In many respects Latin poetry deviates from the linguistic norms of prose; such deviations are often attributed to non-linguistic factors, with the result that the language of Latin poetry is more often studied by literary scholars than by linguists. A frequent poetic phenomenon are discontinuous adjective-noun phrases, which tend to be explained by appeal to various artistic considerations. The present contribution argues that syntactic discontinuity in poetry is not arbitrary, but is subject to linguistic constraints, which are largely the same as in prose. By scrutinising 120 non-discontinuous adjective-noun phrases in the Ciris (out of a total of 531 adjective-noun phrases), it is demonstrated that the overwhelming majority of such phrases fall into a number of specific syntactic and semantic categories, while outside these categories discontinuity appears to be the rule.
One of the most salient features of the language of classical Latin poetry is the high frequency of adjectives, which, moreover, are often discontinuous from their head nouns. Since in prose discontinuity is considered to be a special case usually motivated by pragmatic factors, most studies of adjective-noun phrases in Latin poetry have likewise focused on discontinuous instances. Two important observations have been made: discontinuous phrases serve to extend and/or demarcate intonation units (cola); this function appears to be an inherited feature, paralleled both in early-Latin saturnian verse and in other old Indo-European poetic traditions. Linguists, however, tend to disregard the use of discontinuous phrases in Latin poetry, considering them a largely artificial phenomenon: “In poetry discontinuity is mainly a poetic device, without a pragmatic justification” (Pinkster 2021: 1100).
A notable omission of previous studies of adjective-noun phrases in Latin poetry is that they all but ignore seemingly ‘normal’ cases in which adjectives dutifully remain by the side of their nouns. This is the more surprising as it seems to have been generally recognised that, statistically speaking, discontinuous phrases constitute the norm in poetry. It could be profitable to turn the tables, so to speak, and view non-split adjective-noun phrases as the exception, while enquiring into factors that prevent discontinuity (or at least allow non-discontinuity). This is what I attempt to do here, if in a limited and provisional way.
My corpus is taken from the Ciris, an anonymous narrative poem from the Appendix Vergiliana, some 541 hexameters in length. In compiling a collection of adjective-noun phrases, I adopted a generous definition of ‘adjective’ that includes participles (both perfect and present) and modifiers like aliquis, alius, multus, nullus, omnis, solus, talis, tantus, totus, ullus, uterque (but excludes demonstrative and possessive pronouns). As phrase heads, I only accepted substantives (to the exclusion of pronouns). I disregarded instances in which the adjective does not appear to form a syntactic phrase with its noun but clearly plays a predicative role. I collected 531 adjective-noun phrases that fit my criteria, though in view of the poem’s unreliable manuscript tradition, as well as of a certain arbitrariness of my notion of what counts as ‘adjective’, the figure should be taken cum grano salis. Of these, I consider 120 (∼23%) to be non-discontinuous, in which I include cases of ‘minor discontinuity’ produced by either a function word such as preposition or conjunction (e.g. 401 fluctibus in mediis) or a pronominal modifier attached to the same head (e.g. 133 malus ille puer). In other words, discontinuous phrases take up over three-quarters of all adjective-noun phrases in the Ciris. In what follows, I discuss all non-discontinuous instances one by one, in an attempt to show that the overwhelming majority belong to a number of more or less clearly defined categories.
Before I turn to consider individual examples, a few words are in order on the limitations of the present study. First, as transpires from the above paragraph, it is only based on one relatively short text. To obtain more reliable and more objective results, it will be necessary to investigate a larger and more varied corpus of texts: on the one hand, given the restricted number of contexts in the present selection, certain phenomena could only be identified in a tentative way; on the other, it is quite likely that in some poetic texts adjective-noun phrases do not behave in exactly the same manner as in the Ciris. Second, a comprehensive analysis of adjective-noun phrases in Latin poetry should explore discontinuous instances alongside non-discontinuous ones. The purpose of the present study is merely to suggest that non-discontinuous phrases may yield relevant insights, without claiming that they can answer all questions. Third, I have mostly confined myself to investigating linguistic factors that favour non-discontinuity, without considering stylistic or metrical factors in any systematic way. This was done not to deny the potential importance of the latter, but to demonstrate that genuinely linguistic mechanisms may be operative in poetry to a greater extent than is sometimes assumed. Fourth, although I have made an effort to take into account all cases of non-discontinuous adjective-noun phrases in the Ciris, I largely abstain from giving exact figures or percentages for individual categories, as the boundaries between different groups are far from watertight: some examples may belong to more than one category, while in other cases attribution to a particular category may be tentative. Besides, numbers only become meaningful when comparisons are drawn between different texts. In view of these limitations, the present study and its findings necessarily have a preliminary character; I hope to be able in the future to undertake a larger-scale exploration of adjective-noun phrases in Latin poetry.
The first group includes cases in which the adjective-noun phrase is enclosed on both sides by (a) a clause boundary, (b) a boundary between conjoined constituents, or (c) a verse break; of these, the first two are absolutely impenetrable, whereas the third can on occasion be breached, but as a rule appears sufficiently strong to justify non-discontinuity. It seems worth noting that the majority of such cases are supplied by conjoined constituents. In examples (1)–(9), the adjective-noun phrase is enclosed by a clause boundary on one side, and by its coordinate pair on the other:
Sometimes the two conjoined constituents are connected asyndetically:
Sometimes the adjective-noun phrase is demarcated by a connector on both sides:
In quite a few cases one of the boundaries is produced by a line break (–), though, as (27) below shows, an adjective-noun phrase can in principle be split between lines; it may also be worth noting that several examples (–) involve local satellites (cf. below):
Significantly, only in a minority of cases the verse-internal syntactic boundary is not with a coordinate constituent:
Finally, in a number of cases syntactic and metrical boundaries may leave a potential for discontinuity, but there are clear reasons why it is not realised:
In (24) and (25), the adverbial modifiers longe and suaue almost form a unity with the modified adjectives, not unlike tam in (9) or (28). In (26), the relative quae normally takes first position, and although poetry allows postponement, here its placement is additionally supported by anaphora. A similar explanation should suffice to account for (27), cuius being quite common at the head of a clause beginning at the penthemimeral caesura. In (28), 248 te stands at the head of the clause (after the subordinator) for the sake of contrast with 247 me. In (29), the adverb inde plays a discourse-organisational role, which also favours clause-initial position. In (30), the apposition could in principle split the adjective-noun phrase so as to produce the so-called schema Cornelianum, but even in poetry this word-order pattern is rare. In (31), the anaphoric construction apparently requires the verb to come at the end of the first parallel member. Similarly in (32), the resumed diues at 434 appears to be under a constraint to occupy line-initial position.
One could ask further relevant questions about the examples adduced above, but for our present concerns it will suffice to note that over a quarter of all non-discontinuous adjective-noun phrases avoid separation not for any positive reason, but because no mobile words are (fully) available in their syntactic-prosodic units to split them.
In a study of factors that constrain discontinuity in Latin prose, Bolkestein makes the important observation that the information status of the noun phrase in the sentence plays a crucial role. In syntactic terms, “discontinuous NPs are rarely other than ‘arguments’ of the main predicate, that is they are major constituents of the verb frame (Subject, Object, Recipient etc.)”. In particular, “no optional adnominal genitive phrases are found which are themselves internally discontinuous; very rare are obligatory genitive phrases of which the Head is the intervening element. Rather rare are discontinuous satellites as compared to NPs in other semantic and syntactic functions.” As Bolkestein plausibly suggests, an explanation may be found “in limits to the general cognitive capacities of language users in production and processing. The restrictions on discontinuous satellites and adnominal genitives […] may well be connected to differences in what is more or less easy to keep track of as a unit still to be completed.” In other words, discontinuity requires extra attention – which can fittingly be given to core elements of the message, much less so to its periphery. The same factors appear equally to be operative in poetry: as we shall see, a major part of our collection of non-discontinuous phrases consists of satellites, genitive phrases and – a category not identified by Bolkestein – appositions.
A sizeable portion of non-discontinuous satellites have either a temporal () or a local (–, minus ) function:
Possibly (44) should rather be interpreted as an argument, but it conveys the same local meaning as other satellites and is formally similar to (43) and (46). Another distinct group, which also includes (45) above, are satellites with some sort of modal or instrumental function:
It may be worth noting that the majority of examples in this group follow on the penthemimeral caesura and often are of the shape ‘molossus + dactyl’, as is (33) above. Both formal and semantic similarities between individual examples, besides their sheer number, seem a strong indication that satellite status constitutes a valid constraint on discontinuity, though in poetry this may not so much mean ‘preventing discontinuity’ as ‘allowing non-discontinuity’.
The second syntactic class with constraints on discontinuity – adnominal genitive phrases – was identified by Bolkestein with regard to a rather specific type (“a complex genitive phrase which designates a state of affair”), so it is not clear if the restriction should hold valid for other types of genitive phrases as well. Yet if Bolkestein’s general explanation for such constraints is correct, any type of adnominal genitive is at some distance from the syntactic core of the sentence and, accordingly, should succumb to discontinuity less readily than main constituents. Without complete certainty, I assume that the extrapolation is valid, at least to some extent, though below I also suggest other syntactic or semantic factors that may have contributed to the avoidance of discontinuity in individual instances:
In (61), we may be witnessing the specific constraint noted by Bolkestein against the head splitting the dependent genitive phrase (*patulae species frontis would be metrically possible). In (62), it is no doubt a contributing factor that the genitive phrase belongs to an extended apposition to Corinthon, which appears also to account for the non-discontinuity of (75) florentia regna (see below). The syntactic interpretation of (63) is less transparent, but the filia phrase can probably be taken as an apposition to illa (‘I am she, the once daughter of mighty Nisus’), cf. (76) below; in (64), the similar filia phrase is attached to Carme. On the semantic level, one might speculate that in (62)–(65) the adjectives are not merely ornamental, but almost form part of the name: ‘Cypselides the Great’ (), ‘Nisus the Mighty’ (), ‘Phoenix the Ogygian’ (), ‘Myrrha of Arabia’ (). In (66), the adjective can probably qualify as distinctive (as opposed to descriptive): tortoises are not as a rule ‘ferocious’, only Sciron’s was (for distinctive adjectives as favouring non-discontinuity, see below). In (67)–(69), (71)–(72), the adjectives can be characterised as predicative or situational (again, see below): they refer to a temporary (or otherwise restricted) state rather than to an essential permanent quality of their head nouns (note that three out of five are participles). In sum, even though many of the examples can be accounted for by other factors, this need not imply that genitive phrases are not as such under a constraint against discontinuity.
The third syntactic category that appears to limit the potential for discontinuity are appositions. Bolkestein does not discuss it, but appositions are precisely the kind of element that is at the periphery of the syntactic structure of the sentence. Our examples suggest that this is indeed a valid category:
In the first passage the appositive construction accounts for two non-discontinuous phrases (–), and arguably the same holds true for the second: besides (75), see (62) above. In (76), the syntax is somewhat tricky; I suggest cognato sanguine uobis (ablative of quality) can be interpreted appositively: ‘I am her, (who is) of the same blood as you, Scylla’, cf. (63) above. Examples (77) and (78) can, I think, be construed along similar lines. In (81)–(83), appositions are attached to a head in the vocative, which in (83) I take to be unexpressed.
Two of the syntactic categories with constraints on discontinuity discussed above account for a major share of ablative and genitive phrases in our collection; many of the remaining non-discontinuous phrases are either in the nominative or in the accusative. As noted before, main syntactic constituents, especially the subject and the object, are the most likely to be discontinuous in prose; there is no reason why this should be different in poetry. It seems natural to posit, accordingly, that most instances of non-discontinuity in nominative and accusative phrases are due to semantic rather than syntactic constraints.
It had been proposed that in prose the position of the adjective relative to its head noun is influenced by whether it is descriptive or distinctive, even though it has since transpired that rules governing the position of the adjective are far more complex. As alluded to above, however, it appears that in our corpus distinctive adjectives are under a constraint against discontinuity from their head nouns:
Examples (84)–(86) feature typical distinctive adjectives, derived from proper nouns and singling out specific persons out of a class: ‘the hero of Gortyn’ (), ‘the daughter of Nisus’ (), ‘Neptune’s spouse’ (). In (87), the distinction is focal: ‘if your love is lawful’ (as opposed to illicit). Similarly in (88), the adjective carries the informational load of the phrase: ‘strange, previously unheard, talk’. In (89)–(91), the adjective-noun phrase defines a conceptual unity: ‘mortal hearts’ for ‘humans’ (), ‘the following dawn’ for ‘next day, tomorrow’ (), ‘nocturnal silence’ for ‘the quiet of night’ (). The next two examples are kennings: ‘Cyllenian murmur’ is the sound of lyre (), ‘the familiar grass’ is dictamnus (). In (94), one would normally expect such exotic geographic adjectives as Idaeis and Grais to behave as ornamental epithets, not unlike Dictaeas in (93), but since they are used to produce a contrast (‘neither Idaean, nor Greek witches’), they take on properties of distinctive adjectives. Since distinctive adjectives work together with their head nouns to identify the referent, it stands to reason that they should be less readily detachable than merely ornamental descriptive adjectives.
I suspect that a similar constraint may account for the non-discontinuity of adjectives that can be characterised as predicative or situational: if a distinctive adjective serves to isolate the referent out of a larger class of things referred to by the noun on its own, a situational adjective confines its noun’s referent to a specific moment or circumstance. It is significant that all the clear examples in this category are formed by participles:
Example (95) should probably be taken to mean, not ‘her deceased father’, but ‘her father after his death’; in (96), the adjective does not merely serve to intensify the noun’s meaning (‘looming fear’), but conveys a notion of manner (‘fear by its looming over’); it may also be significant that both adjectives follow their nouns, a position untypical for descriptive adjectives in poetry. In the rest a predicative interpretation is quite natural: Scylla (daughter of Nisus) observes the stars as they slide down (nutantia) towards the western horizon (); Carme swears by the godhead of Dictynna, present (praesentia) to witness the oath (); Scylla raises to heaven her eyes, glowing (ardentia) in indignation (); she asks the winds to calm down their raging (turbantia) gusts (); Carme kisses Scylla’s cheeks, wet (rorantibus) with tears (); Scylla (the monster) preys on sailors caught unawares (deprensos) by a storm ().
One last group of non-discontinuous phrases that can be isolated on semantic grounds appear to be phrases involving the quantifiers talis (–) and tantus (–):
Although (103)–(104) could be accounted for as satellites, as in fact was (57) above, the remaining examples do not easily fall into any of the defined categories. One might speculate that the deictic force of talis and tantus is responsible for their non-discontinuity, possibly by making them behave as distinctive adjectives (or perhaps simply as demonstrative pronouns). Whether or not this explanation is correct, it seems likely that talis and tantus are indeed under a constraint not to be separated from their head nouns.
There remain 11 examples of a non-discontinuous adjective-noun phrase, which require individual discussions: in some instances it proves possible to account for their non-discontinuity, but in a handful of cases no linguistic explanation is forthcoming. First, an example in which the adjacent adjective and noun do not produce a syntactic phrase:
It seems clear that perfusa should be taken as a secondary predicate, besides exornarat, both of which are negated by necdum: ‘the priestess had not yet bathed in sacred water, nor wreathed her head with leaves of the pale olive’.
A noteworthy special case appears to be presented by two occurrences of the same phrase, infelix uirgo:
We already have encountered it in example (19), where non-discontinuity was attributed to the constraint of space – which clearly does not apply in either (111) or (112). If we consider (19) in its wider context, however, another factor transpires: the phrase appears to be used as apposition to the (unexpressed) subject of the main verb (siue etiam iactis speciem est mutata uenenis, | infelix uirgo, ‘or else whether she was disfigured by scattered poison, unlucky girl’, 70–71). In all four remaining occurrences of the adjective in the Ciris, its use can likewise be described as, broadly understood, appositive, conveying the speaker’s attitude to Scylla and her situation. It appears that in such contexts the force of infelix is evaluative rather than objectively descriptive. The same holds true, it would seem, for (111) and (112). In both cases, the subject does not need to be expressed in the main clause, since Scylla is likewise the subject in the preceding subordinate clause; accordingly, infelix uirgo should perhaps be construed appositively in (111) and (112) as well (‘unlucky girl, she…’).
For the remainder of examples, I attempt to come up with some tentative ad hoc explanations, or admit failure:
In (113), one might expect the colour adjective candida to behave as a typical ornamental epithet; however, it clearly is contrasted with roseus in the next line, so it may rather serve as a distinctive adjective, defining a conceptual unity (‘grey hair’) jointly with its head noun; cf. (94) above. One might ask why, then, roseus is disjoined from crinis; a possible answer would be that roseus crinis is more focal than candida caesaries (as is medio uertice, a local satellite, in comparison with tempora circum). The sequence (114)–(116) is striking and possibly to be explained by appeal to artistic reasons (a description of Scylla’s distress requiring less balanced word-patterning?). One might explain (114) as an instance of modal/instrumental satellite; however, in most other examples discussed above (–), the adjective contributes to the modal meaning of the phrase (note that many are participles), whereas Idaeo here is purely decorative. In (115), there seems to be no linguistic reason for keeping bacata monilia together. The non-discontinuity of (116) can probably be attributed to its being a local satellite. In (117), I tentatively suggest that non-discontinuity is intended to strengthen the pause at the penthemimeral caesura, thus effecting a rhythmic conclusion of the period describing Scylla’s metamorphosis (496–507). I see no explanation for (118) and (119). Finally, (120) is likewise inexplicable, but it also belongs to a corrupt passage whose restoration is far from certain.
To sum up, I hope to have shown – if in a very provisional and approximate way – that the normal linguistic factors influencing (non-)discontinuity of adjective-noun phrases in Latin prose are also operative in poetry. It has transpired that the overwhelming majority of non-discontinuous phrases in the Ciris can be attributed to specific syntactic or semantic constraints, while such phrases not under any constraint – more specifically, main constituents consisting of a noun accompanied by a descriptive adjective – are vanishingly few. The rule governing discontinuity in Latin poetry can perhaps be provisionally formulated thus: adjective-noun phrases are discontinuous, unless there is a specific factor allowing non-discontinuity. This hypothesis will of course need to be tested on a larger corpus of texts, with due attention also paid to discontinuous phrases, but one may expect it to remain true that the behaviour of adjective-noun phrases in Latin poetry is amenable to linguistic analysis, rather than being purely a matter of artistic caprice.
Bauer, Brigitte L. M. 2009. Word order. In Philip Baldi & Pierluigi Cuzzolin (eds.), New perspectives on historical Latin syntax (Trends in Linguistics. Studies and Monographs [TiLSM] 180), vol. 1, 241–316. Berlin: Mouton de Gruyter.Search in Google Scholar
Bolkestein, A. Machtelt. 1998. Word order variation in complex noun phrases in Classical Latin. In Benjamín García Hernández (ed.), Estudios de Lingüística latina: actas del IX Coloquio Internacional de Lingüística Latina, vol. 1, 185–202. Madrid: Ediciones Clásicas.Search in Google Scholar
Bolkestein, A. Machtelt. 2001. Random scrambling? Constraints on discontinuity in Latin noun phrases. In Claude Moussy (ed.), De lingua Latina novae quaestiones: actes du Xe Colloque International de Linguistique Latine, 245–258. Leuven: Peeters.Search in Google Scholar
Conrad, Carl W. 1990. From epic to lyric: A study in the history of traditional word-order in Greek and Latin poetry. New York: Garland Publishing.Search in Google Scholar
Dainotti, Paolo. 2015. Word order and expressiveness in the Aeneid (Untersuchungen zur antiken Literatur und Geschichte 121). Berlin & Boston: Walter de Gruyter.10.1515/9783110401028Search in Google Scholar
Habinek, Thomas N. 1985. Prose cola and poetic word order: Observations on adjectives and nouns in the Aeneid. Helios 12. 51–66.Search in Google Scholar
Heyken, Johann. 1916. Über die Stellung der Epitheta bei den römischen Elegikern. Kiel: H. Fiencke.Search in Google Scholar
Hoffer, Stanley. 2007. The use of adjective interlacing (double hyperbaton) in Latin poetry. Harvard Studies in Classical Philology 103. 299–340.Search in Google Scholar
Jong, J. R. de. 1986. Hyperbaton en informatiestructuur. Lampas 19. 323–331.Search in Google Scholar
Mahoney, Anne. 2014. Colometry in Vergil’s hexameters: Structure, style, and sense. New England Classical Journal 41. 320–331.Search in Google Scholar
Markovic, Daniel. 2006. Hyperbaton in the Greek literary sentence. Greek, Roman, and Byzantine Studies 46. 127–146.Search in Google Scholar
Martín Rodríguez, Antonio María. 2010. Quand le signifiant est aussi significatif: Effets de sens dans l’ordre des mots du syntagme nominal chez Ovide. In Olga Spevak (ed.), Le syntagme nominal en latin: Nouvelles contributions, 97–118. Paris: L’Harmattan.Search in Google Scholar
Neubourg, L. de. 1978. Sprachlicher und metrischer Zwang bei der Stellung von Adjektiv und Substantiv im lateinischen daktylischen Hexameter. Glotta 56. 102–122.Search in Google Scholar
Nisbet, Robin G. M. 1999. The word order of Horace’s Odes. Proceedings of the British Academy 93. 135–154.Search in Google Scholar
Watkins, Calvert. 1995. How to kill a dragon: Aspects of Indo-European poetics. New York: Oxford University Press.Search in Google Scholar
Wills, Jeffrey. 1996. Repetition in Latin poetry: Figures of allusion. Oxford: Oxford University Press.Search in Google Scholar
Wohl, Birgitta. 1974. The phenomenon of Sperrung in Tibullus’ elegies. Transactions and Proceedings of the American Philological Association 104. 385–428. https://doi.org/10.2307/2936099.Search in Google Scholar
© 2022 Boris Kayachev, published by De Gruyter, Berlin/Boston
This work is licensed under the Creative Commons Attribution 4.0 International License.