Logical Categories and the Parts of Speech System as Structuring Devices in Pollux’ Onomasticon

Citation with persistent identifier:

Chronopoulos, Stylianos. “Logical Categories and the Parts of Speech System as Structuring Devices in Pollux’ Onomasticon.” CHS Research Bulletin 5, no. 1 (2016). http://nrs.harvard.edu/urn-3:hlnc.essay:ChronopoulosS.Logical_Categories_and_Parts_of_Speech_as_Structuring_Devices.2016

Pollux’ Onomasticon as onomasiological dictionary

1§1 Pollux’ Onomasticon is a Greek dictionary in 10 books (ca. 120.000 words) from the second century CE.* It is written as fluid text, although it mainly contains lists of words. Its author, Julius Pollux, was a professional orator (or “Sophist”) in Roman-Imperial Athens and professor of rhetoric in Athens.[1] He has a rather clear idea about what the purpose of his dictionary is. In the prologue of the first book he entitles it “Onomasticon” (“book of words”) and defines its content: lists of synonyms, that can be used alternatively; and lists of notions together with the words used to denote them (Poll. 1.2: ὀνομαστικὸν μὲν οὖν τῷ βιβλίῳ τὸ ἐπίγραμμα, μηνύει δὲ ὅσα τε συνώνυμα ὡς ὑπαλλάττειν δύνασθαι, καὶ οἷς ἂν ἕκαστα δηλωθείη).[2] This concept is directly comparable to those used to define a specific type of modern dictionaries, the onomasiological dictionaries or thesauruses.[3] M. Roget, the creator of the prototypical modern onomasiological dictionary, the Thesaurus of English Words and Phrases  (1st edition 1852)[4] defines it as “a linguistically oriented wordbook in which general-language vocabulary is organized from a semantic point of view, designed to give guidance on alternative words for similar concepts” (in Hartmann and Braun 2006:1013).

1§2 Onomasiological dictionaries have two main characteristic features that clearly distinguish them from semasiological dictionaries: the starting point for each entry is a notion/concept, not a lexeme; the main body of the entry consists of groups of words used to express this notion, ordered according to logical and/or linguistic categories. The term “microstructure” is used to refer to the internal structure of each entry. In modern onomasiological dictionaries the microstructure is regularly based on the parts-of-speech system (PoS system), the distinction between noun, adjective, verb and adverb.

1§3 The entries in an onomasiological dictionary are not organized alphabetically (as it is usually the case in semasiological dictionaries) but in object domains according to logical/conceptual criteria. The term “macrostructure” is used to describe this overall order of the entries. Modern onomasiological dictionaries combine scientific and common sense categories and categorizations of the world in order to create hierarchical macrostructures, and thus attribute to “each concept a relative place in the wider conceptual system of which it is part (Alexander and Kay 2015:376).[5]

1§4 Pollux’s Onomasticon is a general onomasiological dictionary in the sense that most of its entries have as starting points notions/concepts and that it covers a broad range of vocabulary.[6] But the Onomasticon has no hierarchically organized macrostructure similar to those of modern onomasiological dictionaries. The connection of the different entries is logical in some cases and associative in others; often there is no connection at all between adjacent entries. Pollux makes this lack of a consequent macrostructure explicit in the prologue of the first book, remarking that he starts presenting the vocabulary about gods, as every pious person should do, but he then orders the other entries as they happen to come into his mind (Poll. 1.2.7–9: ποιήσομαι δὲ τὴν ἀρχὴν ἀφ᾽ ὧν μάλιστα προσήκει τοὺς εὐσεβεῖς, ἀπὸ τῶν θεῶν· τὰ δ᾽ ἄλλα ὡς ἂν ἕκαστον ἐπέλθῃ τάξομεν). König (2016:301 and note 11) shows that this claim is a topos in ancient miscellany of the Imperial period and argues that it is “disingenuous,” since both in miscellanistic texts which raise this claim and in the Onomasticon careful and consecutive reading reveals connective threads between different subjects or entries and “clusters of order scattered among” the texts (König 2016:300). In my interpretation, the formulation τὰ δ᾽ ἄλλα ὡς ἂν ἕκαστον ἐπέλθῃ τάξομεν that Pollux uses to refer to the general structuring principle of the subject domains in the Onomasticon refers exactly to the types of connections König recognizes in the Onomasticon: they are connections that emerge as a certain subject domain or a group of subject domains evokes another one in the lexicographer’s mind. Pollux tacitly contrasts this type of connection, prevailing in the macrostructure, to the much more regular, logical and very often explicitly stated connections among word lists related to the same subject domain, that is, on the level of microstructure.

Scope of the paper

2§1 In this paper, I focus on the level of the microstructure in the Onomasticon, on the structures used to order lists of words in entries.

2§2 Pollux uses two very different devices for this purpose: on the one hand, logical and descriptive categories that isolate specific aspects of the object domain that the entry covers and on the other hand the PoS classification system.[7]

2§3 In the first part of the paper I present briefly these two structuring devices in two examples and I propose an explanation about the reason of this difference.

2§4 In the second part I discuss entries in which a structure based on the PoS system is mixed with a structure based on logical/descriptive categories, and I try to show how the attempt to accommodate for needs of a logical/syntactic classification system influences the PoS system Pollux regularly uses. In this connection, I attempt to explain a particularity in the PoS system Pollux uses, the clear and regular differentiation between ὄνοματα (onomata: nouns and adjectives denoting persons, objects and qualities) and πράγματα (pragmata: abstract nouns, denoting mainly actions and states-of-affairs.)

Word-oriented Structures and the PoS System

3§1 In 6.185 Pollux lists vocabulary related to the concept “τρυφᾶν” (‘live luxuriously, softly’):

ὡσαύτως[8] ταὐτόν ἐστι τρυφᾶν, ἡδυπαθεῖν, θρύπτεσθαι ἀνατεθρύφθαι, ἐκδεδιῃτῆσθαι, χλιδᾶν. τὸ γὰρ θερμερύνεσθαι καὶ κιχλιδιᾶν κωμικὰ μὲν, ἐμοὶ δ’ οὐκ ἀρέσκει. βιαζομένῳ δ’ ἂν εἰς χρῆσιν ἔλθοι καὶ τὸ σκιατροφεῖσθαι.[9] τὰ δὲ πράγματα τρυφή, ἡδυπάθεια, θρύψις, ἐκδιαίτησις, χλιδή, σκιατροφία· οἱ δ’ ἐσκιατροφημένοι σκιατροφίαι καλοῦνται.[10] καὶ μετοχαὶ δ’ ἀπὸ πάντων. ὄνομα δ’ ἀπ’ οὐδενὸς ὅτι μὴ χλιδανός, τάχα καὶ ὁ θρυπτικός. ἐπίρρημα δὲ θρυπτικῶς καὶ ἐκδεδιῃτημένως, φαῦλον δὲ τὸ ἐσκιατροφημένως, καὶ βίαιον τὸ χλιδανῶς.

3§2 The structure of the entry is based on relations of synonymy (as the introductory remark states: ταὐτόν ἐστι), and on the PoS system. Most of the lists are introduced by a tag declaring the part of speech the words of the list belong to:

  • headword and list of verbs, without PoS tag
  • pragmata: abstract nouns
  • metochai: participles; the remark that participles from all above-mentioned verbs can be used
  • onomata: nouns denoting persons, with the remark that there are actually no such nouns, with two exceptions
  • adverbs

3§3 The structure of the entry is a clear example of the usage of the PoS system as organizing principle. Except of the term pragmata, which I discuss in the second part, all the other terms belong to the terminology of the standard eight parts of speech (ὄνομα, ῥῆμα, μετοχή, ἄρθρον, ἀντωνυμία, πρόθεσις, ἐπίρρημα and σύνδεσμος) as it is represented, for example, in the Τέχνη Γραμματική (Ars Grammatica) attributed to Dionysius Thrax.[11] Relevant for Pollux’ purposes are onoma, rhêma, methochê and epirrhêma.

3§4 A comparison of the microstructure of the entry tryphan with the microstructure of a similar entry in Roget’s Thesaurus reveals that Pollux’ and Roget’s microstructures are virtually identical.[12] In both entries a predefined linguistic classification system, the PoS system, independent from the peculiarities of the specific concept domain each time, is used in order to present the vocabulary in a structured way.

Object-oriented Structures and Logical/Descriptive Categories

4§1 The microstructure of the entries in Roget’s Thesaurus and in other modern thesauri are always organized on the basis of the PoS system. This does not apply to the Onomasticon. In several of its entries the word lists are organized according to logical/descriptive categories that address specific features of the subject domain.

4§2 The first half of book 6 (1–112) presents vocabulary related to the symposium. The section has a macrostructure of its own based on principles such as the order of actions conducted during a symposium.[13]

4§3 The first word lists in the section can be considered as a distinct entry containing vocabulary related to the place and the persons taking part in a symposium (6.7–8). The entry consists in word lists denoting the following notions: (1) “the place of the symposium”, (2) “the act of the symposium”, (3) “the persons taking part”, (4) “the act of inviting a symposium”, (5) “the public symposia”.

4§4 Specific features of the object domain covered in the entry, that is, the distinction between private and public symposia, the fact that a symposium is an event taking place in designated rooms of a house, the fact that it is an event involving persons that form groups and the fact that it is an event that somebody initiates by inviting people, become the logical basis for the choice and the organization of the vocabulary presented in the entry. The structuring principle of the vocabulary in the entry is object-oriented. Although the words in each word list belong to the same part of speech,[14] an organization according the PoS system would not have been reasonable, given the material the entry presents. It is significant that the words contained in the lists (2) and (4) do not belong to the same word family, as is regularly the case in entries organized according to the PoS system, since they cover different aspects of the event “symposium”. The labels Pollux uses to mark the content of each word list are a mixture of logical categories (1: χωρίον, 2: πρᾶγμα, 4: ἔργον) and descriptive categories (3: τοὺς συνιόντας, 5: αἱ δημόσιαι ἑστιάσεις). At least in the case of the list (4) Pollux could have used the PoS-tag ῥήματα. Instead he avoids mixing the system of the object domain-oriented categories with the PoS system.

4§5 Why does the Onomasticon use two different structuring devices for the microstructure of the entries? A common property of onomasiological dictionaries in general, is that they combine a language-oriented and an object-oriented approach, “[b]lend facts about the language with facts about the world in which the language is used,” as Kay and Alexander (2015:367) remark. This is why they have some common features with the definitely language-oriented semasiological dictionaries, yet they also share some important similarities with the object-oriented genre of encyclopedias.[15] In modern onomasiological dictionaries facts, knowledge and models about the world, and categories that emerge from them are decisive for the shaping of the macrostructure. Facts about the language, on the other hand, and especially the categories of the PoS system and relationships of synonymy and antonymy are decisive for the microstructure. The Onomasticon has no overall macrostructure. And although the macrostructures of sections that cover extensive object domains (besides the section on the symposium cf. also the section on the ship in 1.82–125[16]) are in general organized logically, in several cases there are no clear-cut boundaries between micro- and macrostructure and an object-oriented approach influences the most basic organization level of the wordlists, the microstructure of the entries.

Mixing Structures

5§1 The dichotomy between object-oriented and word-oriented microstructures is not a strict one. In several entries, these two structuring devices are mixed or combined. How this is done varies considerably from entry to entry.

5§2 I discuss below three examples illustrating the three main modes of such mixing of structuring devices: (1) logical and PoS structures coexisting without being merged, (2) extension of the PoS structure by using logical/descriptive categories and (3) merging of logical and PoS structures.

5§3 Example 1: The entry “dispose/sell merchandise” in Poll. 3.124.1–126.2 starts with a list of verbs meaning “to sell” (without PoS tagging), followed by pragmata, onomata, and adverbs. After these lists a new subsection with different structure is introduced: a special phrase used for the selling of slaves; adjectives for slaves who have been sold several times; verbs for sellers who sell expensive and cheap; words to denote the place where slaves and merchandise is sold (ὁ δὲ τόπος).

5§4 This two-fold structure is based on a clear distinction between the word-oriented structure of the PoS system and some categories created ad hoc to present special vocabulary related to processes belonging to the object domain of the entry.

5§5 Example 2: The entry “buy” complements the entry “dispose/sell merchandise” (Poll. 3.126.3–127). It starts also with a list of verbs (παρὰ δὲ τοῦ πιπράσκοντος ὠνήσασθαι etc., ‘from the person who sells, buy etc.’) followed by a list of participles denoting “the buyer,” which should be used instead of onomata; some onomata are also introduced but they are marked as not approved and their use is therefore not recommended. The pragmata (ὠνή, ἀγορά, ἐμπολή, ἀγόρασις, ἀγορασία) follow. The next list, after the pragmata consists of nouns and substantivized adjectives and is introduced by τὰ δὲ πιπρασκόμενα φορτία (‘merchandise that is sold’). The last “list” isn’t actually a list, it’s only the remark that there are no adverbs for the entry “buy”.

5§6 The introduction of τὰ πιπρασκόμενα φορτία seems to interrupt the PoS structure. Actually, this list is the last part of a larger list containing participles and onomata denoting the subjects, the processes and the objects of the act “to buy”, with clear and marked distinction among these three categories. Of course, the markers used for these categories go beyond the PoS terminology and focus on specific aspects of the object domain. It is thus more accurate to describe the structure of the entry “buy” not as the mixture of two different structuring devices, but rather as the extension of the PoS system with logical/syntactical categories stemming from a model for the buying act.

5§7 Example 3: Part of the long section about gods at the beginning of book 1 is the entry “praying and sacrificing to the gods (1.25–30).” The descriptive definition of the domain is mine. (1) Pollux introduces the entry directly with the first wordlist which is marked by δεῖ δὲ προσιέναι τοὺς θεούς (‘one must approach the gods [being]’) and consists in a list of adjectives and participles denoting the state of purity in which those who approach the gods should be. (2) A rather long list follows which is introduced by προσιέναι θεοῖς and includes verbs and verbal phrases denoting what humans do when they pray and sacrifice. The list is extended in two cases with information about what the term ἁγνὰ θύματα, which Thucydides uses for ἁρώματα as sacrificial offerings (1.26.6–8), means and what anathêmata (1.28.1–2) may include. (3) The third list of the entry is introduced by τὰ μέντοι πράγματα and presents nouns denoting acts and processes that humans do when they pray and sacrifice to the gods. (4) The next list is introduced by the description/definition for the smoke that is produced from the sacrifice (τὸ ἀπὸ τῶν βωμῶν ἀπορρέον πνεῦμα) and contains only two words (1.29.3–4). (5) The last list is introduced by the remark “one should bring sacrificial animals to the altar [which are]” (προσακτέον δέ θύσιμα ἱερεῖα) and presents words denoting the properties that animals brought to sacrifice should have (1.29.4–7). (6) The entry closes with the information about the term used for the roast meat of the sacrificed animals (1.29.8: ἰστέον δ᾽ ὅτι τὰ ἐκ τῶν ἱερείων κρέα θεόθυτα καλεῖται).[17]

5§8 Specific features of the object domain “praying and sacrificing,” the fact that different subjects and objects are involved, the humans and the animals, that there are certain important rules to be observed and the rough temporal order of processes conducted during a sacrifice are the logical basis for the choice and the organization of the vocabulary presented in the entry. It is not difficult to imagine this entry expanding into an encyclopedic article or even a technical treatise on sacrificing. The structuring principle of the vocabulary in the entry is primarily object-oriented. Nevertheless, there is to some degree an overlap between the logical/descriptive categories used in the entry and the categories that the application of the PoS system as organizing principle would produce: the lists (1), (2) and (3) contain respectively onomata applying to the worshippers, rhemata and pragmata denoting cultic acts; list (4) contains onomata applying to sacrificial animals. It is significant that the list of rhemata and the list of pragmata include several words belonging to the same word family; they present vocabulary covering the same aspect of the cultic acts. The result can be described as the merging of an object-oriented and a PoS structure, with the object-oriented structure being dominant.[18]

The term pragma

6§1 The decision to list abstract nouns denoting actions and states of affairs separately under the term pragma is crucial in structures mixing PoS classification and logical/syntactic categories. Pollux uses regularly the terms ῥῆμα (verb), μετοχή (participle), ἐπίρρημα (adverb), ὀνομα (noun and adjective) and πρᾶγμα or ὄνομα τοῦ πράγματος to denote parts of speech. All terms can be understood with reference to the PoS system exposed and analyzed in the Ars Grammatica attributed to Dionysius Thrax.[19] It is however significant that πρᾶγμα is not a part of speech in the Ars, but the one of the two semantic categories covered by onomata, the other being σῶμα: ὄνομά ἐστι μέρος λόγου πτωτικόν, σῶμα ἢ πρᾶγμα σημαῖνον, σῶμα μὲν οἷον λίθος, πρᾶγμα δὲ οἷον παιδεία … (D.T. Ars Gram. §12). Lallot translates πρᾶγμα as “action” and remarks that the term’s extension embraces any abstract noun and primarily nouns denoting process, action, activity, and state (Lallot 1998:127–128). The connection between pragma and verb is established clearly in the philosophical and grammatical tradition prior to Pollux. The term pragma is used to refer to the infinitive in Aristoteles Cat. X 12b;[20] Apollonius Dyscolus (Adv. p. 129.16 Schneider) defines the infinitive as ὄνομα πράγματος, καὶ εἰ ἔστιν εἰπεῖν, αὐτὸ τὸ γενικώτατον ῥῆμα;[21] the Scholia to Ars Gram. (Sch. Cod. C at §12) contrast the term pragma to the term ousia, the latter denoting nouns, the former verbs.[22] The words classified as pragmata in the Onomasticon indicate in general that Pollux used the term to refer to abstract nouns denoting activity, action, or states of affairs.[23] The term is used 136 times in total.[24] In all cases abstract nouns denoting actions or states are classified as pragmata,[25] or, in 1.17, 1.101, 3.67, 9.96, as infinitives.

6§2 Nouns denoting actions or states of affairs are not always listed as pragmata in the Onomasticon. In 1.123, for example, the action denoting nouns ἀναγωγή, ἀνάπειρα, πλοῦς, περίπλους, ἐπίπλους, διέκπλους are tagged as onomata. In 1.230 the nouns τὰ ἄνθη, ἡ ἀνθησις, ἡ βλάστησις and ἡ βλαστή, that is, nouns denoting objects and nouns denoting actions, are listed together as onomata. But in several cases, in the context of different types of microstructures onomata and pragmata are clearly distinguished, as for example in 2.112, where the distinction between the onoma θρασύφωνος and the pragma θρασυφωνία illustrates that Pollux uses the two terms to denote two different classes of words: ὥσπερ καὶ θρασυφωνίαν τὸ μὲν πρᾶγμα εἴποις ἄν […] τὸ δὲ ὄνομα θρασύφωνος βίαιον.

6§3 The regular usage of pragma as a distinct category in structures that otherwise follow the PoS system is a powerful decision towards creating entry structures that are more logical: it allows for the possibility that the onoma-class organizes subjects and objects of actions and states, while the pragma-class organizes the processes, the actions and the states themselves. This distinction is already a step towards an organization according to logical categories; at the same time though it does not disrupt the PoS classification system.

Concluding Remarks

7§1 Pollux uses three different devices to organize the word lists of his Onomasticon on the level of microstructure: the PoS classification system (as modern onomasiological dictionaries also do), logical categories that address specific features of the object domain and a mixture of both devices. A systematic study of these devices and the ways they are combined and used in alternation is crucial in the attempt to understand the way the Onomasticon is structured. The most pertinent problem in this attempt is to define the basic unit for the structural analysis each time, that is, to isolate a relatively coherent section, comparable to the onomasiological entry of a modern thesaurus. Because Pollux mixes and combines different structuring devices and because he merges elements of microstructure and macrostructure – if we apply as criterium the way modern onomasiological dictionaries are structured – it is often very difficult to define the beginning and the end of an entry in the Onomasticon. Following exactly how the different structuring devices are put in use is a method that allows at least a precise and formalized description of the material the Onomasticon presents.


Alexander, M., and C. Kay. 2015. “Diachronic and Synchronic Thesauruses.” In The Oxford Handbook of Lexicography, ed. P. Durkin, 367–380. Oxford.

Alinei, M. 1974. La Struttura del lessico. Bologna.

Avotins, I. 1975. “The Holders of the Chairs of Rhetoric at Athens.” HSCP 79:313–324. 

Bettalli, M. 2007. “I ‘militaria’ in Polluce.” In L’Onomasticon di Giulio Polluce: tra lessicografia e antiquaria, ed. C. Bearzot and F. Landucci, 145–154. Milan.

Desideri, P. 1991. “Lessici e thesauri.” In Atti del I Seminario di Studi sui lessici tecnici greci e latini, 8–10 marzo 1990, Messina, ed. P. Radici Colace and M. Caccamo Caltabiano, 383–394. Messina.

Hartmann R. R. K. 2006. “Thesauruses.” In Encyclopedia of Language & Linguistics Vol. 12, ed. K. Braun, 668–676. Oxford.

Hüllen, W. 2004. A History of Roget’s Thesaurus: Origins, Development and Design. Oxford.

———. 2009. Networks and Knowledge in Roget’s Thesaurus. Oxford.

König, J. 2016. “Re-reading Pollux: Encyclopaedic Structure and Athletic Culture in Onomasticon Book 3.” CQ 66:298–315.

Lallot, J. 1998. La grammaire de Denys le Thrace. Traduite et annotée. Paris.

Lyons, J. 1977. Semantics Vol. 2. Cambridge.

Marello, C. 1990. “The Thesaurus.” In Wörterbücher: Ein internationales Handbuch zur Lexikographie. Vol. 2, ed.  R. Gouws, et al. Berlin.

Matthaios, S. 1999. Untersuchungen zur Grammatik Aristarchs: Texte und Interpretation zur Wortartenlehre. Göttingen.

Matthaios, S. 2013. “Pollux’ Onomastikon im Kontext der attizistischen Lexikographie. Gruppen ‘anonymer Sprecher’ und ihre Stellung in der Sprachgeschichte und Stilistik.” In L’Onomasticon de Pollux: Aspects culturels, rhétoriques et lexicographique., ed. C. Mauduit, 67–140. Lyon.

Ophuijsen, J. M. van. 1993. “The semantics of a syntactician : things meant by verbs according to Apollonius Dyscolus ‘Peri syntaxeos’.” In ANRW, II, 34.1:731–770. Berlin.

Radici Colace, P. 1997. “Opere/contenitore del mondo antico e tardo-antico : temi e strutture della letteratura di raccolta.” GIF 49:3–20.

———. 2000. “Dai testi ai vocabolari tra ricordo e nostalgia.” In Il tardoantico alle soglie del Duemila. Diritto, religione, società. Atti del Quinto Convegno Nazionale dell’Associazone di Studi Tardoantichi, ed. G. Lanata, 267–283. Pisa.

———. 2013. “Polluce nell’‘Onomasticon’.” In L’Onomasticon de Pollux: Aspects culturels, rhétoriques et lexicographiques, ed. C. Mauduit, 25–37. Lyon.

Roget, P. M. 1879. Thesaurus of English Words and Phrases, new edn. enlarged and improved, ed. J. L. Roget. Boston.

Schmidt-Wiegand, R. 2002. “Die onomasiologische Sichtweise auf den Wortschatz.” In Lexicology: An International Handbook on the Nature and Structure of Words and Vocabularies, ed. D. A. Cruse, 738–752. Berlin.

Swiggers, P., and A. Wouters. 1997. “Philosophical aspects of the Thechnê grammatikê of Dionysius Thrax.” In Grammatica e ideologia nella storia della Linguistica, ed. P. Berrettoni and F. Lorenzi, 35–83. Perugia.

Tolkiehn, J. 1933. “Lexikographie.” In RE 12.2:2432–2482. Stuttgart.

Tosi, R. 2007. “Polluce: struttura onomastica e tradizione lessicografica.” In L’ Onomasticon di Giulio Polluce: tra lessicografia e antiquaria, ed. C. Bearzot and F. Landucci, 3–16. Milan.


*     A version of this paper was presented in the CHS Fall 2016 Research Symposium. I would like to thank the conveners and the attendees of the Symposium for the discussion, as well as Robin Olson for editing the paper.

[1]     For the chronology of Pollux’ carrier cf. Avotins 1975 and Matthaios 2013:70–2.

[2]     For a brief presentation of the program of the Onomasticon emphasizing the atticistic aspect cf. Matthaios 2013:79–81. König 2016:299–304 emphasizes the encyclopedic aspect. Cf. also Radici Colace 2013.

[3]     For the tradition of onomasiological dictionaries in Ancient Greece cf. Tolkiehn 1933:2437–2440, 2448–2455 and Tosi 2007.

[4]     For a list of the editions of Roget’s Thesaurus up to 2002 cf. http://www.roget.org/Tabular_History2.htm.

[5]                For a general description of the main features of the thesauruses and the terms “microstructure” and “macrostructure” cf. Marello 1990:1083–1094, Schmidt-Wiegand 2002:738–752 and Hüllen 2004:331–347 (especially for Roget’s Thesaurus).

[6]     For Pollux’ Onomasticon as an onomasiological dictionary (thesaurus) and some aspects of its macrostructure see the remarks in Desideri 1991:390–393.

[7]     In a few cases Pollux organizes wordlists using the principle of word families as for example in Poll. 7.25, where the wordlist includes words belonging to different parts of speech, related to the profession of “butcher”: κρεωπώλια καὶ κρεωπῶλαι καὶ κρεωπολεῖν.

[8]     Other lists of synonyms precede this entry.

[9]     σκιαστρ. in Bethe’s edition; obviously, a typo, instead of σκιατρ. It is remarkable that Pollux includes the verb only hesitantly (βιαζομένῳ δ᾽ ἂν εἰς χρῆσιν ἔλθοι καὶ τὸ etc., ‘if someone wants to force its usage, he could also use the etc.’), although the verb is attested in Xenophon and Plato. For the usage of βιάζομαι as terminus technicus in the Onomasticon in the meaning “force the usage of a word (in a certain meaning)” cf. Poll. 1.51, 1.115, 3.144, 4.172, 5.107, 5.125, 7.33, 9.92, 9.156.

[10]  This remark interrupts the order: ἐσκιατροφημένοι is a participle, attested otherwise only in Max. Tyr. 24.7 (2nd century CE), (οἱ) σκιατροφίαι is a noun, attested only through Pollux here and in 4.147. The remark is a comment on the following statement that no onomata are produced from the listed verbs except for χλιδανός and θρυπτικός.

[11]  D. Thr. Ars Gram. §11 (GrGr I, p. 2-32): Λέξις ἐστὶ μέρος ἐλάχιστον τοῦ κατὰ σύνθεσιν λόγου. Λόγος δέ ἐστι πεζῆς λέξεως σύνθεσις διάνοιαν αὐτοτελῆ δηλοῦσα. Τοῦ δὲ λόγου μέρη ἐστὶν ὀκτώ· ὄνομα, ῥῆμα, μετοχή, ἄρθρον, ἀντωνυμία, πρόθεσις, ἐπίρρημα, σύνδεσμος. ἡ γὰρ προσηγορία τῷ ὀνόματι ὑποβέβληται. The date of the Ars Gram.  is debated, dates starting from the second century BCE (Uhlig) and going up to the fourth century CE (Di Benedetto) have been proposed; cf. the discussion in Matthaios 1999:17–24 and Lallot 2004:20–25. Independently from the date of the Ars Grammatica, the system of eight parts of speech is generally accepted and presupposed by grammarians in Pollux’ time. Apollonius Dyscolus (first half of the second century CE), for example, presupposes such a system; cf. Matthaios 1999:191–192.

[12]  Cf. Roget 1879:entry #638, “waste”. For the microstructure of the entries in Roget’s Thesaurus cf. Hüllen 2009:135–137.

[13]  The beginning and the end of this part are marked: Poll. 6.1: ἐπεὶ δ᾽ οὐδὲ τῶν συμποτικῶν ὀνομάτων ἀμελητέον, χρὴ λέγειν […] and Poll. 6.112.4: τοῖς δὲ συμποτικοῖς οὐδὲν ἂν κωλύοι προσκεῖσθαί τινα τῶν σποράδην συνωνύμων. The formulation introduces rather short entries from various object domains. It is characteristic that the last wordlist contains vocabulary for the notion παῦσαι τὴν ἑστίασιν, διαλῦσαι τὸ συμπόσιον etc. (6.112.1–3). For the organization of the section on the symposium cf. Radici-Colace 1997:9–12.

[14]  Lists (1) and (3) contain onomata, list (2) and (5) contain pragmata, list (4) contains verbs.

[15]  Marello 1990:1084 remarks that it is quite common feature for onomasiological dictionaries, to offer “defining glosses, describing the object or concept referred to and ignoring the linguistic behavior [the register to which the word belongs, different meaning of different grammatical forms of the word etc.] of the word used to refer [to it],” thus revealing their “encyclopedic nature.”

[16]  For the part of this domain that is relative to war ships and naval battle cf. Bettalli 2007:146–147.

[17]  For the general structure of the section on the gods see Desideri 1991:391–393.

[18]  Cf. Radici Colace 2000, 279–283, who refers to the model of Alinei 1974 for the logical organization of the lexicon of a language and shows that the PoS structure in the Onomasticon can be mapped on logical categories.

[19]  Two times in book 9 terms denoting a specific kind (εἶδος) of onoma (κτητικὸν σχῆμα τῆς λέξεως in Poll. 9.135 and προσηγορία παρηγμένη κατὰ τὴν κτητικὴν ἰδέαν in 9.151; nouns/adjectives produced to denote possession) are used to refer to adjectives with the ending -ικός. For these terms cf. D.T. Ars Gram. § 26, where κτητικόν is one of the types of derivative onoma: εἴδη παραγώγων … κτητικὸν δέ ἐστι τὸ ὑπὸ τὴν κτῆσιν πεπτωκός, ἐμπεριειλημμένου τοῦ κτήτορος, οἷον † Νηλήϊοι ἵπποι, Ἑκτόρεος χιτών, Πλατωνικὸν βιβλίον and the comments in Lallot 1998:133.

[20]  Cf. Swiggers and Wouters 1997:58 discussing this passage: “The conceptual content of an infinitive (which is semantically equivalent to an abstract noun) is qualified as pragma.”

[21]  For a detailed discussion of the term πρᾶγμα in Apollonius Dyscolus see Ophuijsen 1993:733–739.

[22]  Sch. D. Thr. p. 216.8–10 Uhlig: περὶ δὲ τῆς τάξεως ἄξιον ζητῆσαι, τί δή ποτε τῶν ἁπάντων προέταξε τὸ ὄνομα, τοῦ ῥήματος προγενεστέρου ὄντος τῇ φύσει· ἀεὶ γὰρ τὰ πράγματα τῶν οὐσιῶν προγενέστερά εἰσιν. On the term πρᾶγμα in the Greek grammatical and philosophical tradition in general see the concise discussion in Swiggers 1997:56–58.

[23]  Mapped on the ontological categorization used by Lyons 1977:442–447 the nouns listed as πράγματα correspond to the second-order and the third-order entities, that is, events, processes, and states of affairs in time and abstract entities without spatial and temporal aspects.

[24]  I exclude the usages in 1.33: καὶ τὰ πράγματα τὸ μὲν ἅγιον, καθαρόν … τὸ δ᾽ ἐναντίον ἐναγές, ἐξάγιστον etc., in 2.17: καὶ πρᾶγμα δὲ κορικὸν τὸ παρθενικόν and in 8.154: καὶ τὰ πράγματα θορυβῶδες καὶ ταραχῶδες· ἀμείνω γὰρ ταῦτα τῶν ἀπ᾽ αὐτῶν επίρρημάτων. In all three cases πρᾶγμα is used in the meaning “thing, object” that the listed adjectives modify. In case of 8.154 it must be noticed that just before πρᾶγμα has been used as PoS term: καὶ τὰ πράγματα στάσις, στασιασμός etc. (8.153). Cf. also: 2.90, 5.124, 9.8, 9.10, 9.11, 9.13, 9.52, 9.54, 10.15.

[25]  A borderline case is the usage of the term in Poll. 5.111 to mark nouns denoting weather phenomena in summer and in winter (τὰ δὲ πράγματα τοῦ μὲν δυσχειμέρου χειμών, ψῦχος, ἄνεμος, ὄμβρος etc. τοῦ δὲ δυσθέρου πνῖγος, καῦμα etc.).