzeau's distinction `determining'/`qualifying'. For the order of Attributes in Cato see de Sutter (1986).

Note (end) return to page 41 See for such instances Nägelsbach & Müller (1905: 639–46) and Wilkinson (1963: 213–20).

Note (end) return to page 42 On this pattern see Bermúdez Ramiro (1985).

Note (end) return to page 43 See Solodow (1986).

Note (end) return to page 44 Such lines are sometimes called `Golden Lines', cf. Wilkinson (1963: 213–20). Habinek (1988) points out that there is a relation with a tendency to avoid `colon boundaries'. Pearce (1966) demonstrates that in hexametric poetry discontinuous placement of adjective and noun is used as a means to frame constituents and noun phrases.

Note (end) return to page 45 On the basis of examples of this kind a so-called `scrambling rule' has been developed for Latin in transformational grammar. This rule takes care of the mixing of constituents of sentences that have been generated in the deep structure. See Ross (1967: 41–3) and criticism by Boas (1975).

Note (end) return to page 46 See Aili (1979).

Note (end) return to page 46a [added 01-09: Word order is discussed in rhetorical treatises in the context of sentence rhythm and periodic structure. Cic. de Orat. 3.176 observes that making a sentence more rythmical may be compensated by a more flexible word order (immutatio ordinis). Nihil est enim tam tenerum neque tam flexibile neque quod tam facile sequatur quocumque ducas quam oratio. However, there are certain limits to be observed: Orat. 229: ne aut verba traiciamus aperte, quo melius aut cadat aut volvatur oratio, that is we should avoid too obvious hyperbaton. There is no evidence in our ancient sources for pragmatic factors correlating with word order on the NP level, and as far as the sentence level is concerned, only Quintilian's stressing the importance of the final position seems to point into the direction of pragmatic aspects of word order. For a survey of the ancient testimonies see Scaglione (1972), De Neubourg (1978a), who also criticizes Marouzeau's ideas, and Gutiérrez Galindo (1990). ].

Note (end) return to page 47 Greenberg (1963: 78–9).

Note (end) return to page 48 See [added 01-09: Gutiérrez Galindo (1990)], Lehmann (1979), and Wanner (1987: 391). On the relationship between word order and the structure of nominal compounds see Oniga (1988: 149ff.)

Note (end) return to page 49 On unaccented pronouns see crosssection Of the Romance languages, especially French is viewed as an SVO language. In others, especially Spanish, this order is much less predominant (see Lambrecht 1988 (`The "canonical" transitive clause of the SVO type … hardly ever occurs in actual speech') and Ulrich 1985).

Note (end) return to page 50 Instability according to Ramat (1980: 189). The SVO word order in colloquial Latin is assumed by Adams (1976a) and Panhuis (1982). Lakoff (1968) also assumed SVO for Latin, partly on the basis of the situation in the Romance

-- 286 --

languages. Incorrectly so, see Pinkster (1971: 390–4).

Note (end) return to page 51 Koll (1965) shows the continuity of (Classical) Latin word order.

Note (end) return to page 51a Panchón Cabañeros (1986) has statistical data concerning word order patterns in Caesar and Cicero, which lead him to the conclusion (: 227) that Classical Latin was neither OV nor VO, or rather that there are some correlations for OV and some other for VO:
- OV: V# (= Verb final), Vprin.-Vaux., Adj.N, Gen.N.
- VO: PrepN, Nrel, Comp.Stand.
As for the AN ordering, this is in contrast with Adams (1976) statements and in accordance with Rubio's statements (1976). The Gen.N ordering holds especially for genitive pronouns, but does not hold for NgenN combinations: For these combinations NGen. is more frequent, especially in Caesar. As for combinations of N, Adj. and Gen., it is not true that the Gen. is preferably placed in between the other two, as Adams (1976: 80) states. The statistical data are based on Caes. Gal. I and Cic. Mil.
N.B. Verb final is defined as the last position in its clause independent of whether relative or other subordinate clauses follow.

10 Sentence type, illocutionary force and mood

Note (end) return to page 1 (tu) is in parentheses, because the explicit mention of the addressee of an order is exceptional (but see Gonzalez Lodge (1924), Lexicon Plautinum, s. v. tu 739B and example (4)).

Note (end) return to page 2 Other terms: `mand', `command', `imperative', etc. See Bolkestein (1980a: chapter 5).

Note (end) return to page 3 By `imperative sentence' I also mean the main sentence of a conditional period, e.g. ignosce in
- si quid in te peccavi …, ignosce (`If I have done wrong against you, forgive me', Cic. Att. 3.15.4)
The imperative exceptionally occurs in relative clauses (K.-St. II. 309).

Note (end) return to page 4 There are, of course, other means to make the speaker's attitude explicit, e.g. the `attitudinal Disjuncts' such as fortasse, iure treated in chapter 4.

Note (end) return to page 5 Scherer (1975) is one of few scholars who do systematically discuss the various sentence types (in his chapter VI, `Satzarten'). He remarks, without using the notion illocutionary force, that sentence types can be used indirectly, i.e. with another function than they `normally' have. As an example he mentions the so-called rhetorical question: such an utterance is formally a request for information, but intended as a statement or an exhortation (146–7).

Note (end) return to page 6 See Rubio (1968); Scherer (1975).

Note (end) return to page 7 See Vairel-Carron (1975: 117). She mentions one exception:
- nam mihi quidem hercle qui minus liceat deo minitarier populo … quam servolo (`For why, by Hercules, should I, a god, not be allowed to threaten men just like a slave', Pl. Am. 986–7)
In her view, this is a rhetorical question. It might be better to say that hercle emphasizes mihi (as it often does).
Bolkestein (1980a: 70–1) points out that so-called inferential utterances (e.g. `He has got to be ill') cannot be used in interrogative sentences.

Note (end) return to page 8 See Calboli (1968: 467); Lehmann (1973: 21–34); Löfstedt (1966: 12–20); Prat (1975); Scherer (1975: 76–7); Sonnenschein (1910: 7); Touratier (1977: 372). For a discussion of exceptional occurrences of non see Löfstedt (1966: 12–20) and Pinkster (1986a).

Note (end) return to page 9 ne … igitur in Cic. N. D. 1.88 is exceptional.

Note (end) return to page 10 See Löfstedt (1966: 111) and reservations in Touratier (1977: 394 ff.).

Note (end) return to page 11 Examples in Löfstedt (1966: chapters 4–5).
[added 01-09: On the use of the indicative in directive utterances see Petersmann (1977: 204-205), who among other examples discusses Petr. 33.2 permittitis tamen finiri lusum (v.l. permittetis) (Trimalchio speaking). See also Soffritti (1963: 57), who discusses a few other disputable examples from epigraphical and literary texts, as well as many cases in the Incantamenta magica and several in Apicius.
On the `präskriptive Funktion' of various non imperative expression forms in medical texts see Önnerfors (1989: 134). ]

Note (end) return to page 12 See Pinkster (1972c: 138–41) and Scherer (1975: 153).

Note (end) return to page 13 I leave out of consideration the (historical) infinitive (see crosssection 11.3.4).

Note (end) return to page 14 See Lyons (1977: 816). He distinguishes `counterfactive', `non-factive', `factive'. Perret (1964: 209–10) describes the difference between present and imperfect subjunctive as `actualisation médiocre' : `actualisation faible'.

Note (end) return to page 14a [added 01-09: Gilmartin (1975) discusses the use of the second person singular subjunctive verb form in Roman historians from Sallustius to Ammianus (putares, cerneres, etc.). ]

Note (end) return to page 15 See Handford (1946: 32–3). In other languages a translation with the pluperfect is often more to the point; cf. my translation of (15). K.-St. (I.180) point out that in Latin the pluperfect subjunctive cannot be `potential', but is always an `irrealis'. It is, however, practically impossible to make this distinction. See also crosssection

Note (end) return to page 16 See Vairel-Carron (1980: 123). She mentions the following passages: Pl. As. 491; Capt. 53; Epid. 528; Rud. 305 (in a subordinate clause).

Note (end) return to page 17 For this view see Thomas (1938: 309).

Note (end) return to page 18 See Vairel-Carron (1978; 1980). She explains the element of `downtoning' (`moindre actualisation') on the basis of the fact that the perfect, being the tense of

-- 287 --

anteriority, by implication indicates `past tense'. She remarks that in many languages `past' tenses are in reality used for non-actual matters; cf. Pinkster (1983b: 299).

Note (end) return to page 19 See Pinkster (1986a). K.-St. (I.177) mention as examples of the second person Cic. Leg. 3.1 laudaveris (`You will have praised') and Cic. Fam. 11.24.2 non erraris (`you will not have erred'), but these are, rather, instances of the future perfect. I do not know any other instances. Furthermore, one notes that the number of instances of dixerim without a negation and/or words such as facile is very small. In brief, the expression seems highly idiomatic. The example mentioned in the grammars, Cic. Marc. 4, contains a future form: tamen adfirmo et hoc pace dicam tua, followed by an AcI, a construction that we do not find with pace tua dixerim (`And yet I assure you and will tell you, by your leave, that … ').

Note (end) return to page 20 On the low frequency and gradual disappearance of -ne see Janson (1979: 104). On the differences between questions introduced by these particles in terms of discourse cohesion see Bolkestein (1988a) and Finke (1987).

Note (end) return to page 21 As is well-known, Latin does not have words for yes and no. Instead, we find expressions such as (non) ita, ita est; sane, vero, etc. See K.-St. II.531–2; Pinkster (1972c: 138–41); Thesleff (1960).

Note (end) return to page 22 According to Hoff (1983), requests for information and assertive questions differ in terms of the question words used:
Request for informationAssertion
cur nonquidni; quin
I am not sure about this.
On `rhetorical questions' see further Calboli (1981); Hoff (1983: 123–5), Orlandini (1980: 111–12); Scherer (1975: 166–7); for English Quirk et al. (1985: 1478–9); for German Schmidt-Radefeldt (1977: 381–3). Following Schmidt-Radefeldt (1977), Orlandini (1980) points at the frequent occurrence of comparative constructions (quam; also nisi). The distinction between requests for information and rhetorical questions was already made in antiquity (`percontatio' : `interrogatio'). For ancient views on the effect of rhetorical questions see Calboli (1981: 135–6), who cites, among other passages:
- interrogatio non omnis gravis est neque concinna, sed haec, quae, cum enumerata sunt ea, quae obsunt causae adversariorum, confirmat superiorem orationem (`Not all interrogation is impressive or elegant, but that interrogation is, which, when the points against the adversaries' cause have been summed up, reinforces the argument that has just been delivered', Rhet. Her. 4.22)
A good example of the failure of a rhetorical question is given by Seneca the Elder (Contr. 7 praef. 7) (see Orlandini 1980: note *).

Note (end) return to page 23 See also Löfstedt (1966: 188–91) and Scherer (1975: 160).

Note (end) return to page 24 One of the proposals found in the abundant literature on this line is to consider dono a relic of an old subjunctive form in -o.
[added 01-09: For the use of the indicative mood in sentences like quid ago see Alfonsi (1965). ]

Note (end) return to page 25 See K.-St. I.181, where instances are cited such as
- vis vocem huc ad te? (`Should I summon him to you?', Pl. Capt. 360)
K.-St. view the use of non in sentences of this type as `Begriffsnegation' (see above p. 192).

-- 288 --

Note (end) return to page 26 The text is not entirely certain, but this is not relevant for the problem under discussion. See for the example Ernout & Thomas (1953: 379); Lehmann (1973: 113); Scherer (1975: 153).

Note (end) return to page 27 An exception is:
- nolito edepol devellisse (`Stop plucking them, by God', Pl. Poen. 872)
See also example (50) on p. 199. Combinations with obsecro, quaeso, etc. do occur.

Note (end) return to page 28 A more detailed subclassification is given by Risselada (1989). A meticulous treatment of directive expressions in German is given by Hindelang (1978).

Note (end) return to page 29 See also Handford (1946: 86); Lehmann (1973: 172); Löfstedt (1966: 11); Touratier (1977: 379–81), in different terms. For the restrictions in general see Bolkestein (1980a: 43 ff.).

Note (end) return to page 30 Sometimes we find the negation non (Scherer 1975: 159). The borderline between sentences of this type and counterfactual expressions is not clear (in fact,the parts in (31) not cited here also include si-clauses).

Note (end) return to page 31 See for -dum Janson (1979: 105); TLL s. v. ago 1403.80 ff.; s. v. dum 2201.8 ff.; s. v. modus 1300.19 ff.; K.-St. I.199–202.

Note (end) return to page 31a [added 01-09: In those cases in which tu is not used to mark a change of Addressee and/or Agent the use of tu with directive expressions is limited to authoritarian situations (Risselada, unpubl.). ]

Note (end) return to page 32 Lehmann (1973: 165–70) elaborately discusses the distinction between `optative' and `volitive' sentences, calling wishes only those imperative sentences that are introduced by utinam.

Note (end) return to page 32a [added 01-09: It is strange to read in Núñez (1991: 220) that I should not be aware of the fact that wishes may contain States and Processes (but see also his page 222). ]

Note (end) return to page 33 Some manuscripts offer a different text.

Note (end) return to page 34 Imperative sentences with a concessive illocutionary force are rarely found in the second person. K.-St. I.190 interpret Ter. An. 892 viceris as concessive. Pl. Men. 295 perieris and Verg. A. 11.153 velles are regarded as wishes (I.183–4).

Note (end) return to page 35 For the infinitive in directive utterances see Sz. 266 and Löfstedt (1966: chapter 6).

Note (end) return to page 35a [added 01-09: I leave out of account the so-called imperative use of the infinitive, instances of which are found in non-literary texts (K.–St. I.666; Sz. 366)) and the use of the indicative (see (72) below). ]

Note (end) return to page 36 Facias is seen as a milder form by Thomas (1938: 123–6) and Vairel-Carron (1975: 128; 237–8). Often the second person is in reality an indefinite person (`one'), certainly in Cicero (K.-St. I.185).

Note (end) return to page 37 Mildness is not involved according to K.-St. I.186; Sz. 335; Handford (1946: 42); Löfstedt (1966: 113) and Touratier (1977: 376–7).

Note (end) return to page 38 Touratier (1977: 376–7) mentions as an example
- (Cicero to Terentia) si est spes nostri reditus, eam confirmes et rem adiuves. sin, ut ego metuo, transactum est, quoque modo potes, ad me fac venias (`If there is some hope for my return, support and further it. If, however, as I fear, all has been decided, come here in whatever way you can', Cic. Fam. 14.4.3)
It would, however, be misleading to describe fac (venias) as a normal imperative. Other examples: Pl. Bac. 416–8; Mil. 1099–1102; Truc. 432–3; Ter. Hec. 753–5; Liv. 3.48.4; Verg. A. 4.495; Prop. 1.1.31; Tib. 1.1.37. More instances in Blase (1903: 124–5).

Note (end) return to page 39 In pre-classical Latin there are some instances, but in comparison with the total number of orders very few; in Cicero and others highly limited.

Note (end) return to page 39a [added 01-09: Risselada (1993: 138–55) shows that the 2nd person present subjunctive is more likely to be stronger than the `normal' imperative which can be used in all sorts of pragmatic situations. The 2nd person present subjunctive is more contextually embedded than the imperative. ]

Note (end) return to page 40 See, apart from Vairel-Carron (1975: 237), who is reacting against Riemann (1886), also Löfstedt (1966: 22–7) and in general Lyons (1977: 746–79). Arguments in favour of the future character are: temporal adverbs; occurrence with relative clauses; occurrence in series with the present imperative in order to indicate a later moment; occurrence in main clauses with a si-clause with future indicative (si rogabit, respondeto/respondebis). See also K.-St. I.196; Sz. 340–1; Lebreton (1901: 195–200).

Note (end) return to page 41 See Vairel-Carron (1975: 239–44; 258–60). She notes that some grammars incorrectly assume that Cato's use of-to-forms is similar to that in juridical texts. In reality, Cato addresses his son, in the second person.

Note (end) return to page 41a [added 01-09: Risselada shows that the -to imperative is used in temporally and conditionally less immediate situations, frequently directed to `general' second persons (1993: 122–38). ]

Note (end) return to page 41b [added 01-09: Risselada (1993: 155–8) shows that the subjunctive perfect prohibition is the most neutral counterpart of the `normal' imperative. Noli is almost exclusively used in requests as well as in advisory directives, which somehow reflects the original lexical meaning of this expression. ]

Note (end) return to page 42 See Vairel-Carron (1975: 189 ff.). She mentions as a seeming exception Pl. Mer. 321. For further literature on this view see Calboli (1983: 95). Some linguists argue

-- 289 --

that the type ne fac expresses a permanent prohibition, but see Löfstedt (1966: 61).

Note (end) return to page 43 In Plautus and Terence most instances of ne facias are pseudo-Purpose clauses (see p. 34). In Cicero there are only three instances of ne facias, and Lebreton (1901: 302) even tries to explain them in a different way. Coleman (1975: 133) is of the opinion that ne facias is typical for Cicero's letters and ne feceris for his `formal writings', but this cannot possibly be determined in view of the limited frequencies, see Calboli (1966: 294).

Note (end) return to page 44 Perfect subjunctive categoric: see Löfstedt (1966: 127–31); Thomas (1938: 123, 133); Vairel-Carron (1978: 308–9; 1981a). Aspectual difference: see Handford (1946: 45); Löfstedt loc. cit.; Sz. 337; Thomas (1938: 117–19). Greek influence: Thomas (1938: 134). Thomas also assumes influence of metrics and clausula.

Note (end) return to page 45 Löfstedt (1966: 75) mentions an instance in which two types of prohibition occur in the same sequence:
- sta ilico, noli avorsari neque te occultassis mihi (`Come here, stand still, do not turn away and do not hide from me', Pl. Trin. 627).

Note (end) return to page 46 The distribution of ne feceris/noli facere in Cicero's letters is as follows:
Cic. Att.; Q. fr.Fam.
ne feceris225
noli + inf.1813

Note (end) return to page 46a [added 01-09: Risselada (1993:296–300) defines the noli-prohibition type as `less authoritarian' than the ne feceris type. ]

Note (end) return to page 47 See Hoff (1983: 126–7).

Note (end) return to page 48 Sometimes only an infinitive, if the Subject constituent is known from the context.

Note (end) return to page 49 Vairel-Carron (1975: 100–4) cites some exceptional instances of an evaluative adjective as Subject Complement in the AcI, always amplified by adeo … ut, e.g.
- adeon me fuisse fungum, ut qui illi crederem (`That I was so stupid as to trust him', Pl. Bac. 283).

Note (end) return to page 50 See Vairel-Carron (1975: 24–31; 116–8). Note that the subtype with the AcI resembles interrogative sentences (particularly sentence questions), owing to the possibility of adding -ne and the restrictions on expressions such as edepol (see Vairel-Carron 1975: 113–38).
[added 01-09: The exclamative infinitive is relatively frequent in Terence (54 examples). Plautus (33 exx.) and Cicero (61 exx.) also have a considerable number of instances. It is much more rare in later authors (Livy has 15 exx., 10 of which in indirect speech). Melzani (1974) counted ca 200 examples in the literature up to Apuleius. The Subject is lacking in 45 instances. The particle -ne is found in 88 instances, almost always attached to pronouns or adverbs. Future infinitives are rare (4 examples) and among the perfect infinitives passive ones are in the minority. The exclamative sentence covers al sorts of emotions. Often one or more elements of the exclamative sentece are (mutually) contrastive. ]

Note (end) return to page 51 See Sonnenschein (1910: 4) and Touratier (1977: 378–9).

Note (end) return to page 52 See also Löfstedt (1966: 103; 112) and Scherer (1975: 229–31).

Note (end) return to page 53 See Bolkestein (1977c: 67–8); Lyons (1977: 738) and Risselada (1989). These `parenthetical verbs' naturally also allow `normal' embedded predications, e.g. obsecro ut … (see crosssection 7.6.2. on p. 139 ff.).

Note (end) return to page 54 These verbs occur only in the first person singular of the present indicative. This use is called `performative': when someone says `I promise' he also makes a promise. The notion was introduced by Austin (1962).

Note (end) return to page 54a [added 01-09: For this class of parenthetical verbs see Haverling (1988: 228ff.). Rogo as a parenthetical verb replaced oro, quaeso, and obsecro in colloquial Latin of the early empire. Examples in imperative sentences with the imperative can be found in Petronius (67.1), Martial (2.14.18) Claudius Terentianus (CPL 252.17) and Vind. (Inv. 88/946, 13 and other). See Bowman et al. (1990: 47).
For a discussion of predicates that are used as deontic illocutionary verbs see Job (1992). He distinguishes four types: (i)the class of iubere and imperare (occasionally also praecipere, and a few other); (ii) that of vetare and interdicere); (iii) that of permittere, and, rarely, (iv) solvere, absolvere, remittere. ]

Note (end) return to page 55 [deleted 01-09: The idiomatic nature of quaeso also appears from the fact that we hardly ever find other forms of the same lexeme (only Cic. Leg. 1.6 quaesumus). ]

Note (end) return to page 56 (66) has been taken from Calboli (1966: 264–5); (67) from Handford (1946: 58). Both examples are debatable: in (66) the sit-clause could also be adhortative. In that case, par videtur has a directive illocutionary force. With regard to (67) it may be noted that oportet is not always `directive' (Bolkestein 1980a).

Note (end) return to page 57 See Bolkestein (1980a: 41; 117; 131). But see also note 7.

Note (end) return to page 58 For English see Brown & Levinson (1987).

Note (end) return to page 58a [added 01-09: See now Risselada (1993). ]

Note (end) return to page 59 For a gradation of this kind see Franck (1980: 121).

Note (end) return to page 60 Cf. English John's coming home tonight $ John's coming home tonight (isn't he)? $ Is John coming home tonight?. For certe `interrogantis vel opinantis' see TLL s. v. 930.63 ff. There are no convincing instances of sentences with this certe and an

-- 290 --

interrogative particle (-ne, num).

Note (end) return to page 61 Further instances in Gonzalez Lodge (1924), Lexicon Plautinum, s. v. non, p. 193 (end). In interrogative sentences with a directive illocutionary force non is very frequent, nonne very rare, -ne quite frequent.

Note (end) return to page 62 See also Handford's distinction between a `should/would', a `can' and a `may' potential subjunctive (1946: 99–115).

Note (end) return to page 63 See also Scherer (1975: 67–81).

Note (end) return to page 64 Sz. also mentions the Indo-European injunctive; Scherer (1975: 85) notes that the Latin subjunctive has also adopted some uses from the Indo-European indicative.

Note (end) return to page 65 A reason for classifying the deliberative subjunctive in this way is the occurrence of expressions with debere that are at first sight comparable (Sz. 338). See for the difference between debere and the jussive subjunctive Bolkestein (1980a: 38–40).

Note (end) return to page 66 `etwas als gedacht, als vorgestellt bezeichnen'.

Note (end) return to page 67 See Lakoff (1968: 157) and Scherer (1975: 76).

Note (end) return to page 68 See Sonnenschein (1910: 19–20; 54–8). Related is Handford's (1946: 29) approach.

Note (end) return to page 68a [added 01-09: For a critical discussion of the various theories with respect to the subjunctive see Núñez (1991: ch. 6). ]

Note (end) return to page 69 This table is rather like that of Rubio (1968: 86), but he assumes that the indicative also occurs in the imperative sentence type; incorrectly, since ne facit does not exist.

Note (end) return to page 70 See Orlandini (1980: 121 ff.; 132 ff.)

Note (end) return to page 71 This argumentation could also be reversed: the use of the subjunctive is superfluous if the meaning of the governing predicate clearly defines the status of the subordinate clause. In this one might also see the cause of the disappearance of the possibility of opposition of moods in later Latin (see Bailard 1980: 9). For the semantic motivation of the subjunctive in subordinate clauses see Kiss (1982: 42–50).

Note (end) return to page 72 For the difficulty in explaining ut (non) with verbs meaning `to happen' see Haudry (1973: 163).

Note (end) return to page 73 See for this Stephens (1985).

Note (end) return to page 74 There are several similarities between Purpose satellite clauses and embedded imperative predications, e.g. the restriction on tense (not, however, that on controllability).

Note (end) return to page 75 Adams (1976b: 98), on the other hand, remarks that in subordinate clauses the subjunctive seems to be on the decrease, pointing at the use of the indicative in subordinate clauses within the AcI in Anonymus Valesianus II.

Note (end) return to page 76 See Perret (1964: 209–19).

Note (end) return to page 77 For relative clauses in the subjunctive see Lambertz (1982: 465–72), Lavency (1981a), Maurais (1980), Scherer (1975: 250), Touratier (1983b) and E. Vester (1988).

Note (end) return to page 78 For the factors that determine the choice between qui and ut see Elerick (1985).

Note (end) return to page 79 Scherer (1975: 250) calls these instances `erläuterende Relativsätze', in contrast with `Konsekutivsätze'.

Note (end) return to page 80 Iordache (1977: 265–8) mentions the following contextual factors that indicate a causal interpretation: evaluative adjectives (stultus (`stupid'), miser (`unhappy')), adverbs (male (`badly'), iniuste (`unjustly')) and nouns (flagitium (`disgrace')). Often we find an emphatic pronoun tu or intensifying particles such as quidem. The main sentence is often a question or an exclamation. The relative clause sometimes contains particles such as ut, utpote, quippe. In the (factive!) perfect the indicative predominates.

Note (end) return to page 81 See Iordache (1977) and Touratier (1980a: 344–5).

-- 291 --

11 The Latin tense system

Note (end) return to page 1 The text of this chapter is largely based on Pinkster (1983b), to which I refer as little as possible.

Note (end) return to page 1a [added 01-09: In the Spanish edition the verb 'to read' is replaced by 'to ask (pedir)'. ]

Note (end) return to page 2 The sentence can possibly be interpreted as `in three hours the boy read a (whole) book, from cover to cover'.

Note (end) return to page 3 E. Vester (1983: 23–5) has not found instances of desinere/desistere with momentaneous states of affairs. In English (and Dutch) a test with `almost' may be used in order to distinguish terminative and non-terminative predicates. See Pinkster (1983b: 281).

Note (end) return to page 4 See for the classification of states of affairs in general Dowty (1979); see for Latin also E. Vester (1983). De Groot (1983) contains critical remarks as to the classification of states of affairs given here. See also François (1985). Serious doubts as to the possibility of a well-defined classification of states of affairs may already be found in Pfister (1936).

Note (end) return to page 5 For the term Aktionsart see below p. 223. Sz. also distinguishes a `perfektische' Aktionsart.

Note (end) return to page 6 In Sz. and most older treatments preverbation is considered a productive means to change the Aktionsart of the verbum simplex (Sz. 304). The most far-reaching approach is that of Barbelenet (1913). A good refutation is van der Heyde (1926). For active: passive see E. Vester (1985).

Note (end) return to page 7 A survey of the research is given by Sz. 302–4. Critical remarks in Pinkster (1983b: 284–5). For the classes of verbs that in Caesar occur in imperfect or perfect see Reinhold (1956).

Note (end) return to page 8 The past tense is used as a polite form in many languages. See Brown & Levinson (1987). In Dutch child language the past tense is, for example, used for fictitious, `acted' situations (Dan was jij koning (`Then you were the king')). A number of the effects of the use of the imperfect is discussed by Mellet (1987; 1988: chapter 13).

Note (end) return to page 9 More fully: that which is expressed by the predication.

Note (end) return to page 10 Exceptions will be dealt with below.

Note (end) return to page 11 For two-dimensional systems of this kind and the redundancies present in them see Bull (1960) and King (1983).

Note (end) return to page 12 See Ruijgh (1971).

Note (end) return to page 13 Examples (19a-b) and (20a-b) (and their translations) are from Rijksbaron (1984: 46, examples (93) and (95); 102–3, examples (244) and (243)).

Note (end) return to page 14 See Moralejo (1988); Serbat (1975: 384–5; 1988a); Touratier (1983a). [added 01-09: For a repeated exposition of the arguments in favour of regarding the present tense as a-temporal see Serbat (1991).  ]

Note (end) return to page 15 See Pinkster (1985c: 187–8).

Note (end) return to page 16 I do not further define this term. See also crosssection 12.1.

Note (end) return to page 17 Virgil closes some books of the Aeneid with a historic present, among which the last book. I regard the end of a book as a boundary signal.

Note (end) return to page 18 According to K.-St. (I.177–8) quaerat aliquis/quispiam does not occur. Also with dicere, fortasse dixerit quispiam (Cic. Sen. 8) is more common.

Note (end) return to page 19 The semantic correlation between present subjunctive and future indicative is reflected in the historical developments. Probably the Latin e-future in the third and fourth declensions is formally a continuation of the Indo-European subjunctive (Leumann 1977: 573). A similar correlation may be observed in the development of Latin to Romance (see Fleischmann 1982 and Lyons 1977: 816 ff.).

Note (end) return to page 20 In Cicero the use of the future perfect without anterior meaning is limited. Most are of the type de me videro. See K.-St. I.148; Schuetz (1929); and Thomas (1938: 162–98).

Note (end) return to page 20a [added 01-09: Heine (1990: 13–14) argues that hyperbolic expressions are so frequent in Vergil and others that Vergil might have made Dido say adii, although agreeing with this specific interpretation. ]

Note (end) return to page 21 K.-St. (I.120–1) remark that the conative interpretation of present, imperfect or present participle only occurs in the case of an intended action (`beabsichtigte

-- 292 --

Ausführung einer Handlung'). An example of a conative present participle is:
- … si Lycomeden … multis cum lacrimis iter suum impedientem audire voluisset (`If he had wanted to listen to L., who was trying to stop him with many tears', Cic. Amic. 75).
[ERROR: no link :]21a [added 01-09: For literary exploitation of the offerebantur nisi iussisset type of expression see Nesselrath (1992). ]

Note (end) return to page 22 See Serbat (1976b: 330–40).

Note (end) return to page 23 Occisus sum, though, of course, formed with the present of esse, functions as one unit. See already Gellius:
- at cum dico: `factum est', `subruptum est', quamquam `est' verbum temporis est praesentis, confunditur tamen cum praeterito et praesens esse desinit (`When I say "factum est", "subruptum est", the verb form "est", though being a present-tense form, forms one whole with the past and stops being present', Gel. 17.7.7)
Besides the forms occisus sum/eram formally past-tense periphrastic forms developed at a relatively early stage: occisus fui/fueram. In the Romance languages forms corresponding with occisus est are present passive forms.

Note (end) return to page 24 Kühner & Holzweissig (1912: 805–6) give a number of instances of fuit, etc. that can be considered more or less equivalent to factum/factus est.

Note (end) return to page 25 For the aspectual explanation see also Handford (1946: 46–7) and Calboli (1966: 293–308); objections in Grassi (1967: 220–4) and Coleman (1975: 133), who points at the unequal distribution of the various uses over individual authors. As an argument for the aspectual view, it is pointed out that archaic Latin uses only so-called sigmatic (aoristic) forms such as ausim and faxim for the potential subjunctive of the perfect (Sz. 333). Objections against this formal argument may be found in Vairel-Carron (1980).

Note (end) return to page 26 The grammars devote a great deal of attention to the question of the boundary between the `potential' subjunctive and `counterfactual' subjunctive (irrealis), a distinction that is often difficult to make in practice. See K.-St. II.399–401; Sz. 662–3. For the historical development see Harris (1986).

Note (end) return to page 27 Gerund and gerundive are left out of account, as they do not have a merely temporal value.

Note (end) return to page 28 For Cicero see Laughton (1964: 38–41). See also chapter 8, note 10.

Note (end) return to page 29 This holds a fortiori for perfect participles used on the noun phrase level. See positis in (64b).

Note (end) return to page 30 Sz. (391) points out that this phenomenon is connected with the fact that these verbs have what he calls an `imperfective Aktionsart', i.e. are non-terminative.

Note (end) return to page 31 See Coleman (1975: 133).

Note (end) return to page 32 Instances such as exquisisse oportuit (`You ought to have asked') are to be explained as attraction (temporal harmony). See K.-St. I.135; Bolkestein (1980a: 160, note 16).

Note (end) return to page 33 For further examples see TLL s. v. licet 1360.45 ff.; s. v. possum 155.57 ff.

Note (end) return to page 34 See, among others, Weinrich (1964).

Note (end) return to page 35 Cf. carpebant in 522.

Note (end) return to page 36 This general rule cannot be applied mechanically. Virgil begins the second book of the Aeneid as follows:
- conticuere omnes intentique ora tenebant./inde toro pater Aeneas sic orsus ab alto (`All were silent and held their gaze bent upon him. Then from his lofty couch father Aeneas thus began', Verg. A. 2.1–2)
Conticuere is perfect, because it represents a new step in the story after Dido's invitation to Aeneas to relate what had happened at Troy (at the end of the first book). Tenebant describes the state resulting from conticuere and forms the framework for the perfect sic orsus. Tenebant must, therefore, be seen in relation with what follows rather than with what precedes (in spite of the coordination).

-- 293 --

Tenebant could, incidentally, also be interpreted as a kind of elaboration of conticuere (see also example (70) below).

Note (end) return to page 37 Quintilian (Inst. 9.3.11) considered the use of the historic present a kind of metaphor (transferuntur). For the historic present see Kravar (1969).

Note (end) return to page 38 Examples of the use of the historic infinitive in Livy are given by Viljamaa (1983: 38–51).

Note (end) return to page 39 See Chausserie-Laprée (1969: 403–10).

12 Beyond the sentence

Note (end) return to page 0 [added 01-09: For a speech act approach to Plautus see Cohen (1993). ]

Note (end) return to page 1 Henceforth, I will use `text' as a general term for `suprasentential structures', where others might prefer the notion `discourse'.

Note (end) return to page 2 Speakers and hearers tend to assume such a relation, and are willing to look for one if it is not immediately apparent (Grice's so-called `cooperative principle' (1975: 45)).

Note (end) return to page 3 The term `asyndeton' is also used to describe the absence of connecting constituents within a sentence. See crosssection 12.4. on p. 257 below.

Note (end) return to page 4 Cic. Att. 1.1–5 (178 lines Oxford Classical Text); Liv. 1.48–53 (195 lines Oxford Classical Text); Sen. Ep. 1–5 (181 lines Oxford Classical Text). From table 12.1 it may also be deduced that in this sample Seneca has the shortest, Livy the longest sentences.

Note (end) return to page 5 See Longacre (1982: 35) for the low frequency of connectors at climaxes of narratives.

Note (end) return to page 6 In Sen. Phil. the percentage of `coordinating conjunctions' in comparison with the total number of words is 7.86%, in Sen. Trag. 6.07%. `Coordinating conjunctions' also include coordination on the noun phrase level, but the difference between the percentages is nevertheless remarkable, considering the large number of words (see app. on `Liège' lexicons). For transitions between episodes and expressions used for this by historians see Chausserie-Laprée (1969: 15–124).

Note (end) return to page 7 Austin (1971, referring to TLL s. v. interea 2183.52 ff.) argues that in Virgil interea often means `and now' for successive actions. See, however, Kinsey (1979). Even if interea has its normal meaning `meanwhile', it is still a typical signal for the transition to a new episode. For the connecting function of temporal markers in Cicero see Hilton (1989).

Note (end) return to page 7a [added 01-09: Heine (1990: 2–3) rightly refers to repetition in juridical and similar types of text, as well as to Cicero's
- si Fabius oriente canicula natus est, Fabius in mari non morietur (`$&3squo;, Cic. Fat. 12) ]For examples see Mendell (1917: 22; 36–7).

Note (end) return to page 8 For examples see Mendell (1917: 22; 36–7).

Note (end) return to page 9 See TLL s. v. homo 2882.13 ff.; Mendell (1917: 36). See also the use of res in Caesar as in qua re animadversa (`Observing this', Caes. Civ. 1.83.5).

Note (end) return to page 9a [added 01-09: For res see Orlandini (1994). ]

Note (end) return to page 10 Examples of anaphorically used action nouns may be found in Rosén (1983: 187–9).

Note (end) return to page 11 For the type of cohesion seen in (18)–(20) see Conte (1981:49–51), with literature, and Hannay (1985), who uses the notion `subtopic' and Prince (1981)'s notion `inferability'.

Note (end) return to page 11a [added 01-09: Harris (1980: 77) is certainly wong, when saying that `where ... intolerable ambiguity might ensue - as with certain oblique pronominal uses - or where a particular emphasis or adversative effect was sought, forms were available to be pressed into service, forms from the paradigm of the anaphoric is, ea, id'. ]

Note (end) return to page 12 See the category of the `comparatives' in Halliday & Hasan (1976: 76–87).

Note (end) return to page 13 See Sz. 569–70.

Note (end) return to page 14 To some extent similar to the use of the term in Halliday & Hasan (1976: 88–90).

Note (end) return to page 15 See Sz. 513–14.

Note (end) return to page 16 K.-St. II.549 ff.; Sz. 822–6. I use the notion `ellipsis' in the sense in which it is commonly used in modern linguistics, so to denote `Ergänzung aus der Zusammenhang', which Sz. does not regard as ellipsis (Sz. 824). See also Tuomi (1975).

Note (end) return to page 17 For more examples see K.-St. I.3–7.

-- 294 --

Note (end) return to page 18 A survey of cases of omission of the predicate that are not due to textual cohesion is given by Sz. 419–25; see also K.-St. I.8–15.

Note (end) return to page 19 (37) might also be interpreted as one sentence.

Note (end) return to page 20 For examples see Mendell (1917: 98).

Note (end) return to page 21 Coordinators also link parts of sentences, see p. 257.

Note (end) return to page 22 For the difference between connectors and adverbs see Pinkster (1972c: 153–63).

Note (end) return to page 23 See for exceptional instances Pinkster (1972c: 157).

Note (end) return to page 24 See K.-St. II.1–145; Sz. 469–526. For individual words see TLL ss. vv.

Note (end) return to page 25 For a description of differences and similarities between et, atque and -que see Coseriu (1968).

Note (end) return to page 26 For the differences between these connectors see Bolkestein (1989) and Kroon (1989a), who point out pragmatic differences between the sentences linked by these connectors.

Note (end) return to page 27 For differences between these connectors see Kroon (1989a).

Note (end) return to page 28 A number of observations may be found in Pinkster (1972c: 155–64). Combinations of connectors are treated by Sz. (799–800) as `pleonastic'. No detailed research has been done into the syntactic properties of connectors.

Note (end) return to page 29 For quidem see K.-St. I.623–4 and Solodow (1978).

Note (end) return to page 29a [added 01-09: For these particles see Bolkestein (1988). ]

Note (end) return to page 30 For the text-structuring function of the tenses see Bolkestein (1987) and Beaugrande & Dressler (1981: 69–70). See also Rosén (1980: 48) on the relation between Pliny the Younger's use of the imperfect and the low frequency of connectors. For the imperfect in causally related sentences see Mellet (1988: chapter 14). For Apuleius see Dragonetti (1981).

Note (end) return to page 31 For the hierarchy of syntactic functions of arguments (Subject > Object > 3rd argument) see Dik (1978).

Note (end) return to page 32 See for these verbs and the principle of `harmony' Bolkestein (1985) and Bolkestein & Risselada (1985).

Note (end) return to page 33 See Bolkestein & Risselada (1987), Granger (1983: 292–9) and Pinkster (1985a).

Note (end) return to page 34 Example (48) is from Bolkestein (1983b); see Bolkestein (1981a) for a survey of the attested instances.

Note (end) return to page 35 An excellent classification of et may be found in J. B. Hofmann's TLL article.

Note (end) return to page 36 One often tends to translate this `introductory' et with `also' or `and, moreover, … '.

Note (end) return to page 37 See Bolkestein (1989) for some paratactic and hypotactic causal relators.

Note (end) return to page 37a [added 01-09: Moreno (1977) has statistical information about the cohesive devices used by five Roman historians (pronouns, situating adverbs, coordination, parataxis, explicit coreference). The authors discussed are Caesar, Sallustius, Livius, Tacitus and Suetonius. ]

-- 295 --

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