io Introduction
1.14 Authors and works
For references to authors and their works, I follow the abbreviations in the Oxford Latin Dictionary wherever possible. For Christian authors and texts I primarily use the abbreviations in Blaise's dictionary, which more closely resemble the form of abbreviating adopted in the OLD than do the abbreviations in the Thesaurus Linguae Latinae.
CHAPTER 2
Basic grammatical concepts
This chapter introduces the terms used most frequently in this Syntax to describe the structure of Latin sentences and texts. Details will follow in later chapters.
2.1 Clause, sentence, phrase, word, clitic, and constituent
2.2 The clause
The clause is denned as an expression that minimally consists of a verb form, or a. comparable entity, such as an adjective with or without a form of the verb sum 'to be'—for details see below in ex. (d). Usually a clause also contains one or more other words or phrases in addition to the verb; while some of these follow necessarily from the meaning of the verb, others do not. This is illustrated by (a).
(a)   At enim illi noctu occentabunt ostium...
("Well, they'll serenade your door at night...' PI. Per. 569)
This clause centres on the verb form occentabunt. The verb occento 'to serenade at or to' requires two participants who are direcdy implicated in the action of the verb: a person who performs the serenade {in this example: illi 'those people') and an entity the serenade is directed to (in this example: ostium 'door'). Accordingly, the verb occento is a two-place verb; that is, it requires two words or phrases to complete its meaning. For elements of a clause that are semantically obligatory in this way, the term argument will be used. The arguments illi and ostium in combination with the verb, occentabunt, form the core or the nucleus of the clause (further details and complexities follow in § 2.7 and in Chapter 4).
In this particular example, the future tense of occentabunt signals that the nucleus refers to a state of affairs that will come into being at a time later than the communicative situation in which the speaker utters the words. Furthermore, the speaker presents his utterance as a fact by his choice of mood: he uses the indicative mood of the verb.
14   Basic grammatical concepts
Clause, sentence, phrase, word, clitic, and constituent 15
(g) Quid si igitur reddatur illi unde empta est? # Minime gentium. / # Dixit se redhibere, si non placeat.
('Then how about returning her to the man I bought her from? # Not at all. # He said he'd take her back if I didn't like her.' PI. Mer. 418-19)
Exx. (d)-(g) show various ways of combining clauses to form complex clauses,-where one clause is part of another clause, either as a satellite or as an argument. Another form of clause combining is illustrated by (h). Ex. (h) has two finite clauses which are linked by the coordinator et 'and' and so form what is called a compound clause in this Syntax (a more traditional term for et is 'coordinating conjunction'). In contradistinction to the complex clauses dealt with above, neither of the clauses contains the other or is contained in the other. In (d) above we also had two finite clauses: non sino and neque... sinam. Here the two clauses were linked by neque, which combines the functions of a negation word and a coordinator. Another example of two coordinated clauses is (i). When a compound clause is transformed into an accusative and infinitive clause, all coordinated clauses become accusative and infinitives.
(h) {Odi} et (amo.)
('I hate and I love.' Catul. 85.1)
(i) Quia {ego hanc amo) et {haec med amat}... ('Because I love her and she loves me.' PI. As. 631)
Coordination and subordination are two different means of forming multiple clauses.1 Where necessary, the term simple clause will be used for clauses that are not multiple (complex or compound).
2.3 The sentence
'Sentence' is a problematic concept in linguistic theory and is avoided in certain approaches.2 In other approaches, 'clause' is used as more or less equivalent to 'sentence'. In this Syntax, the two terms will be used to refer to different phenomena. This difference can be illustrated by the following exchange between the slave Tranio and the old man Theopropides about what happened to a certain sum of money.
(a) Tr.: Aedis filius / tuos emit.
(b) Th.: Aedis?
(c) Tr.: Aedis.
("Your son's bought a house. # A house? # A house.' PL Mos. 637-8)
1 The terms 'multiple', 'compound', and 'complex1 are taken from Quirk et al. (1985: 987).
2 See Longacre (2007), who defines sentences as combinations of clauses. For the definition of 'sentence', see also Bodelot (2007).
Ex. (a) is a grammatically complete clause with all the arguments required by emit present. In actual use (either oral or written) it is marked as a complete unit of communication. In (a), this follows from the use of the period. Although in spoken Latin there may have been something comparable to the falling tone at the end known from many languages, including English, we are not able to recover much information about Latin intonation.3 Ex. (b) is a question (or exclamation) in reaction to the surprising preceding statement. It is an incomplete clause but a complete unit of communication (in modern texts the question mark is used to indicate this; in spoken Latin the intonation may have been different from (a), for example in having a rising tone, as in English). Ex. (c) is also an incomplete clause, but like (a), it is a statement (marked by a period). Such complete units of communication are called sentences in this Syntax.
A clause (simple or multiple) may coincide with a sentence. This is the case in (a) above and in the examples (d)-(i) cited at the end of the preceding section. Sentences may also be shorter than a complete clause, as is shown by (b) and (c) above. This is further illustrated by (d)-(g). In (d), the sentence Vobis... volt consists of a relative clause that is uttered as a reply to the preceding question and which is the subject of the verb that has to be supplied from that question (est 'is'). In (e), the one-word sentence consisting of the noun malum 'evil' is a reply to the preceding question, from which something like dabo 'I will give' must be supplied to govern the object malum. Similarly, the adjective scitula in (f) goes with the noun facie in the preceding question. In these three instances, the preceding question contains an element that offers a slot to be filled by the answer. In (g), by contrast, while the speaker probably expects an answer to his suggestion (the question mark in the OCT may be exaggerated), the precise form of the reaction is unpredictable. This is even more true for nugae in (h), a one-word sentence serving as a comment on the preceding words of the interlocutor.
(d) Quis est? / # Vobis qui multa bona esse volt.
{'Who is it? # Someone who wants you to have lots of good things.' PL True. 116-17)
(e) Si huius miseret, / ecquid das qui bene sit? # Malum.
('If you feel pity for her, will you give her anything from which all will go well for her? # A thrashing.' PL Cur. 518-19)
(f) Qua sunt facie? # Scitula.
('Of what appearance are they? # Pretty.' PL Rud. 565)
(g) Fortasse tu hue vocatus es ad prandium, / ill' qui vocavit, nullus venit? # Admodum.
('Perhaps you were invited here for lunch and the one who invited you didn't come? # Exactly.' PL Rud. 142-3)
5 Ancient grammatical and rhetorical treatises manifest a keen interest in intonation, and they also recognize various types of sentences, but they do not seem to pay attention to the relationship between these areas. Augustine (Doctr. chr. 3.3.6) gives instructions to pronounce information-requesting and rhetorical questions differendy in order to avoid ambiguity (as pointed out by Branden Kosch, p.c). For ideas about intonation in Antiquity, see Luque (2006).
16   Basic grammatical concepts
(h)   Tace modo. Deus respiciet nos aliquis. # Nugae.
(')ust be quiet: some god will look after us. # Nonsense!' PI. Bac. 638)
Latin has four different types of sentences (sentence types), three of which are illustrated above. When using a declarative sentence the speaker asserts a state of affairs, as is illustrated by the answers in (d)-(g) and the second sentence of (h). Ex. (d)-(g) also illustrate interrogative sentences (or 'questions'). Ex. (h) illustrates the imperative sentence type. Finally, Latin has exclamatory sentences.
the number of utterances that can be used as sentences is infinite. However, there are certain words that cannot be used as sentences. Examples are the preposition ad 'to', the coordinator et 'and', the connector nam 'for', and the interactional particle enim 'for (as you know)'.
Application of the content of the preceding paragraph to the Latin material is not unproblematic, not least because we only have written material, and the spaces and punctuation marks that were in use served a different purpose from punctuation in modern languages. Capitals or something similar were not used to mark the beginning of sentences in a systematic way. (For details on the evidence see Chapter 24.4) The punctuation in our printed texts is the result of modern interpretation and varies from one national tradition to another. For the purposes of this Syntax, capitals and periods are used to demarcate words and sequences of words as sentences. Although the texts used for illustration are essentially those from the Teubner series and the Corpus Christianorum series, the punctuation has been adapted whenever necessary. Sentences start with a capital and end with a period or question mark. In practice this means that many semicolons have been replaced by periods.
Appendix: In actual pronunciation, sentences, certainly those of a greater length, were segmented into smaller units, which may, but need not, coincide with grammatical units (words, phrases, or clauses). These segments may also be sequences of grammatical units between which no immediate syntactic relation exists. A very prosaic reason for this segmentation is the need to take a breath, but it may also serve other purposes, such as giving emphasis and the production of certain rhythmical structures. We may assume that such segments were also marked by intonation. For such segments the term colon will be used. Details are discussed in Chapter 24.
Clause, sentence, phrase, word, clitic, and constituent 17
more elements modifying it, belonging to various classes. In principle, the noun phrase as a whole has the same structural potential as the head by itself.
The elements modifying a noun are called attributes, but sometimes also in a more general way: modifiers. At this point, it suffices to give an example of a modifying possessive adjective, as in (a), and another with a noun in the genitive, as in (b). In (a), the possessive adjective nostrum functions as the attribute of hortum, which is the head of the noun phrase hortum nostrum. In (b), the genitive noun horti modifies the noun ostium, which is its head. In both cases, omission of the modifying element does not make the remainder of the structure ungrammatical.
(a) Abii iliac per angiportum ad hortum nostrum clanculum.
('] secretiy went that way through the alley to our garden.' PI. Mos. 1045}
(b) Ostium quod in angiporto est horti, patefeci fores...
('I opened the wings of the garden door that is in the alley.' PI. Mos. 1046)
Simple illustrations of an adjective phrase and an adverb phrase are (c) and (d). We see the degree adverb valde 'very' modifying the adjective bonis 'good' and the adverb bene 'well', respectively. Here, too, omission of the modifying element does not lead to an ungrammatical expression.
(c) Explicat orationem sane longam et verbis valde bonis.
('He unrolls a speech that is very long and characterized by very fine words.' Cic. Agr. 2.13)
(d) Rem te valde bene gessisse rumor erat.
{'There is a report that you have had a highly successful campaign.' Cic. Fam. 1.8.7)
With prepositional phrases the situation is different. In (a) above, ad hortum nostrum is a prepositional phrase. Ad seems to be the 'dominant' element in the phrase, since the accusative case form of hortum nostrum depends on it. However, neither the preposition ad nor the noun phrase hortum nostrum can be omitted without making the remaining structure ungrammatical. (The relation between the preposition and the other constituent in a prepositional phrase is discussed further in § 12.23 and in § 12.25.)
2.4 The phrase
The term phrase will be used for combinations of two or more words that behave as a unit with respect to other elements of the structure to which they belong. In this Syntax, four types of phrases are distinguished: noun, adjective, and adverb phrases on the one hand, and prepositional phrases on the other. The term 'verb phrase' will not be used in this Syntax. A noun phrase consists of a noun (the head) and one or
4 A good introduction to pause and punctuation in Latin texts is ch. 1 of Parkes (1992).
2.5 The word and the clitic
A familiar definition of word is: the smallest linguistic segment in an utterance that can (in principle) be moved independently. In this respect, a word differs from a morpheme (a word-internal segment) and from a phrase (a segment containing two or more words). More importantly, words may stand at the beginning of a sentence, which distinguishes them from clitics. Most clitics have a meaning of their own, but they are bound to a word that functions as their 'host' with which they form a
18   Basic grammatical concepts
The nucleus of the clause 19
phonological unit. Although clitics have a certain freedom as to which words they are attached to, they can never stand in a sentence-initial position (for details see § 3.28).
Modern editions use a space to separate individual words. The Romans often marked word boundaries in some way or other, but there was both individual and diachronic variation (see Chapter 24). Sometimes the order and relative position of words in a phrase is such that one may hesitate whether a given segment is a word or a phrase. In the course of time, a phrase may develop into a word, This can be illustrated with the segment magno opere I magnopere in its meaning 'greatly' or 'particularly'. It is rare in Early Latin, but there are a few instances in Plautus. In the OCT it is printed as two words in (a) (manuscripts vary). There are also instances with the order reversed, as in (b). Obviously therefore, in Plautus' time it was still a phrase. After Plautus, the reverse order is not attested any more (except with a different meaning). Cicero has two instances of separation—(c) and (d)—but in the instances where the words occur together, editions of his texts vary in printing magno opere or magnopere. This is also the case in editions of later authors, most of which do not feature instances of separation.5
(a) Edictum est magno opere mihi, ne quoiquam hoc homini crederem... ('And I was ordered explicitly not to entrust this to anyone...' PI. Per. 241)
(b) Vos omnis opere magno esse oratos volo...
('I want to appeal earnestly to you all...' PI. Cos. 21)
(c) Cum puerorum igitur formas et corpora magno hie opere miraretur...
('When, therefore, he was greatly admiring the figures of the boys, and their bodies,...' Cic. Inv. 2.2)
(d) ... magnoque opere abs te peto cures ut is intellegat...
('... I earnestly request you to make him realize...' Cic. Fatn. 13.34.1)
rnnre complex structure without specifying to which category they belong, or which function they fulfil, the term constituent has become common in contemporary linguistics; it will be used throughout this Syntax.6
In (a), some of the constituents have an internal structure of their own. The noun phrase magno opere, for example, consists of two constituents and the cures clause also consists of two constituents: cures and the clause ut is intellegat, which in turn can be analysed in more detail. A syntactic argument for regarding magno opere as one constituent in its clause (and not two) is that it could be replaced by one word (for example, valde 'strongly'); similarly, the clause cures ut is intellegat could be replaced by the single word auxilium 'help'. These replacements would still result in a correct grammatical structure: valde abs te auxilium peto and would leave the syntactic structure of the sentence unchanged.
2.7 The nucleus of the clause
In § 2.2 the nucleus (or core) of the clause was defined as the combination of the, verb (or similar element) and its arguments. In example (a), repeated below, the verb occento requires two elements to create a meaningful clause. The nucleus refers to a state of affairs, in this particular case an action, in which an agent (illi) and a patient (ostium) are involved. In the next section, some further terminology is introduced concerning the nucleus of the clause.
(a)   At enim illi noctu occentabunt ostium...
('Well, they'll serenade your door at night..." Pi. Per. 569)
2.6 Constituents
Clause (a) below, repeated from the preceding section, may in a first analysis (ignoring -que) be divided into four parts, indicated by brackets.
(a)   ... [[magnoque opere]] [[abs te]] [[peto]] [[cures ut is intellegat]]... ('... t earnestly request you to make him realize...' Cic. Fam. 13.34.1)
These four parts belong to different categories: a noun phrase, a prepositional phrase, a verb, and a subordinate clause, respectively. They also fulfil different functions in the clause. Magno opere, for example, is a manner expression and the subordinate clause functions as the object of peto. As a neutral term for referring to 'parts' of a
2.8 Valency and verb frame
The number of arguments a particular verb requires depends upon its meaning. The term valency is used to describe this. The verb occento is two-place or^ bivalent; as shown above, it requires an argument that performs the action (an agent), and a second argument on which the action is performed (the patient). The verb ambulo 'to walk' is a one-place or monovalent verb: it requires only an agent to complete its meaning. The verb do in its basic meaning 'to give' is three-place or trivalent (one entity gives another entity to a third entity). And, finally, pluit 'it rains' does not require any element at all to complete its meaning; hence it is zerovalent.
If a verb has more than one meaning the number and the type of arguments may vary accordingly. The verb dico, for example, has several distinct meanings, each with
5 For details, see TLL s.v. opus 854.1 Off.
6 The term is not found frequently in scholarship in Classics; for an exception, see Habinek (1985: passim).
20   Basic grammatical concepts
its own constellation of arguments or, in the terminology of this Syntax, its own verb frame Jhis is illustrated by (a)-(d). In (a) dico means 'to speak', 'to give a speech'; in this meaning it requires one argument (the agent) (note the parallelism with vivendi). In (b) it means 'to say' (two arguments), in (c) 'to tell' (three arguments), in (d) 'to appoint' (also three arguments, but of a different type and with a different pattern of case forms).
(a) ... idem erant vivendi praeceptores atque dicendi...
('... the same men were teachers of ethics and of speaking...' Cic. de Oral. 3.57)
(b) Egone istuc dixi? # Ita.
('I said that? # Yes.' PI. Bac. 806)
(c) Egone ;'sfHC dixi tibP. I # Mihi quidem hercle.
('I said that to you? # Yes, to me, by Hercules.' PI, Mer. 761-2)
(d) Postero die dictator...magistrum equitum dicit L. Tarquinium...
('On the following day the dictator... named as his master of the horse Lucius Tarquinius...' Liv. 3.27.1}
The frame or frames a verb is associated with do(es) not only depend on the meaning of the verb. Even with one and the same meaning, a verb may have multiple frames. The lexical properties^ the argument (s) play a role as well. Thus, with a human first argument the verb incipio 'begin' is usually bivalent ('a person begins something') but with another type of first argument, inanimate autumnus 'autumn' for example, incipio is monovalent, as in (e).
(e) Nam desinit aestas, / incipit autumnus media sub Virgine utrimque.
('For at the middle of the Virgin summer on one side ceases and autumn on the other begins.' Man. 2.176-7)
The verb facio in its meaning 'to make, to produce' will normally be considered bivalent ('someone produces something'), but with an emotion noun as its patient, such as dolorem 'grief, the resulting causative expression requires a recipient in the dative case form, as in (f).
(f) Sed augeo commemorando dolorem et facio etiam tibi.
('But I make the pain worse by dwelling on it, and give you pain also.' Cic. Att. 11.8.2)
Conversely, the verb do 'to give' is a typical trivalent predicate, requiring a recipient in the dative, but not so if the second argument is something like motus 'movements' (further details are discussed in Chapter 4) 7
Valency is essentially a quantitative notion: it indicates how many arguments a verb normally requires. However, the arguments also have a semantic relationship
7 For factors influencing the presence or absence of a recipient constituent with a number of verbs, see Banos (1998: 28-39).
1
The nucleus of the clause 21
with respect to the verb (functioning, for example, as the agent) and a syntactic function (functioning, for example, as the subject). Semantic and syntactic functions are dealt with in § 2.12.
Appendix: The notion of valency is relatively recent, but can be traced back to ancient notions of transitivity and intransitivity. It was introduced into general linguistics in a systematic way by Tesniere (1959) and into Latin linguistics by Dressier (1970).8 Although the notion is present in some form or other in most contemporary linguistic models, there remain a number of difficulties in its practical application (see 4.1-4.6). It has been suggested that assigning one or more precise valencies to every verb is impossible in view of the number of borderline cases and that it is better to regard valency as a scalar concept: some entities are more obligatory or optional than others.9 Similarly, some scholars consider the strict division between arguments and satellites impracticable and/or unnecessary.
Objective tests to determine the valency of predicates have been developed for several languages, including Latin.10
The special role assigned to the verb (or a comparable entity) is justified because the structure of the nuclear predication is determined by its meaning. This does not exclude the possibility of verbless utterances that in all other respects look like 'normal' clauses. Nor does this say anything about the prosodic status of verbs in general or of particular verbs. The fact that notably the monosyllabic forms es 'you are' and est 'he/she/it is' were elided and are also often found in the shadow of salient words, resembling the behaviour of clitics, is also irrelevant for the concept of valency."
The notion of valency is also applicable to non-verbal categories, such as adjectives and nouns. The adjectives similis 'alike' and dissimilis 'unlike' in (g) are both bivalent; the same holds for the noun dux 'guide', 'leader' in (h), whose valency is like that of duco.
(g) Sic dicitur similis homo homini, equus equo, et dissimilis homo equo.
('Thus a human being is said to be like a human being, and a horse to be like a horse, and a human being to be unlike a horse.' Var. L. 10.4)
(h) Cuius legationis Divico princeps fuit, qui bello Cassiano dux Helvetiorum fuerat.
('The leader of the deputation was Divico, who had been commander of the Helvetii in the campaign against Cassius.' Caes. Gal. 1.13.2)
Nouns, adjectives, and verbs that share a particular semantic feature often have the same valency and allow the same types of elements to be combined with them. This phenomenon of transcategorjai. parallelism is illustrated by (i)-(k). The verb cupio 'to desire' governs an infinitival clause, as in (i). The same type of embedding is
References in LSS § 1-2. For Latin valency from a typological perspective, see Lehmann (2002).
For the application of the concept of scalarity to valency, see Himmelmann (1986). 10 See Happ (1976: 346-428); LSS § 2.1; Banos etal. (eds) (2003), especially the contribution by de la Villa (2003); Torrego et al. (eds) (2007), with references.
" On the prosodic status of verbs, see Fortson (2008: ch. 9).
22   Basic grammatical concepts
found with the noun cupido 'desire', as in (j), and with the adjective cupidus 'desirous', as in (k).
(i)   Cum hoc... cupivit... bellum componere...
('With him... he wished to arrange a truce..." Nep. Han. 6.2) (j)   ... si tanta cupido / bis Stygios innare lacus...
('... if there is such a great desire / to swim the Stygian lake twice..." Verg.
A. 6.133-4)
(k)   ... cupidus falsis attingere gaudia palmis,..
('... yearning to touch with unreal hands his heart's delight.' Prop. 1.19.9)
2.9 States of affairs
The verb and its arguments refer to some situation in the observable or imaginary world in which we live. The structure of such situations may be quite diverse. Human beings may be actively involved, exercising their will over what is taking place, or something may take place or be the case without some form of active involvement. Further, the situation may or may not involve change (for example, John grew old or John was old), etc. From Aristotle onwards, terms have been proposed to describe and characterize the most important types of situations. As a general term for the various types of situations this Syntax will use the term state of affairs (SoA). The following distinctions will be made:12 (i) SoAs that are controlled or not controlled by some entity denoted by the first argument (usually Fnuman being, but also animals or forces of nature may be depicted as controllers, exercising their will over some situation); (ii) SoAs that do or do not imply change (or, in other words, SoAs that are 'dynamic' or 'non-dynamic'). The combination of these two features control and change results in four types of states of affairs, as indicated in Figure 2.2.
English examples of these four types are: John went home (action), John stayed home (position), John grew up in London (process), and John knows Greek (state).
	Dynamic (+ change)	Non-dynamic (- change)
	ACTION (jo)	POSITION {stay)
Not amlrollcd {- control)	PROCESS (grow)	STATE (know)
Figure 2.2 Typology of states of affairs (simple version)
i2 Essentially this follows LSS § 2.4 with adaptations from Haverling (2000: 22-31), where further references to the literature can be found. Some of the English examples are taken from her book See also Haverling (2010a: 284-340). The term 'state of affairs' is taken from Dik (1997:1.105-26).
The nucleus of the clause 23
States of affairs that are controlled can occur in the imperative and may be used in subordinate clauses depending on verbs of ordering and wishing, like impero and volo. They can furthermore be combined with manner and instrument expressions and also with beneficiaries. This is briefly illustrated by (a)-(c). In (a), abi is an imperative form. It is combined with the manner satellite cito. In (b), dormitum ut abeas depends on volo, to be understood from the preceding question. Ex. (c) illustrates a beneficiary satellite (maioribus natu). Details and problems follow in § 6.30 (imperative) and in 5 10.42 (process adjuncts).
(a) Abi cito.
('Off with you, quickly!' PI. Cist. 781)
(b) Numquid v«? # Dormitum ut abeas.
('Do you want anything? # Only that you go off and sleep.' PI. Ps. 665)
(c) Ut maioribus natu assurgatur...
('... as, for example, rising out of respect to elders..,' Cic. Inv. 1.48)
The absence or presence of the feature 'change' is especially relevant for the combin-ability of an SoA with various expressions of time and for the interpretation of certain tense forms. In fact, the binary distinction +/- dynamic has to be refined further. Dynamic states of affairs either may be brought or come to a natural end (in that case they are called terminative or 'telic') or may not (non-termmative or 'non-telic'). English examples are John is walking along the beach (non-terminative) and John is walking to school (terminative). A further distinction can be made into momentaneous and non-momentaneous states of affairs. An example of a momentaneous SoA is John finds a book on the beach. The distinction +/- momentaneous operates especially within the type of terminative states of affairs, but a non-terminative SoA like John coughed is momentaneous as well. Non-momentaneous SoAs are called accomplishments, momentaneous ones achievements. These distinctions are represented in Figure 2.3.
State of Affairs
- dynamic John is tying an the beach?
y dynamic
- terminative John is walking along the beach?
- momentaneous John is walking to school?
+ momentaneous John finds a book on the beachß
Figure 2.3 Typology of states of affairs (more elaborate version)
24  Basic grammatical concepts
Latin illustrations are (d)-(g).
(d) Quis hie est senex / qui ante aedis nostras sic iacet?
("Who's this old man lying in front of our house like this?' PI. Am. 1072-3 non-dynamic)
(e) In foro infumo boni homines atque dites ambulant.
('In the lower forum citizens of repute and wealth stroll about.' PL Cur. 475 non-terrrunatrve)
(f) Quo ambulas tu qui Volcanum in cornu conclusum geris?
('Where are you going, you who are carrying Vulcan locked up in your horn?' PI. Am. 341 non-momentaneous)
(g) Interii, si non invenio ego illas viginti minas.
('I'm dead if I don't find those twenty minas.' PL As. 243 + momentaneous)
With non-dynamic states of affairs, an expression of duration is possible, as in (h). Such expressions are excluded with terminative states of affairs, which however allow an expression of the time within which, as in (i) (for examples, see 10.31-10.38).
(h) Duodequadraginta annos tyrannus Syracusanorum fuit Dionysius... ('For thirty-eight years... Dionysius was tyrant of Syracuse.' Cic. Tusc. 5.57)
(i) Ipse... Tarraconem paucis diebus pervenit.
('In a few days he himself arrives at Tarraco...' Caes. Civ. 2.21.4)
Momentaneous terminative states of affairs cannot be combined with phasal verbs such_as desino 'to stop' and incipio 'to begin'.'3 Sentences containing anon-momen-taneous terminative state of affairs in the perfect tense may be ambiguous if combined with paene 'almost', as in (j): was the action of destruction interrupted before final destruction was reached, or did the action not take place at all?
(j)   Cogitate quantis laboribus fundatum imperium... una nox paene delerit.
('Think how one night almost destroyed the empire founded by such toil.' Cic.
Catil. 4.19)
2.10 Satellites
Clauses often contain more than just the verb and its obligatory arguments. Ex. (a), repeated here from § 2.2, contains apart from its nucleus a connector (at), an interactional particle (enim), and a satellite (noctu).
(a)   At enim illi noctu occentabunt ostium...
('Well, they'll serenade your door at night...' PL Per. 569)
13 See Vester (1983: 22-7); Haverling (2010a: 305).
Satellites 25
Several types of satellites can be distinguished. In the first place, there are satellites that add details to, or specify in some way, the SoA described by the verb and its arguments: in (a) noctu situates the action illi occentabunt ostium in time. In the same way, one could add that they will do this magna voce 'in a loud voice', inplatea 'on the street', or suis sodalibus 'for the benefit of their comrades'. For this type of specifying satellites, this Syntax uses the term^ADjUNcr. Some of these adjuncts occur only, or predominantly, with states of affairs of a specific type. Instrument adjuncts, for example, occur typically with controlled states of affairs. Time adjuncts like noctu, by contrast, have no restrictions.
Another type of satellite is illustrated by (b) and (c). Certe in (b) expresses the speaker's subjective evaluation of the content of the clause. The clause ut... veniamus in (c) indicates the status of the following information in the ongoing conversation and how that information has to be understood. The two expressions do not give further details about the SoA; they relate rather to the communicative situation of the speaker and hearer. For this type of satellite this Syntax uses the term disjunct. Further discussion will follow in §§ 10.97-10.107.
(b) Certe hie insanu'st homo.
('This man is certainly mad.' PL Men. 282)
(c) Ut vero iam ad ilia summa veniamus, quae vis alia potuit... homines unum in locum congregare...
('To come, however, at length to the highest achievements of eloquence, what other power could have been strong enough... to gather... humanity into one place Cic. de Orat. 1.33)
A composite clause based on exx. (a)-(c) can be represented in a graph as in Figure 2.4, showing the hierarchical position of the various components. The verb is the most central
CLAUSE
NUCLEUS
SATELHTE(S)
VERB
ARGUMENT(S) ADJUNCT(S)
DISJUNCT(S)
occentabunt        illi        ostium   noctu   magna voce     certe      ut ad summa
venjamus
(Ut ad summa veniamus, certe Uli noctu magna voce ostium occentabunt.)
Figure 2.4 The hierarchical structure of the clause (more elaborate version)
26   Basic grammatical concepts
element, completed by its arguments. This combination is then further specified by adjuncts, and the resulting whole 'commented on' by disjuncts.
2.11 Extra-clausal constituents
Besides constituents that belong to clauses (whether complete or not), Latin has three types of extra-clausal constituents. In combination with (complete or incomplete) clauses they form sentences, without having a syntactic relation to these clauses. The first type, so-called theme constituents, precede the remainder of the sentence. The function of themes is to draw attention to, or to announce or introduce a new entity. An example is (a). The second type, tail constituents, function to give additional information about a constituent in the preceding clause. An example of a tail constituent is (b), where the tail constituent viginti minae gives a specification of argentum in the preceding clause (the traditional term is 'apposition'). Ex. (c) illustrates the third type, a parenthesis, which is inserted in the middle of a sentence. In this particular instance, it justifies the repetition of the same statement.
Sed urbana plebes, ea vero praeceps erat de multis causis. ('As for the city populace, they in particular acted with desperation for many reasons.' Sal. Cat. 37.4)
Hercle te hau sinam emoriri, nisi mi argentum redditur, / viginti minae. ('By Hercules, I won't let you die, unless I'm paid the money, twenty minas.' PL Ps. 1222-3)
Sumus enim natura, ut ante dixi—dicendum est enim saepius—studiosis-simi adpetentissimique honestatis...
('We are by nature, as I have said before—and indeed it must be said quite often—the most eager for and desirous of honour...' Cic. Tusc. 2.58)
(a)
(b)
(c)
2.12 Semantic, syntactic, and pragmatic functions
For a full grammatical description of the way constituents function in Latin clauses and sentences and of their relation to other sentences in the context or to the extra-linguistic world, at least three types of concepts are needed. In the first place, a certain amount of terminology is required to describe the semantic relations between the constituents of a clause. In this Syntax, the following semantic functions of arguments will be distinguished:
(i) agent for entities wilfully instigating an SoA, as in (a), an active sentence, and in (b), a passive clause. Agents are especially human beings or entities equated with them, such as animals and sometimes forces of nature. Satellites may also function as agent, as in (c) with monovalent moriuntur, but this is very infrequent.
Semantic, syntactic, and pragmatic functions 27
(a) Pater vocat me...
('My father calls me...' PI. Am. 991)
(b) C. Gracchus ad M. Pomponium scripsit... haruspices a patre convocatos. ('Gaius Gracchus wrote to Marcus Pomponius that the soothsayers were called together by his father.' Cic. Di'v. 2.62)
(c) Moriuntur non alter ab altero, sed uterque a patre.
(They died, not at each other's hand, but both at their father's.' Sen. Con. 5.3.1)
(ii) cause for entities that bring about an SoA without wilfully controlling it, as in (d). Satellites may also function as cause, as in (e).
(d) Eius amor cupidam me hue prolicit per tenebras.
('Love for him is driving my eager self out here through the darkness.' PI. Cur. 97)
(e) Is amore misere hanc deperit mulierculam...
('He's dying wretchedly with love for this woman...' PI. Cist. 131)
(iii) patient for entities undergoing an SoA, as in (a)-(d). The term 'patient' covers both so-called apfectel^ objects (as in John restored the house) and so-called effected objects (as in John built the house). The term experiencer is used for entities that undergo a physiological or psychological SoA, as hanc mulierculam in (e).
(iv) recipient and addressee for entities that function as third arguments in three-place states of affairs referring to handing over or communicating something, respectively. Examples are (f) and (g), respectively.
(f) Diabolus Glauci filius Clearetae / lenae dedit dono argenti viginti minas...
('Diabolus, the son of Glaucus, has given twenty silver minas as a gift to the madam Cleareta...' PI. As. 751 -2)
(g) Em tibi! Hie mihi dixit tibi quae dixi.
('Here you go! He told me what I told you.' PI, Mil. 365)
(v) associative for entities that actively participate in a two-place SoA together with a principal agent, as with depugno (cum) 'to fight (with)' in (h). This function is also fulfilled by satellites, in combination with verbs that do not require them by their meaning, as in (i).
(h) Ter depugnavit Caesar cum civibus, in Thessalia, Africa, Hispania.
('Three times Caesar fought it out with his countrymen, in Thessaly, Africa, and Spain.' Cic. Phil. 2.75)
(i) ... ut cum exercitu / hinc profectus sum ad Teloboas hostis eosque ut
('... since I went away with the army from here to our enemy, the Teloboians, and defeated them.' PI. Am. 733-4)
28   Basic grammatical concepts
(vi) direction, place, and source for entities that function as arguments in SoAs implying motion or rest, like the source argument in (j). These functions are also fulfilled by satellites.
(j) Puer aberravit inter homines a patre.
('The boy strayed from his father among the crowd.' PI. Men. 31)
The terms mentioned in (i) and (iii)-(v) are especially useful for the description of states of affairs in which human beings are involved, as the terms themselves suggest. Many linguists use different labels, and also sometimes a more extensive set of categories.14 In this Syntax, this particular set of labels will help explain certain facts about Latin; it is not suggested that they cover all semantic relations between the verb and its arguments. More discussion will follow in Chapter 4.
The most common semantic functions of satellites (apart from agent, cause, and associative—see above) are:
(vii) beneficiary (the entity to whose advantage or disadvantage an action is undertaken).
(viii) instrument (the tool, usually inanimate, used in performing an action).
(ix) manner (indicates the way in which an SoA is carried out or takes place).
(x) reason (indicates why an SoA is carried out or takes place).
(xi) purpose (indicates with what goal in mind an action is performed).
(xii) Several spatial specifications.
(xiii) Several temporal specifications.
A fuller list can be found in the table of contents of Chapter 10 and Chapter 16. The number of functions (and their nomenclature) varies in the literature.
Arguments and satellites can furthermore be described in terms of their syntactic functions. The only syntactic function on which Latinists seem to agree is that of the subject, commonly defined as the constituent in a finite clause with which the verb form shows agreement.15 This definition is then often extended to non-finite constructions, such as the accusative and infinitive clause and the ablative absolute clause. Examples are (k)-(m). In the active clause (k) laudat agrees with its subject pater: it is third person singular. In the passive clause (1) the verb agrees with filius. In the (bracketed) accusative and infinitive clause in (m) there is another type of agreement (number, case, and gender) between the subject filium and the infinitive laudatum esse.
(k)   Pater filium laudat.
('The father praises his son.')
14 See for example Dik (1997:1.117-24) and Givon (2001: 117-61). Part of the terminology is taken from them, and some definitions are adaptations from these sources.
15 For discussion of the status of the subject in Latin, see Lavency (1994).
Semantic, syntactic, and pragmatic functions 29
(1)   Filius a patre laudatur.
('The son is praised by his father.')
(m)   Dicunt {filium a patre laudatum esse}.
('They say that the son was praised by his father.')
Most Latinists will cail filium in (k) the (direct) object of the clause. Some will restrict the use of the term object to instances precisely like (k), that is, to the second argument in the accusative case that can also be used as the subject in a passive counterpart, as in (1). As a consequence, in this view, the second arguments in (n) and (o) do not count as objects: pecunia mca in (n) is in the ablative and passivization is excluded; magnos articulorum dolores in (o) is in the accusative, but passivization is nevertheless excluded. Another instance of an accusative which is not passivizable is me in (p) with the deponent verb sequor. Other Latinists will take only the accusative as the defining criterion.
In this Syntax, the term 'object' will be used for all four types of second argument, (k) and (n)-(p). If necessary, for instances like (n) the term 'ablative object' is used.
(n)   Pecunia mea tot annos utitur P. Quinctius.
('Quinctius has had the use of my money for so many years.' Cic. Quinct. 43) (o)   Terentia magnos articulorum dolores habet.
('Terentia has a bad attack of rheumatism.' Cic. Att. 1.5.8) (p)   Num ilia me nunc sequitur?
('She isn't following me, is she?' PI. Cos. 936)
The term indirect object is commonly used for the third argument of verbs of transfer or communication, that is, for an argument that functions semantically as a recipient or an addressee (see above). Examples are (q) and (r). Some scholars also use it for the second argument of verbs that govern a dative, such asfaveo, as in (s). In this Syntax, however, that constituent is called a 'dative object'.
(q)   Prandium uxor mihi perbonum dedit...
('My wife gave me a very good lunch.' PI. Mos. 692) (r)   Em tibi, hie mihi dixit tibi quae dixi.
{'Here you go! He told me what I told you.' Pi. Mil. 365) (s)   Hominum nobilium non fere quisquam nostrae industriae favet.
('There is hardly one member of the old families who looks kindly on our activity.' Cic. Ver. 5.182)
Adopting the traditional term 'indirect object' raises the question of what to call other third arguments—for example, the ablative constituent with verbs of supplying, as in (t). (NB: this verb is also used with an accusative + dative [object + indirect object]
30   Basic grammatical concepts
pattern.) There is no generally accepted term available. In this Syntax, the description of such structures will be given in an ad hoc way.16
(t)   Ubi igitur est crimen? Quod eum Pompeius civitate donavit.
('Wherein then does the accusation consist? In this: that Pompeius has honoured him with citizenship.' Cic. Balb. 7)
The remaining syntactic function of arguments is that of the complement. Two types of complements are distinguished, the subject complement and the object complement.1' In some English grammars these constituents arc called 'predicates'. Examples are (u) and (v). In (u), uliginosum and lutulentum are used in combination with the copula sum (which requires two arguments). Being adjectives, they show agreement with the subject (caprile). In (v), lutulentos is the third argument of reddo and shows agreement with the object multos homines.
(u)   Id, ut pleraque, lapide aut testa substerni oportet, caprile quo minus sit uliginosum ac lutulentum.
('That, and in fact most of them, should be floored with stone or tile, to prevent the goat-house from being wet and muddy.' Var. R. 2.3.6)
(v)   Scio ego, multos iam lucrum lutulentos homines reddidit.
('I know that profit has already made many men corrupt.' PI. Capt. 326)
As for the syntactic functions of satellites, two types of satellites are distinguished in § 2.10, those functioning as adjunct and those functioning as disjunct.
The last syntactic function of clause constituents to be discussed in this section is the secondary predicate. Two examples are (w) and (x), the first with a present participle, the second with an adjective. The terminology for these constituents varies from language to language and from author to author.18 They are optional constituents, which makes them resemble satellites. While they often refer to properties of nouns and other nominal constituents, they are not part of the noun phrase. In (w), for instance, flentes is related to principes but it is not an attribute—for which see § 2.4. Secondary predicates also occur with implied constituents (especially subjects), such as 'we' in (x). In (y),flens is related to the relative pronoun quae, the subject of its clause. Details are discussed in Chapter 21.
(w)   Postridie in castra ex urbe ad nos veniunt flentes principes.
('The next day their leaders came from the city to our camp, crying.' PI. Am. 256) (x)   ... scandentem moenia Romanae coloniae [et] Hannibalem laeti spectamus.
('.. .but now that Hannibal is scaling the walls of a Roman colony, we look on with
indifference.' Liv. 22.14.7)
,s In LSS the term 'complement' was used, but this has its own disadvantages. In this Syntax, complement will be used differently, as described immediately below.
17 The terms are taken from Quirk et al. (1985). 1S They are called 'praedicat.va in LSS: ch. 8.
Semantic, syntactic, and pragmatic functions   31
(y)   Is amore misere hanc deperit mulierculam / quae hinc modo flens abiit.
('He's dying wretchedly with love for this woman who went off crying a moment ago.' PL Cist. 131-2)
The third type of function used in this Syntax are pragmatic functions. Three factors play an important role in the structure of a sentence and in the choice of specific expressions. The first is the role of a particular constituent in its surround ing context—for example, whether it is already present in the preceding context or will be continued in the following context. The second factor is the speaker's or writers estimate of what the addressee already knows. The third factor concerns the importance attached to a particular constituent by the speaker and/or addressee for their communicative interaction. Constituents of a clause therefore differ in their form and degree of topicality and focality: normally a clause contains at least one entity that the speaker considers to some extent known to, or at least accessible for, the addressee (the topic of the clause), and which is therefore a good starting point for providing new or unexpected information (the focus of the clause).
In this Syntax, various types of topic and focus are distinguished. These pragmatic functions are especially relevant for word order, the choice between active and passive voice, and the selection of anaphoric expressions. Exx. (z)-(ab) illustrate the use of these terms for describing word order phenomena. In (z), Terentia, Cicero's wife, is of course well-known to Atticus, the addressee of the letter, so it is unproblematic for Cicero to take her as the starting point for the new information given in the remainder of the sentence. The initial position in the sentence is very common for topical constituents. In the immediately following sentence—here (aa)—there is no need to mention Terentia, duly introduced, again. That Cicero's daughter Tullia joins in for the greetings is an additional (see et 'also') and salient piece of information and placed at the end of the sentence. Ex. (ab) illustrates a specific focalizing structure to emphasize the role of Clodium (by contrast with alios) (a so-called cleft construction).
(z)   Terentia magnos articulorum dolores habet.
(Terentia has a bad attack of rheumatism.' Cic. Att. 1.5.8)
(aa) Et te et sororem tuam et matrem maxime diligit salutemque tibi plurimam adscribit et Tulliola, deliciae nostrae.
('She is very fond of you and of your sister and mother, and sends her best love, as does my darling litUe Tullia.' Cic. Att. 1.5.8)
(ab) Neque tu eras tarn excors tamque demens ut nescires Clodium esse qui contra leges facere, alios qui leges scribere solerent.
('And yet you were not so senseless and so infatuated as not to know that it was Clodius' part to act in defiance of the laws, and the business of others to formulate them.' Cic. Dom. 48)
32   Basic grammatical concepts
Details on pragmatic functions (and some of the problems involved) are given in Chapters 5, 23, and 24.19
In other studies in which systematic attention is paid to the pragmatic structure of clauses and sentences other terms are used for 'topic' and 'focus', for example 'theme' and 'rheme'. The term 'focalization also plays a role in narratological studies, in a sense that is not entirely unrelated to its use in this Syntax, but with which it should not be confused.
2.13 Discourse and text type
Generally speaking, people communicate with one another in units that consist of more than one sentence. Alternatively, in the case of an exchange between two or more participants in a communicative interaction, people form sentences together with the sentences of the interlocutor(s) (e.g. question-answer pairs). For all units above the level of the individual sentence, this Syntax uses the term discourse. Viewed from the top down, a discourse unit (a poem, a letter, a chapter, a book) may consist of smaller subunits (episodes and paragraphs), which themselves consist of a coherent set of sentences.
The linguistic properties of a unit of discourse (and hence ultimately of its sentences, clauses, words, etc.) depend upon a number of factors. In the first place, there are external factors related to its time and place, to the properties of its participants (sex and age. for example), and to their mutual relations. As illustrations of what is meant by properties of the participants, one may think of the intellectual level and/or linguistic ability of the participants (for example, Augustine's use of Latin in his Sermones differs from the use in his works for an audience of intellectuals whom he assumed knew 'Classical' Latin).20
Next, there are internal factors, depending on certain decisions made by the language participant(s). Among these are:
(i) medium (spoken or written), indirectly relevant to the study of Latin (even though we have only written material, think of drama and orations vs narrative and letters);
(ii) participation (monologue vs dialogue): in his Partitiones Oratoriae Cicero chooses the form of a dialogue between himself and his son;
(iii) formality, as imposed by social (including literary) convention: Cicero's orations against Antony are quite different from his orations before a jury;
19 For a description of Latin word order along pragmatic lines, see Spevak (2010a) and Hoffmann (2010).
30 See, for example, Dokkum (1900).
Discourse and text type 33
(iv) communicative purpose, which results in various text types, such as narrative texts, argumentative texts, didactic texts, poetic and prose texts of various types (this coincides partly with 'literary genre'). The effect of text type on the linguistic features of texts is vastly underestimated in Latin scholarship; one example: quoniam 'since' occurs some hundred and forty times in Lucretius and 'only' less than fifty times in Livy's much longer work. Lucretius' is an argumentative text, Livy's a narrative; almost all his instances occur in the orations, the most argumentative parts of his work;
(v) styi e, defined as the set of individual choices made by the speaker/writer(s) within limits imposed by items (i)-(iv), such as periodic vs non-periodic sentences, archaic vs contemporary expressions, analogy vs anomaly, adopting foreign elements (for example, Greek) vs purism.21
The coherence of discourse derives in the first place from the fact that the 'world' it describes has its own internal coherence, which is manifested in the semantic relations that exist between the individual sentences and paragraphs. These must share some overarching discourse topic. But there are also specific linguistic devices that enhance discourse coherence. In (a) at the beginning of this chapter, the connector at was mentioned. Connectors play an important role in marking inter-sentential relations and relations between other discourse units, such as paragraphs. A few further devices were mentioned in the paragraph on pragmatic functions in § 2.12. Anaphoric expressions, active/passive variation, the use of the tenses, etc. contribute to discourse coherence. Further discussion will be found especially in Chapter 24.
" For a discussion of such 'discourse decisions' see Dik (1997: U.415-22).