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

Next Section

9 Word order

Latin is often characterized as a language with a `free' word order. The term `free' suggests that in Latin the word order does not matter and is not determined by any rules. In this chapter it will appear that this suggestion is not correct. [1] This chapter must necessarily be of a tentative nature. In spite of the existence of a large number of studies on individual issues, research into Latin word order is still relatively underdeveloped, owing to the fact that little attention has been devoted to the function of word order in a coherent text. I discuss the following points. In crosssection 9.1. I make some remarks on the notion `word order' and the most current approach to word order. In crosssection 9.2. I deal with a number of factors that may determine word order, particularly the syntactic and pragmatic function of constituents and their internal structure. crosssection 9.3. concerns the relative order of constituents on the sentence level, crosssection 9.4. on the noun phrase level. In crosssection 9.5. I mention a number of literary/stylistic aspects of word order. crosssection 9.6. contains a brief typological exposition.

9.1. Definition of the notion `word order'

The term `word order' suggests that what is at stake is the position of individual words in a sentence. In reality, chapters on word order deal with constituents consisting of one word (e.g. the position of the adverb valde (`very')) as well as constituents consisting of more than one word (noun phrases, subordinate clauses.). I use `word order' in a general sense for `constituent order'. In most cases, the order of constituents is studied without their function in the sentence being taken into consideration, e.g. their relation with regard to the Predicate (whether they are argument or satellite) or their syntactic function. An exception is the Subject, which does receive special attention (see Sz. 401–2). In most cases, incidentally, the position of the Subject is merely understood as the position of the noun or noun phrase in the function Subject, rather than, for instance, embedded predications in the function Subject. Furthermore, the position of a constituent is often examined in isolation, without taking into consideration the structure of the whole sentence in which such a constituent occurs. Thus, much attention is devoted

-- 164 --

to the initial, middle and end positions of a sentence and to the distribution of the constituents over those positions. Clearly, however, such an approach can possibly be useful only in the case of sentences consisting of three constituents, and much less so in the case of sentences with fewer or more than three constituents. An important factor for the relative order of Subject and Predicate, for instance, can be the presence or absence – apart from a Subject constituent and a predicate – of a satellite which locates the state of affairs in time or space (see p. 29). In other words, the relative order of constituents is determined in part by the number of constituents that may occur in a certain position. In this respect little research has been done for Latin. [2]

It is all the more difficult to make statements on the order of constituents in Latin due to the phenomenon of discontinuity. Two types may be distinguished. First, there are words in Latin that are more or less automatically put in a certain position and, as it were, do not care whether or not they split up another constituent in doing so. An example is the connector enim (`for'), which as a rule occupies the second position in the sentence (sometimes the third position, especially if words of certain types precede, such as monosyllabic words; for details and exceptions see TLL s.v.). It need not, however, be moved towards the end of the sentence in order to preserve the continuity of a noun phrase; see instances such as (1):

(1) de civitatis enim iure … disceptamus (`For we are discussing civil law', Cic. Balb. 29)

A second type of discontinuity may be discerned in cases where a constituent is split up for pragmatic or other reasons. A well-known kind of example of discontinuity of a noun phrase is (2).

(2) hic optimus illis temporibus est patronus habitus (`In those days he was considered the best lawyer', Cic. Brut. 106)

In instances of this kind it is very difficult to say what the position of, for example, the Subject Complement is: is this merely the Head (patronus), or does the Attribute optimus play a role as well? (see also crosssection 9.5.).

9.2. Factors influencing the word order

Apart from the number of constituents, there are several other factors that influence the order of the constituents in a sentence. These factors in part involve the syntactic function, in part the pragmatic function of constituents within a sentence (or noun phrase). Furthermore, the lexical category (or categories) to which a constituent belongs and the internal structure of a constituent (e.g. its size and its internal complexity) can influence the word order. Finally, the sentence type (see chapter 10) also plays a role.

-- 165 --

9.2.1 Syntactic and pragmatic factors

English is usually considered a typical fixed word order language, in which syntax seems to play a predominant role. Thus, the order of the arguments with regard to the predicate seems to be quite fixed, as in (3):

(3) Mary gave the visitor a glass of milk [3]

Any inversion of this order would be strange, unless some situational or contextual element were to make a different order more appropriate. In English the finite verb has a kind of privileged position. The Subject normally precedes, and the two Objects follow. However, so-called fronting of non-Subject constituents is acceptable in certain pragmatic conditions. Quirk et al. (1985: 1377) mention (4):

(4) Most of these problems a computer could take in its stride

Here an Object constituent precedes the finite verb. As is suggested by the presence of these, the Object contains known, topical, information. A computer, as appears from the presence of the indefinite article, does not represent information introduced earlier.

From the examples given above it appears that also in English the word order is to some extent `free', in the sense that the flow of information contains as a rule certain factors that favour or necessitate a certain order. Our knowledge of English enables us to identify such factors. In this connection, intonation is an important element. For Latin, data concerning intonation are lacking, and consequently it is much more difficult exactly to determine which factors underlie a given word order. For this reason, a given word order often seems to be arbitrary, or the impression is created that anything is possible. Further research will have to show to what extent the diverging data on Latin word order can be explained on the basis of pragmatic factors such as those mentioned above, and to what extent, moreover, syntactic factors play a role. In crosssection 9.3. and crosssection 9.4. a number of aspects are discussed.

9.2.2 The influence of the lexical category and the internal structure of constituents

Constituents of different categories are subject to different placement rules. To connectors (igitur, itaque (`therefore'), etc.), for instance, other rules apply than to the – semantically closely related – anaphoric adverbs (ideo, idcirco (`for that reason'). Also igitur and itaque themselves are subject to different word order rules. Sometimes, for constituents belonging to different lexical categories different placement rules apply, even if they fulfil the same syntactic function in the sentence. To illustrate this I give two examples. An English example is (5):

(5) Mary gave to the visitor a glass of milk

-- 166 --

In many respects, this sentence is similar to (3); it differs in that here the function Indirect Object is fulfilled by a preposition phrase (to the visitor). This preposition phrase turns out to be more `mobile' than the constituent the visitor in (3), cf.:

(5') Mary gave a glass of milk to the visitor

(3') * Mary gave a glass of milk the visitor

Something similar may be said with regard to German. There arguments in the accusative occupy different positions in the same sentence with regard to other arguments, depending on whether a reflexive pronoun, another type of pronoun, a definite noun phrase or an indefinite noun phrase is involved.

Latin grammars occasionally pay attention to this phenomenon. For example, Purpose Adjuncts in Caes. Gal. behave differently, according to their syntactic category. Purpose Adjuncts of the type pacis petendae causa (`in order to ask for peace') precede the finite verbs, whereas Purpose Adjuncts in the form of a subordinate clause follow the finite verb. An example of the latter type is (6):

(6) Sittius est … missus ut eam provinciam perturbaret (`S. has been sent in order to stir up trouble in that province', Cic. Sul. 56)

Possibly, this difference in behaviour not only has to do with the difference in category (preposition phrase/subordinate clause), but also with the greater complexity of the Purpose clause, a factor that will be treated below. Another indication for the existence of differences in word order behaviour between constituents belonging to different categories may be found in the observation that the behaviour of embedded predications of the type AcI differs from that of the corresponding embedded predications in the form ut + subjunctive. [4]

The influence of the internal complexity of a constituent on the word order can be illustrated in more detail with the aid of the following examples:

(7a) ?John gave Peter a book that had received lavish praise from reviewer X in last week's TLS on his birthday

(7b) On his birthday John gave Peter a book that had received lavish praise from reviewer X in last week's TLS

(7c) John gave Peter a book on his birthday that had received lavish praise from reviewer X in last week's TLS

Owing to the length of the relative clause the Object constituent in (7a) is so `heavy', that this word order is generally avoided, although all syntactic and semantic rules have in principle been applied correctly. The speaker then has two alternatives. He may order the constituents in such a way that the whole Object constituent is placed at the end of the sentence (7b), or he splits up the Object constituent into two parts (thus forming a discontinuous constituent). The choice between (7b) and (7c) is in its turn determined by pragmatic

-- 167 --

factors. This is an example of a general tendency of languages to determine the placement of constituent in a sentence according to their complexity. In Latin, there are examples which point in the same direction, but one wonders whether the influence of internal complexity is as strong in Latin as in English: there are also quite a few counter examples. I first give two examples that are parallel to (7b) and (7c):

(8) illi mihi disseruisse videntur de poena eorum qui patriae, parentibus, aris atque focis suis bellum paravere (`They seem to me to have spoken of the punishment of those who have waged war against their own country, parents, shrines and homes', Sal. Cat. 52.3)

(9) allata est enim epistula Athenis ab Archino uni ex his Archiae, qui tum maximum magistratum Thebis obtinebat, in qua omnia de profectione eorum perscripta erant (`For then a letter was brought from Athens to one of them, Archias, who at the time held the highest office at Thebes, in which everything about their departure had been written down', Nep. Pel. 3.2)

Instances of discontinuity such as exemplified in (9) are avoided by `Classical' authors (see Sz. 692); also, the instances in which a constituent consisting of Head + relative clause is placed after the finite verb seems fewer than those in which such a complex constituent precedes the finite verb. Statistically, therefore, instances such as (10) predominate: [5]

(10) sed filii familiarum, quorum ex nobilitate maxuma pars erat, parentes interficerent (`But of a number of families the sons, largely belonging to the nobility, would kill their parents', Sal. Cat. 43.2)

That the tendency observed for English cannot simply be applied to Latin also becomes clear from the fact that ut-clauses occupy different positions in the sentence, depending on the question of whether argument or satellite clauses are involved. [6]

9.2.3 The influence of sentence type and the distinction main sentence/subordinate clause

In some languages, such as Dutch and German, there is a difference between the position of the finite verb in main sentences and subordinate clauses. Furthermore, in interrogative and imperative sentences in such languages the finite verb occupies a different position from that in declarative sentences. As for Latin, we know that in imperative and concessive sentences the finite verb is often placed at the beginning of the sentence (for examples see Sz. 403; K.–St. II.598–9). See (11):

(11) sit fur, sit sacrilegus, sit flagitiorum omnium vitiorumque princeps; at est bonus imperator … (`Let him be a thief, a profaner, a champion of all

-- 168 --

crimes and disgraceful acts; yet, he is a good leader', Cic. Ver. 5.4)

The same is often the case in interrogative clauses. For subordinate clauses this is less clear. For Caesar it has been demonstrated that in subordinate clauses finite verbs are placed at the end even more often than in main sentences (Sz. 403), but this is merely a difference of degree. In other authors, too, only a small difference has been found, which need not necessarily be related to the difference main sentence: subordinate clause. [7]

9.3. Order of constituents on the sentence level

Above it has already been remarked that relatively little systematic research has been done on the relative order of constituents on the sentence level and the factors influencing this order. First, I give a brief summary of the standard approach to be found in the large grammars ( crosssection 9.3.1), and then offer some comments on some aspects of this approach ( crosssection 9.3.2). These comments are not based on enough research for them to be accepted as `facts'. Rather, they are intended as indications as to how more reliable data might be obtained.

9.3.1 Survey of the standard approach

The description of word order in Sz. (397 ff.) is to a large extent determined by considerations with regard to the prehistoric Indo-European situation. Thus, in Indo-European certain categories of words, viz. `particles', [8] `conjunctions' (i.e. connectors and subordinators), (certain) pronouns and other `unaccented' words, are said to have been placed `clitically' in the second position of the sentence, and consequently as a rule connected to the Subject constituent, for which the initial position had more or less naturally been reserved (Sz. 401; see K.–St. II.592; 597). According to the grammars, to some extent this also holds for Classical Latin, although some words belonging to the categories mentioned above usually occur in initial position. Furthermore, it is remarked that the initial position is often occupied by non-Subject constituents, which are, as a result, `emphasized'; in more modern terms, we might say that such constituents have (contrastive) Focus or Theme function in the sentence. The Subject constituent, on the other hand, may be placed at the end of the sentence, and thus receive `emphasis', i.e. have Focus (Sz. 402; K.–St. II.597–8). [9]

On the position of the finite verb in Indo-European there is little agreement among scholars. Sz. (402) and K.–St. (II.598) assume that the finite verb did not have a fixed position. Others, however, assume either a fixed position for the finite verb (viz. end position) or a certain relative order of the finite verb and its arguments Subject and Object. Some postulate the order S(ubject) O(bject) V(erb), others SVO. [10] In Latin the finite verb is said to be most often found at the end, both of main sentences and of subordinate clauses. There are, however, great differences between individual authors (or genres) and

-- 169 --

even within authors. In Plautus, for instance, the finite verb is very frequently followed by other constituents (among which also an Object constituent). In Caesar the finite verb is placed at the end of 84% of main sentences (Gal. II). This percentage varies from 32% to 52% in his contemporary Cicero, and is quite low in Varro (33%). In topographical digressions, however, Caesar does not place the finite verb at the end of the sentence (e.g. Gal. 1.1.5–7). [11]

According to K.–St. (II.611–12) non-Subject arguments as a rule precede the finite verb. Thus, SOV is considered the `normal' order for Classical Latin. Also the Praedicativum [12] and the Subject Complement precede the finite verb (K.–St. II.611). The least fixed position with regard to the finite verb seems to be that of satellites (`Adverbien und adverbiale Bestimmungen'; K.–St. II.613, in a very carefully formulated statement).

9.3.2 Discussion of the standard approach

I now discuss some of the `rules', or rather tendencies, summarized above: the position of sentence-connecting words and the special position of pronouns ( crosssection; the initial position of the sentence ( crosssection; the final position of the sentence ( crosssection; the relative order of constituents ( crosssection; presentative sentences ( crosssection Finally, I make some concluding remarks in crosssection Position of sentence-connecting constituents and pronouns

(a) Sentence-connecting constituents

Sentence-connecting constituents usually occupy a position at the beginning of the sentence. Examples of such constituents (for a more elaborate discussion see chapter 12) are connectors (enim, nam (`for')) and various categories of anaphoric words, e.g. anaphoric adverbs (ideo (`therefore'), tunc (`then')), anaphoric pronouns (is (`this'), ille (`that'), talis (`such')). The difference between connectors and anaphoric words is that connectors merely link sentences, whereas anaphoric words both create a connection between (constituents of) two sentences and themselves fulfil a function in the sentence, e.g. Subject, Adjunct or Attribute. In correlation with the latter property of anaphoric words, constituents formed (in part) by such a word show a greater mobility than connectors. The latter as a rule occupy the first or second position of the sentence (see also above, p. 165): sed (`but')/autem (`however'), nam/enim (`for'), itaque/igitur (ergo) (`therefore'). This position at the beginning of the sentence, especially the occurrence in the second position, is generally explained in the grammars as a relic of the rule assumed for Indo-European that certain unaccented words were placed immediately after an accented word in the first position (e.g. the Subject). Another indication for the continued existence of this rule in Latin is seen in the preference of words of different types, e.g. est, quis, quisque, for the second position of the sentence. Synchronically, the placement of connectors and anaphoric words relatively

-- 170 --

early in the sentence can best be explained on the basis of their sentence-connecting function. There is no synchronic explanation for their specific (first or second) position.

The more or less automatic placement of connectors in the first or second position is often abandoned, in poetry especially for metrical reasons, but also in those cases in which the prominent position at the beginning of the sentence is reserved for another constituent. We see, for instance, that connectors that normally occur in initial position (sed, nam, itaque) are preceded by a word or noun phrase (sometimes even noun phrases) with Focus function. See example (12):

(12) pro ingenti itaque victoria id fuit plebi (`The people saw this, therefore, as an enormous victory', Liv. 4.54.6)

Connectors `normally' occurring in the second position (e.g. enim) are also sometimes found later in the sentence. Something similar applies to subordinators and interrogative pronouns, which normally occupy the first position of their clause, but are not seldom placed away from the immediate beginning of the sentence, generally in order to make room for a Topic or Focus constituent (for examples see K.–St. II.614–15). This often happens if the subordinate clause contains anaphoric constituents, as in (13), [13] but there are also instances such as (14), where omnem (and, consequently, the entire preposition phrase per omnem urbem) is emphatically placed at the beginning of the sentence:

(13) de his rebus cum ageretur apud Caesarem (`When these things were being discussed with C.', Caes. Civ. 3.109)

(14) per omnem urbem quem sum defessus quaerere (`I have exhausted myself looking for him all over the city', Pl. Epid. 197)

The occurrence of words of these categories in the second or later position is rather caused by the need to place other words in the first position of the sentence, than by the fact that the words themselves are unaccented. [14]

(b) Pronouns

As we have already seen in (a), anaphoric adverbs and pronouns qualify for the first position in the sentence (P1) on account of their cohesive function. In the lexicon on Livy this clearly appears for words such as ibi (`there'). One notes that pronouns of various kinds do not very often occur after the finite verb and rarely at the end of the sentence (see Koll 1965: 247; Wanner 1987: chapter 3). This still holds true for the Romance languages (in varying degrees), in which unaccented pronouns in the function Object [added 01-09: often]occupy a proclitic position with regard to the finite verb (SOV), although nominal Objects usually follow the finite verb (SVO). By some this placement of pronouns is seen as a relic of the pre-Romance SOV situation in Latin. Others

-- 171 --

view this phenomenon as an example of the general tendency of relatively `light' constituents to be placed at the beginning of the sentence. [15] First position for a `true' sentence constituent

For a long time already, the first position has been viewed as the pre-eminent position for constituents that convey salient information (see Sz. 401), [16] leaving out of account the connectors treated in the preceding section and Theme constituents (see crosssection 4.2.). As an example of such a constituent I repeat example (25) (see p. 37):

(15) de forma, ovem esse oportet corpore amplo (`As for the form, a sheep ought to have a large body', Var. R. 2.2.3)

The constituent de forma has no syntactic relation with the `main sentence', but does indicate the frame within which the remainder of the utterance is to be interpreted: we are talking about the forma ovis. Even though the grammars do not say so explicitly, it may be assumed that they, too, leave these two types of constituent out of consideration in discussing the first position of the sentence.

As has already been remarked above, according to the grammars the Subject constituent is most qualified for the first position of the sentence. They also create the impression that in a majority of sentences a Subject constituent does in fact occupy the first position. This impression is not correct. It is a characteristic of Latin that for the first and second persons the finite verb shows that the Subject is the speaker or addressee of the situation at stake. The presence of the Subject pronoun ego/tu usually involves Focus or Topic change. The identity of the third person (= non-speaker, non-addressee) must be inferrable from context or situation. The rule is, however, that the Subject need not be indicated explicitly if the preceding context makes clear who (or what) is concerned. In many texts, therefore, there is no (explicit) Subject constituent, and in such sentences the first position is occupied by a non-Subject constituent. [17] Naturally, the rule could be that an explicit Subject constituent, if present, occupies the first position. An exemplary analysis of Cic. Att. 1.5 shows that this is not the case (see figure 9.1). [18] The analysis demonstrates that in only 3 out of 27 sentences in this letter the Subject of the sentence occupies the first position (the numbers corresponding to these sentences are printed in italics). Sentences 1 and 26, too, have a `simple' explicit Subject (tu and Tulliola, respectively), but these are not in the first position. In a number of instances a constituent from an embedded predication occupies the first position. This constituent is not always the Subject of the subordinate clause. There is a high number of relative and interrogative pronouns in the first position. Relative pronouns, of course, are immediately linked to previous information and are, therefore, Topic. Interrogative pronouns are, of course, Focus. In this letter three sentences are introduced by a subordinate clause or a preposition phrase that announces the

-- 172 --

next subject, and thus indicates, as it were, a frame for the content of the sentence that follows, i.e. is Theme (sentences 4, 9 and 13). [19] It turns out that – leaving aside connectors and Theme constituents – the first position of the sentence is in fact reserved for Topic and Focus constituents. This may be the Subject constituent, but in most instances this is not the case here. In this connection, the subsequent sentences 23 (Quintum fratrem (Object)) and 24 (Terentia (Subject)) are interesting. Sometimes a constituent has not yet been mentioned in the letter, but can nevertheless be interpreted as Topic on account of the familiarity of Cicero and Atticus. This not only applies to Quintum fratrem and Terentia, but also to Epiroticam emptionem in sentence 20. [20]

Figure 9.1: Cic. Att. 1.5

-- 173 --

Figure 9.1 continued

-- 174 --

Figure 9.1 cont.

1 The first position is occupied by the Object constituent of a dependent question within the embedded predication with potes (N.B. explicit Subject tu).

2 Complement (personal pronoun), possibly to be interpreted as contrastive with regard to tu in sentence 1.

3 Satellite (connecting relative).

4 Satellite (Theme); subsequently Subject Complement.

5 Subject of embedded predication (AcI) (connecting relative).

6 Satellite (Cause) (anaphoric pronoun).

7 Satellite (translated by Shackleton - Bailey as Theme).

8 Satellite (Time Position).

9 Satellite (Theme); subsequently independent relative clause in the function Object.

10 Predicate (one-place; `heavy' Subject).

11 Satellite (temporal clause); after subordinator Satellite (Duration).

12 Satellite embedded predication (Place from which); (preposition phrase with personal pronoun).

13 Satellite (Theme); subsequently Predicate.

14 Auxiliary verb.

15 Subject (personal pronoun).

16 Subject of dependent question (question word).

17 Object in satellite clause (connecting relative).

18 Satellite (translated by Shackleton - Bailey as Theme).

19 Object of embedded predication (AcI).

20 Subject of embedded predication (AcI).

21 Object in independent relative clause in the function Object.

22 Subject (personal pronoun).

23 Object.

24 Subject.

25 Object (personal pronoun).

26 Object (N.B. explicit Subject). a. The text is uncertain; I follow Shackleton - Bailey

27 Predicate (imperative); (`heavy' subordinate clause).

-- 175 --

The picture created by this analysis of a coherent text from Cicero is enhanced if we examine in a number of texts the word order behaviour of a certain lexeme in the function Subject or Object. The tables given below contain the required information. In Cic. Att. 14 instances have been examined of frater in the function Subject, and 8 instances of fratrem in the function Object (see table 9.1). [21] Frater (Subject) is not in the first position if there is another constituent that has Topic function or is placed in initial position for reasons of contrast (Focus) and/or if frater has Focus function:

(16) (nihil mihi nunc scito tam deesse quam hominem … qui me amet …) abest enim frater ἀφελεστατος et amantissimus. Metellus … Tu autem … Ita sum ab omnibus destitutus (`You should know that there is nothing that I miss so much as a man who loves me. My brother, so utterly artless and beloved, is absent. Metellus … You, however, … Thus all have left me', Cic. Att. 1.18.1)

(17) de Metello scripsit ad me frater quantum speraret profectum esse per te (`As for Metellus, my brother wrote to me that it had gone for you as he hoped', Cic. Att. 3.22.2)

In (16) the `absence', already introduced in the preceding context, is the Topic and frater … amantissimus Focus. Note that subsequently the other absent persons (Metellus and tu) are placed contrastively at the beginning of the sentence. In (17) the constituent de Metello represents a new name in an enumeration. Its pragmatic function is difficult to determine. De Metello clearly indicates the person about whom something is predicated in the sentence and thus seems to have Topic function (cf. the definition of Topic on p. 4). Although in this letter Metellus has not been mentioned before, Cicero and Atticus naturally both know him. De Metello might be said to be in contrast with those mentioned in the preceding context (contrastive Focus?). Incidentally, this difficulty in determining the pragmatic function (especially Topic function) is a recurrent problem in Cicero's letters, as Cicero is concise in his statements and abruptly switches from one subject to another. The six instances of fratrem in the first position all occur in sentences without an explicit Subject. The two other instances occur in sentences with an explicit Subject, viz. examples (18) and (19). The former contains ego (Topic in the sentence,) which corresponds with te at the beginning of the letter. The latter contains a Subject reinforced by quidem, with which de fratre autem, somewhat further on, corresponds; moreover, fratrem has Focus function, in view of the preceding question (see also crosssection

(18) ego aestivis confectis Quintum fratrem hibernis … praefeci (`after the campaign I have put my brother Q. in charge of the winter camp', Cic. Att. 5.21.6)

(19) Quem relinquam qui provinciae praesit? Ratio quidem et opinio hominum postulat fratrem (`Whom shall I leave behind to govern the province? My brother would be the most obvious and expected choice', Cic. Att. 6.3.1)

-- 176 --

Table 9.1 Position of frater (subject)/fratrem (Object) in Cic. Att.a

The following passages have been scored: Position in the sentence (regardless of connectors):

fratrem S5.21.6.1+ S S6.3.1.12+ S S S S S
-S: no explicit Subject in the sentence
+ S: explicit Subject

I have chosen forms of the same lexeme in order to exclude any possible variables that might influence the word order.

We come to a similar conclusion if we look at the position in the sentence of tu (Subject) and te (Object) in Cic. Att. 1 (table 9.2).

From these (limited) data it may be concluded with some caution that the first position of the sentence – leaving out of account connectors and Theme constituents – tends to be reserved for constituents with the pragmatic function Topic. Topic constituents are often, but not necessarily, Subject constituents. [22] Other candidates for the first position of the sentence are constituents with contrastive Focus. An example is the sentence with numquam in Att. 1.5.3 (sentence 8 in figure 9.1). (Also constituents that form part of a noun phrase can have Focus function, e.g. attributively used adjectives, see crosssection 9.4. on p. 185). This regularity can be represented in the following formula:
Rule 1 (conn) P1 (conn) {Pred, arg, sat} where conn = connector, P1 = first position of the sentence, Pred = Predicate, arg = arguments(s) and sat = satellite(s).

-- 177 --

Table 9.2 Position of tu (Subject)/te (Object) in Cic. Att. 1a

a Position in the sentence (regardless of connectors)

te S1.1.5.3- S1.3.2.1+ S S1.9.2.7+ S1.19.1.3- S S1.16.9.9+ S S1.16.16.4- S S
-S: no explicit Subject in the sentence
+ S: explicit Subject
imp.: imperative sentence

This formula should be read as follows: a sentence is often, but not always, linked to the preceding context by a connector. One privileged position (P1) is available for a Topic or contrastive Focus constituent, which may be the Predicate, an argument or a satellite. Then the other sentence constituents follow. 22a

The grammars devote special attention to the occurrence of the finite verb in the first position (for examples see K. – St. II. 598–601; Sz. 403). [23] Apart from the occurrence of the finite verb in the first position in interrogative, imperative and concessive sentences, this placement is said to serve to create cohesion between sentences, e.g. in the form of an explicative, consecutive or causal relation, or merely to create discourse continuity. An example of a consecutive relation given by K.–St. (II. 600) is (20):

(20) (The Helvetians were forced to abandon their attempt to cross the Rhone) relinquebatur una per Sequanos via (`Only the one road through the territory of the Sequani remained', Caes. Gal. 1.9.1)

K.–St. translate by `es blieb also übrig'. Note that at the beginning of the passage (in 1.6.1) Caesar has remarked erant omnino itinera duo (`In all there were two roads'). One of these is now said to have become impossible. Logical calculation then naturally leads to the content of (20), but this need not necessarily be the reason of the placement of the finite verb in the first position.

-- 178 --

Incidentally, the grammars give so many different possible reasons for placement of the finite verb in the first position that any individual instance can be explained. In crosssection on p. 179 ff. I return to (20) and possible explanations of the relative order of Predicate and Subject. [24] The final position of the sentence

As has already been pointed out, the finite verb often – albeit with varying frequency – occurs in the final position of the sentence. The grammars point out that the final position can also be used in order to emphasise certain constituents, especially Subject constituents (see also below crosssection

In Cic. Att. I have examined the final position in 126 sentences. This examination has resulted in the data in table 9.3. Table 9.3 The final position of the sentence in a number of letters from Cicero to Atticusa
Clause LevelType of ConstituentNumber of Instances
Main clause:Finite verb52
Embedded predication:66
Total number of instances126

a Cic. Att. 1.3–10.

Then I examined the 8 non-finite verb constituents (table 9.3, no. 2) to see why they occur in the final position of the sentence. Not all are easy to explain, though most do, indeed, involve constituents containing salient information (Focus constituents). A clear example (Object constituent in the final position) is (21):

(21) et (Quintus frater) secum habebat χρηστομαθη hominem, D. Turranium (`And he brought with him a scholarly man, D.T.', Cic. Att. 1.6.2)

In this sentence Quintus frater is Topic (not explicitly present). The most salient information (Focus) is indeed in the final position Habere is a relatively `colourless' verb. What Cicero wants to say is not that his brother has brought a scholar with him, but who or what kind of person he has brought with him. A clear example of a satellite in the final position is (22):

(22) verum haec audies de Philadelpho (`But this you will hear from P.', Cic. Att. 1.11.2)

Here, too, de Philadelpho contains the most important information (Focus). (23), where we find a Time Position Adjunct in the final position, however, is problematic: [25]

-- 179 --

(23) pater nobis decessit a.d. VIII Kal. Dec. (`Father died on the 23rd of November', Cic. Att. 1.6.2)

If the type of explanation we have adduced for the two preceding instances were applicable also to (23), the most salient part of Cicero's statement to Atticus would be the date of his father's death, whereas a modern reader might expect the news of his death to be more important than the date. We must assume that Cicero had already told Atticus of his father's imminent death, so that this was no longer the most important information. [26]

I have limited myself to those instances in which the finite verb is not in the final position of the sentence. From a methodological point of view, it is naturally not correct to examine only the non-finite verb constituents for Focus function. If in those cases the final position is reserved for Focus constituents, one should also examine whether the finite verb has Focus function in all those instances (52 in this sample) in which it occurs in the final position. A priori, considering the very high frequency of the finite verb in the final position and the considerable differences between authors, this is unlikely. Moreover, it should also be demonstrated that constituents of the type treated above are always placed after the finite verb (and even in the final position of the sentence) if they have Focus function. Such research has not yet taken place on a large scale (see also p. 182). [27]

In the preceding sections sentences were analysed in terms of topical and focal constituents. In reality, however, the problem is even more complex than described already. This can be illustrated with the following sequences:

(24a) What did you do today? :: I bought a hat

(24b) What did you buy? :: I bought a hat

(24c) You don't like hats, do you? :: I even bought a hat today.

We see that the same sequence can either function as a pragmatic unit or be split up. It is to be expected from what we have said so far that the `split' cases will be placed in final position. However, what about example (24a)? For Latin, as far as I know, this kind of question has not been asked. The relative order of constituents

It is very difficult to get an insight into the `normal' order of the constituents of main sentences, since most sentences contain various kinds of embedded predications. There are relatively few simple sentences of the type:

arg 1 (Substnom) – arg2 (Substacc) – Sat (adverb) – Vf

(where Vf = finite verb).

Within this limited number there are differences in the number of arguments (valency), active/passive, explicit/non-explicit Subject and in the number and

-- 180 --

type of satellites. Few data are available in which these factors have been taken into account. By way of illustration of the type of approach necessary for a more thorough and detailed description of Latin word order, I present two tables (tables 9.4 and 9.5 below) which indicate the relative order of arguments, finite verb and satellites in simple sentences. In Cic. Att. 1 there are c. 100 sentences without embedded predications. Of these, 52 are active and have a two-place predicate with an explicit Object or Complement constituent (21 also have an explicit Subject constituent). A number of them contains one or more satellites, at the most three in one sentence. The satellites have been numbered according to the order in which they occur, rather than according to the type to which they belong. [28]

If the data obtained from this sample were representative, the following word order `rule' could be formulated for sentences with a two-place predicate:

Rule 2 (conn) – (S) – (sat) – O/C – (sat) – Vf

Where S = Subject, O = Object, C = Complement and `()' = may be omitted. This rule must be read as follows: in sentences with a Subject, an Object (or Complement) and a finite verb these three constituents as a rule occur in the order S – O/C – V. If the S is non-explicit, the order is O/C – V. Connectors, if present, have a fixed position at the beginning of the sentence. Satellites, if present, may occur in various positions, but usually precede the finite verb. Examples are:

(25) Terentia magnos articulorum dolores habet (`T. has severe pain in her joints', Cic. Att. 1.5.8)

(26) nos hic incredibili ac singulari populi [de] voluntate de C. Macro transegimus (`here I have dealt with the case of C.M. with the unbelievable and exceptional assent of the people', Cic. Att. 1.4.2)

Table 9.4 Thirty - one subjectless sentences with two - place predicates in Cic. Att. 1
O/CVbfinSat. 1Sat. 2Sat. 3
Sat. 171541
Sat. 2221
Sat. 3
N.B. The constituent on the vertical axis precedes the constituent on the horizontal axis; e.g. the Object/Complement constituent precedes the finite verb in 24 cases.

-- 181 --

Table 9.5 Twenty - one sentences with explicit Subject constituent and two - place predicates in Cic. Att. 1
SO/CVbfinSat. 1Sat. 2
Sat. 12794
Sat. 2144
N.B. the constituent on the vertical axis precedes the constituent on the horizontal axis. There are exceptions to the `rule' (rule 2), as appears from tables 9.4 and 9.5. The following orders are statistically deviant: O/C – S: example (27); Vf – O/C: example (28); Vf – sat: example (29); Vf – S: example (30):

(27) provincias praetores nondum sortiti sunt (`The praetors have not yet drawn lots for the provinces', Cic. Att. 1.13.5)

(28) obieris Quinti fratris comitia (`You will be present at the election of my brother Q.', Cic. Att. 1.4.1)

(29) ego de meis ad te rationibus scripsi antea diligenter (`I have already exactly informed you of my business', Cic. Att. 1.2.1)

(30) (incredibilis erat severitas …) nihil impetrabat reus, plus accusatory dabatur quam postulabat (`The severity was incredible; the defendant did not get anything done, but to the accuser more was given than he was asking for', Cic. Att. 1.16.4)

For the most part these exceptions can be explained pragmatically. In (27) the argument placed in the first position is Topic. In (29) antea diligenter is placed at the end of the sentence as an emphatic reproach, corresponding with tam diu nihil litterarum (abs te tam diu nihil litterarum (`For so long from you not a single word')). Quinti fratris comitia in (28) contains the most salient information (Focus). [29] In (30) nihil is placed contrastively in the first position, just like plus. Reus is in the final position, because it represents the most spectacular part of the information. One would have expected a defendant at least to get something done.

In this limited material exceptions can, therefore, for the most part be regarded as pragmatically determined deviations from the basic order Subject – Object/Complement – finite verb, or, in a different formulation, first argument – second argument – finite verb. Satellites are normally inserted somewhere between the first argument and the finite verb. Note that the position of the finite verb in Subjectless sentences (table 9.4) is less fixed than in sentences with an explicit Subject (table 9.5).

-- 182 --

I have examined the other active sentences in the material, with one – and three-place predicates, to see whether they can be regarded as variants of the two-place predicates. [30] Most of the instances do indeed turn out to fit in, while the Indirect Object may occur both before and after O/C. Exceptions are:

(31) salutemque tibi plurimam adscribit et Tulliola, deliciae nostrae (`And best wishes you get also from T., my darling', Cic. Att. 1.5.8)

(32) prensat unus P. Galba (`Canvassing is done only by P.G.', Cic. Att. 1.1.1)

(33) (nosmet ipsi … cotidie demitigamur) instat et urget Cato (`I myself am softening every day, but Cato is pressing the case with energy', Cic. Att. 1.13.3)

Like (27) – (30), these exceptions, too, can as a rule be explained on pragmatic grounds.

In crosssection–3 it has been stated that there are good reasons for assuming that the first position of the sentence is reserved for Topic constituents or constituents that contrast with preceding constituents (contrastive Focus); for the time being it is not altogether clear whether a position is also reserved at the end of the sentence (after the finite verb) for Focus constituents. Yet, in this section – with the reservation on account of the limited number of examples – it turns out that statistically a certain order predominates, which may be formulated as arg 1 – arg 2 – Vf. The conclusions of the preceding sections, represented in rule 1 on p. 176), can be combined with those of this section into the following formula:

Rule 3 (conn) – P1 – (conn) – (arg 1) – (sat) – arg 2 – (sat) – Vf

In most cases, at least in simple sentences, P1 is occupied by arg1, if explicitly present. [31]

Many linguists assume that in a sentence the information is ordered in such a way that old, known information precedes new information. In rule 3 above this view may be recognized, since P1 is reserved for constituents with Topic function. We have seen, however, that the placement of the other constituents in the sentence is relatively free: satellites may precede or follow the second argument. The finite verb is usually at the end of the sentence, but Object constituents and satellites may also follow it. These data are difficult to reconcile with the idea of a progression from old to new information. In the framework of the so-called Functional Sentence Perspective of the Prague School Panhuis (1982) has tried to demonstrate that this progression from old to new information is indeed relevant. He has examined the word order in 57 sentences from Caesar with three-place predicates which indicate movement (e.g. mittere (`to send')). He arrives at the following rule for the `normal' order of constituents:

arg 1 – arg 2 – arg 3 – sat – Vf [32]

-- 183 --

Pragmatic factors may cause other orders. For Plautus he has done similar research. There he sees as the basic order:

arg 1 – Vf – arg 2 – arg 3

Thus, apart from the position of the finite verb, the order of the arguments is the same. (I return to the difference between Plautus and Caesar in crosssection 9.6.). According to Panhuis, this order is not coincidental. And example from Caesar is (34):

(34) Helvetii … legatos ad eum mittunt (`The H. sent ambassadors to him', Caes. Gal. 1.13.2)

Panhuis remarks [33] that the action of mittere is successively specified by legatos and ad eum, something which is, according to him, reflected by the order. Satellites, in their turn, would follow the arguments, narrowing down the meaning even further. For the time being I cannot see why with a verb such as mittere the Direction Complement would be `more specific' than the Object. The reverse is just as conceivable. Further research is required. [34] Presentative sentences

So far we have discussed sentences in which a distinction can be made between topical and focal constituents. However, in many situations such a distinction is difficult or even artificial. An example is the following question – answer sequence:

(35) What happened? :: A bird flew against the window

The answer contains only elements that are unknown to the hearer and a distinction between Topic and Focus is impossible. Stories or episodes often start with entirely new information. Examples of this type of `brand new' sentences are the following:

(36) erant omnino itinera duo (`in all there were two roads', Caes. Gal. 1.6.1)

(37) relinquebatur una per Sequanos via (`Only the one road through the territory of the Sequani remained', Caes. Gal. 1.9.1)

Sentences of this kind, in which new entities are brought into the discourse, are often called `presentative sentences'. They often contain verbs like existential esse, apparere (`to appear'). Sometimes they contain some situating constituent in initial position, as in:

(38) apud Helvetios longe nobilissimus fuit et ditissimus Orgetorix (`Among the H. O. was by far the most respected and the richest,' Caes. Gal. 1.2.1)

In this sentence Orgetorix must have been the most unknown element in the sentence. Therefore, we might call it Focus, taking apud Helvetios as topical

-- 184 --

information. However, it may also be described like other presentative sentences.

It has been observed in the literature that in sentences with esse and other intransitive verbs Subjects are found more frequently in final position. In some of the Romance languages the same word order can be found as in Latin. Consider the following Italian and Spanish sentences, respectively.

(39) mi ha morso un cane (literally: me has bitten a dog; `A dog has bitten me' → `I have been bitten by a dog')

(40) le ha mordido un perro a mi madre (literally: her has bitten a dog to my mother: `A dog has bitten my mother' → `My mother has been bitten by a dog')

The presentative word order can be expected with more than average frequency in specific types of texts in which new situations or new entities are introduced which the reader cannot be expected to know. This is the case in the late fourth-century Peregrinatio ad loca sancta, in which examples like (41) abound:

(41) et trans vallem apparebat mons Dei Syna (`And across the valley the mountain of God, Syna, appeared', Peregr. 1.1)

This text is often cited for its high frequency of non-final verbs and as some intermediate stage in the development of SVO order as exemplified in French (see crosssection 9.6. below). However, given the nature of the text, the large number of final Subjects, and hence non-final verbs, need not surprise us. More research in this direction is necessary. [35] Concluding remarks

It cannot be ruled out that a number of the generalizations found in the grammars (see the summary at the beginning of this section) is correct for (main) non-complex sentences. But these generalisations are not generally applicable, in that in the texts, at least in the more literary Classical texts, complex sentences predominate, while, moreover, many sentences do not have an explicit Subject constituent. Placement of a constituent in the first position of the sentence is, however, determined rather by a pragmatic factor (Topic) than by a syntactic factor (Subject). It is useful to distinguish at least two kinds of Focus: `newness' and `contrast'. [36] Furthermore, we have seen that two types of sentence are to be distinguished: sentences with a Topic – Focus distribution and all – new sentences.

9.4. The order of constituents on the noun phrase level

In this section I limit myself to noun phrases (for preposition phrases see crosssection 5.3.). The grammars agree on a number of generalizations with regard to the

-- 185 --

`normal' position of certain categories of Attribute relative to a Head (see K.–St. II.605–11; Sz. 406–9). These generalizations are illustrated in table 9.6. Table 9.6 Position of Attributesa
Position of Attributes with regard to Heads
Type of AttributeBeforeAfter
Demonstrative pronoun+
Possessive pronoun+
`Determining' adjective+
`Qualifying' adjective+
Complex Attributeb+
Noun in the genitive+

a There is a difference of opinion as to the position of the Attribute (casu quo adjective) in Indo - European. K.-St. (II.605) concluded that according to the Indo-European rule the Attribute was traditionally placed before its noun (but see their note 2). Sz., on the other hand, remarks (406) that in Indo - European the adjective seems usually to have been placed after the noun. Rubio (1976: 20–1) assumes an order `determinans – determinatum', without the subdivisions given in table 9.6.

b See p. 83.

Furthermore, the grammars agree that Attributes with monosyllabic Heads are predominantly placed after the Head. [37] Deviation from the normal word order is possible if the Attribute has Focus function. Examples from the grammars are:

(42) altercatio … cui pro tuo studio non est alienum te interesse (`A discussion, in which you, in view of your interests, might very well participate', Cic. N.D. 1.15)

(43) nec me angoribus dedidi … nec rursum indignis homine docto voluptatibus (`I did not let myself go in sadness, or in pleasures that would be unworthy of a learned man', Cic. Off. 2.2)

One of the many problems in table 9.6 is formed by the notions `determining' [38] and `qualifying'. [39] These terms suggest that some adjectives belong to the former class and other adjectives to the latter, and that this determines the `normal' placement of those adjectives. In view of the fact, however, that all adjectives may occur both before and after a noun, it might be preferable to say that in Latin adjectives follow the Head unless pragmatic factors such as Focus cause them to be preposed. Some adjectives are – on account of their lexical meaning – more suitable candidates for Focus function than others, e.g. bonus (`good') more so than Romanus (`Roman'). Further research is required on this point. [40] Placement before or after the Head is one of the means by which an Attribute can be marked as having Focus function. Another means is discontinuity. Two examples are:

-- 186 --

(44) dedi … symbolum servo tuo … epistulam … :: meo tu epistulam dedisti servo? (`I gave a token to your slave, a letter. :: To my slave you gave a letter?', Pl. Ps. 1200–3)

(45) pro di immortales! tantamne unius hominis incredibilis ac divina virtus tam brevi tempore lucem adferre rei publicae potuit? (`By the immortal gods! Is it possible that one man, with his incredible and divine virtue, has brought so much good to this country in so little time?', Cic. Man. 33)

9.5. Stylistic factors

In poetry, but at times also other text types, the word order is largely determined by metrical and/or aesthetic factors. This is a literary convention, which, as it were, overrides syntactic and pragmatic factors that normally determine the word order.

In the grammars a tendency may be discerned to put word order patterns that are exceptional, but can as such be explained, on a par with intentionally exceptional but hardly grammatical word order patterns. Thus, (46) is adduced as an instance of a complicated and statistically remarkable, but nevertheless grammatically correct, embedding: [41]

(46) si quis qui quid agam forte requirat erit (`If there is someone who happens to ask what I am doing', Ov. Tr. 1.1.18)

Conversely, there are examples that constitute an intentional violation of normal placement rules, e.g. (47)–(50):

(47) grandia per multos tenuantur flumina rivos (`Great streams are channelled into many brooks', Ov. Rem. 445)

(48) quem tibi candidi primo restituent vere Favonii … beatum (`Whom the radiant Zephyrs will give back to you happy … at the beginning of the spring', Hor. C. 3.7.1–3) [42]

(49) valui poenam fortis in ipse meam (`To my own detriment I have been brave', Ov. Am. 1.7.26)

(50) raucae, tua cura, palumbes (`The cooing wood-pigeons, your pets', Verg. Ecl. 1.57) [43]

In (47) we find an intentional distribution of adjectives and nouns according to the pattern adj 1 – adj 2 – noun 1 – noun 2. [44] In (49) poenam is in an exceptional position, having been moved from the preposition phrase to the beginning of the sentence. In (50) we find an unusual placement of an appositive constituent. For examples see Sz. (689–94).

Examples of this kind have often been adduced in order to show that Latin word order – presumably owing to the presence of cases – was completely free.

This is incorrect. [45]

-- 187 --

As for prose texts, prosodic considerations have received attention from antiquity onwards (Quint. 9.4.26 is an interesting example). There are certainly intriguing observations to be made. Livy, for example, has fuerunt much more often in clausula position than erant, which is in accordance with his preference for polysyllabic final words. [46] 46a

9.6. Typological and diachronic factors

The publication of Greenberg (1963) generated interest in so-called typological phenomena. On the basis of the languages he had examined (a sample of 30 languages) Greenberg noted, for example, the following `regularities': [47]

Universal 2: In languages with prepositions, the genitive almost always follows the governing noun, while in languages with postpositions it almost always precedes.

Universal 4: With overwhelmingly greater than chance frequency, languages with normal SOV order are postpositional.

Universal 5: If a language has dominant SOV order and the genitive follows the governing noun, then the adjective likewise follows the noun.

(N.B. SOV = Subject – Object/Complement – finite verb.) For this kind of regular correlation he introduced the term `Universal'. In modern linguistics a great deal of attention is devoted to this approach, which has, of course, also been criticized and amended, as regards both the actual material and its methodological aspects. [48] Also for Latin and later Romance developments research has been done. From a typological point of view, in Classical Latin the following word order rules seem to apply (but see the reservations made above).

(a) SOV

(b) prep(osition) – N(oun) (exceptions: mecum (`with me'); Tauro tenus (`as far as the Taurus')

(c) N – Adj(ective) / Adj – N (see crosssection 9.4. on p. 185)

(d) N – N gen(itive)

One would not expect the combination of (a) and (b) on the basis of Greenberg's Universal 4. The uncertainty of (c) conflicts with Universal 5. The combination of (b) and (d) is in accordance with Universal 2. Furthermore, it should be noted – according to typologists – that the Romance languages are `prepositional' and have SVO word order. [49] Some of the conflicts could be solved by assuming for (Classical) Latin an SVO word order. On the basis of these developments and the facts mentioned in (a) and (d), some scholars assume that Latin word order was `unstable', showing elements from various systems. Others assume that as early as Plautus the normal word order in

-- 188 --

colloquial Latin was SVO. [50] The SOV word order found in the more formulaic inscriptions and in the so-called Classical prose – with Caesar as its most extreme representative – is viewed as a conservative, stylistic order without any relation to spoken Latin. In view of the data presented above this opinion does not seem very convincing. SOV does predominate in the simple sentences from Cicero examined here, but there are deviations. In complex sentences the word order is by no means unequivocally SOV. In Plautus, on the other hand, the word order is not unequivocally SVO. The most striking element is the strong variation one observes between individual authors and text types (a variation which persists until far into the Middle Ages) which makes any statement of the type `Latin was an X-language' unfounded. Further research is necessary. [51] 51a

Bibliographical information

The best survey of the `state of the art' in Latin word order research until the early 1950s is Marouzeau (1953). Panhuis (1982) analyses word order on the sentence level within the framework of the so-called Prague School. For the position of Subject constituents see de Jong (1988). Word order is an important field in modern linguistics, especially from a typological perspective. Recent surveys of the literature are offered by Molinelli (1986) and Wanner (1987). General studies are Comrie (1981a: chapters 4 and 10), Dik (1989), Hawkins (1979; 1980; 1984), Mallinson & Blake (1981: chapters 3 and 6). For Latin typological research has been done by Adams (1976a), Harris (1978), Lehmann (1979) and Renzi (1984).

-- 189 --

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