Particles in Ancient Greek Discourse
Table of Contents
I.1 General introduction
I.2 From σύνδεσμοι to particulae
I.3 Approaches to particles and discourse markers
I.4 General conclusions
I.5 Particle frequencies in Homer, Pindar, Aeschylus, Sophocles, Euripides, Aristophanes, Herodotus, and Thucydides
I.7 Particle index
I.8 Index locorum
II.2 Discourse acts: The domain of particle analysis
II.3 Moves: Particles at discourse transitions
II.4 Discourse Memory: The negotiation of shared knowledge
II.5 Particles and Anaphoric Reference: A discourse perspective on particles with third-person pronouns
III.2 Varying one’s speech: Discourse patterns
III.3 Reusing others’ words: Resonance
III.4 Speaking in turns: Conversation Analysis
III.5 Reflecting emotional states of mind: Calmness versus agitation
IV.2 Multifunctionality of δέ, τε, and καί
IV.3 Discourse segmentation
IV.4 Tracking voice and stance
IV.5 Analysis of four excerpts
V Online Repository of Particle Studies
I.3 Approaches to particles and discourse markers
§1. I.2 offers an overview of the scholarship on Greek particles up to the beginning of the Renaissance. The first authors in this period who paid considerable attention to the description of Greek words were Budaeus (Budé) in his Commentarii Linguae Graecae from 1529, and Stephanus (Estienne) in his 1572 Thesaurus Linguae Graecae. Devarius, a Greek scholar working in Rome, was the first modern scholar to devote an entire monograph to Greek particles in 1588.  Subsequently, Vigerus (Viger) with his 1627 De Praecipuis Graecae Dictionis Idiotismis Liber (a work not specifically on particles) and Hoogeveen with Doctrina Particularum Linguae Graecae (1769) made further major contributions to this scholarship. In the nineteenth century, the number of publications increased considerably; the main works were the monographs by Hartung 1832-1833 and Bäumlein 1861.  The twentieth century produced eight more particle monographs,  among many other studies.
§2. All these works are still available today and have been valuable to our research.  These studies, together with the ancients’ descriptions of particles (see I.2), may be viewed as forming a diachronic, horizontal axis along which our work can be plotted—the gradually developing body of research on which we continue to build in this monograph.
§3. The current chapter concerns the vertical axis of our research, which is formed by modern linguistic approaches to particles and functionally similar words in different languages. I (AD – see I.1 §25) will first describe the main theoretical issues that are important in this kind of research, and give an overview of several frameworks that are often applied. Subsequent sections will briefly discuss other relevant background from modern linguistic studies, as well as existing applications of the aforementioned approaches to Greek and Latin. The survey is intended, first and foremost, to provide a context for our research by familiarizing the reader with the cross-disciplinary antecedents to our work.  Secondarily, we aim, by providing this methodological overview, to suggest further avenues for application to the ancient languages.
§4. It is useful to be informed about general principles of verbal communication when investigating Greek particles, because ancient Greek, regardless of specific genres, was used for communication just like any language spoken and written today. Specifically, words functionally similar to Greek particles are found in languages around the world, and the research on these kinds of words has grown extraordinarily in the last few decades. Classicists can therefore profit from contemporary approaches to particles and discourse markers, in working to refine understanding of the communicative role of Greek particles. 
§5. For the overview four approaches have been selected that are employed especially frequently in discourse-marker studies or that have seemed particularly productive to us. First, coherence approaches, which cover a variety of pragmatic frameworks, aim to explain how particles and discourse markers help speakers and hearers construct relations in discourse. Second, Conversation Analysis explains how particles and discourse markers work within the regularities and social negotiations of talk-in-interaction. The third approach is Relevance Theory, which focuses on constraints imposed by particles and discourse markers on the hearer’s inferential processes. Fourth and finally, Construction Grammar describes all linguistic knowledge as pairs of form and meaning, and in this way makes it explicit which parts of the co-text are crucial for interpreting a particle or discourse marker. For each approach I will sketch its theoretical substance and provide a sample analysis. The descriptions will inevitably be selective, but are written in each case to capture a sense of the variety, in terms of the assumptions, questions, and methodologies used, that we have encountered. Before describing the four approaches, however, I will outline certain theoretical issues common to all of them.
3.2 Terminology, definition, and classification
§6. The same three related issues recur in many studies of particles and discourse markers across languages: terminology, definition, and classification.  They are inextricably linked, and will therefore be discussed together in this section (terminology in §§7-10, definition and classification in §§11-15). Each of these issues has troubled scholars working in this field, regardless of the selected language.
§7. First, how to call words that, rather than modifying propositional content, mainly have pragmatic functions? Many scholars point to this problem of terminology, noting that many different terms are used for roughly the same group of words in several languages. The term “discourse marker” is the most well-known and frequently used, but many other terms exist.  The four most frequent terms are “discourse marker,”  “pragmatic marker,”  “discourse particle,”  and “pragmatic particle.”  The important terminological distinctions, then, are between “marker” and “particle,” and between “discourse” and “pragmatic.”
§8. Different scholars use “marker” and “particle” in different ways: they may refer to the same class of words, to two different, possibly overlapping classes, or to a class and its subclass.  For example, Jucker and Ziv (eds.) 1998b:2 see “discourse marker” as an umbrella term, comprising “discourse particles” as a subclass. Hölker 1990:81 and Andersen and Fretheim (eds.) 2000:1 also consider “marker” to be a broader term than “(pragmatic) particle.” In contrast, Lenk claims that “discourse markers” are a subgroup of “particles” (1997:1; 1998:1, 37).  Schourup 1999:229 and Fischer 2006b:4, among others, argue that “particle” concerns the form and syntactic behavior of the words, whereas “marker” is a functional term. Particles are small, uninflected words that are only loosely integrated into the sentence structure, or not at all. Fischer adds that the term “particle” implies a contrast with clitics, full words, and bound morphemes, as well as with larger entities, such as phrasal idioms.
§9. As for the difference between “discourse” and “pragmatic,” choosing one or the other often has theoretical and methodological implications.  Scholars who speak of “discourse markers” tend to consider these words as primarily playing a role in coherence,  whereas those who use the term “pragmatic markers” often focus on how they constrain a hearer’s or reader’s inferential processes in utterance interpretation.  Notably, Andersen and Fretheim 2000:2-3 claim that the term “discourse marker” implies that the words in question have textual functions only, and that the methodology is confined to corpus research.  As a result they adopt the term “pragmatic marker,” which they consider more neutral. Lenk 1997:1 writes that studies of pragmatic markers focus more on interactional aspects between participants than studies of discourse markers, which tend to investigate the structural organization of discourse.
§10. It is evident, then, that although the four main terms (discourse vs. pragmatic, and marker vs. particle) refer to functionally comparable words, they are—consciously or unconsciously—associated with different perspectives. For economy, I will henceforth use “discourse marker,” the most common term, as an umbrella term in this chapter, except when referring to specific studies that employ a different term.
§11. How to define discourse markers as a category is another issue. Different scholars often stipulate different criteria, even when they use the same terms.  This issue is closely intertwined with that of delineating the category. Regardless of the criteria chosen, all scholars consider some members of the class prototypical, that is, as satisfying all criteria, and others peripheral, satisfying only some.  It is therefore hard to decide which words should be included and which should not. If one takes a functional approach, there can be no finite and exclusive list of lexical items, since many words can be used both propositionally and as discourse markers. Often intuition seems to be involved, as researchers tend to be native speakers of the language under discussion. One classification strategy is to contrast discourse markers with other categories, such as conjunctions, but this dichotomy has been criticized.  In sum, while scholars disagree about which criteria to use, they tend to agree that the boundaries of the discourse-marker category are fuzzy: it is unclear where exactly the category, however defined, ends.
§12. One commonly-used criterion for discourse markers is the idea that they do not contribute to the propositional or truth-conditional content of their host utterance.  In addition to that characteristic, Hölker 1990:78-80 lists the following criteria to define what he calls “pragmatic particles” in French: they do not have a referential or denotative function; and they are very loosely integrated into the sentence, that is, they are syntactically flexible.  Brinton 1996:33-34 adds the following features as prototypical of what she calls “pragmatic markers” in English:  they are used more frequently in spoken than in written discourse; they have a high frequency in spoken discourse; they are negatively evaluated in written or formal discourse; they tend to be short items, often unstressed; they may form separate tone groups; they are often sentence-initial; they often have no clear grammatical function; they are syntactically optional; they are marginal in terms of word class; they are multifunctional; and they seem to be used more in women’s than in men’s speech. With Brinton, Onodera 2011:620-623 also stresses the criterion that discourse markers occur predominantly in initial position in units of talk. She compares the markers (cross-linguistically) to traffic signs that mark a speed limit at the entrance rather than the end of a street. However, Brinton 1996:33 notes that pragmatic markers actually occur in other positions as well.  In addition to these criteria, several discourse-marker analyses emphasize the importance of prosody: different prosodic realizations of the same lexical item may distinguish between the item’s use as a propositional adverb and its use as a discourse marker, or between different discourse-marker uses. 
§13. Defining criteria are usually based on English discourse markers, and are not necessarily valid for functionally analogous words in other languages. For example, many particles in German and Dutch cannot occupy the first position of an utterance (e.g., in some of their uses, German halt, ja, zwar, and Dutch even, maar, nou).  More fundamentally, the English language and the cultural values attached to it are often mistakenly seen as the general human norm, as e.g. Wierzbicka 2006:11-13 discusses.  However, studies have appeared on discourse markers in many other languages, among them several non-Indo-European ones. 
§14. Mišković-Luković and Dedaić 2010 summarize the definition issue as follows. After noting that there is little consensus about which words exactly should fall within categories such as “discourse particles,” they add:This perspective is close to the one taken by Andersen 2001:41, who considers the most central feature of pragmatic markers their ability to “guide the hearer in utterance interpretation and constrain the identification of the intended explicit and implicit meaning of an utterance.”  Wierzbicka 1986b:524 makes the “pointing” or “guiding” metaphor explicit: some particles (in her terminology) may function as “road signs in conversational exchanges or in discourse structure.” 
What seems uncontroversial, though, is the versatile nature of such linguistic phenomena — morphologically, syntactically, distributionally and functionally, they do not form a class. However, they do form a “class” in another important respect — these linguistic encoders facilitate the process of utterance understanding not as syntactically-integrated constituents of the proposition expressed by an utterance, but as pointers to the ways the basic proposition or message should be taken by the addressee.
Mišković-Luković and Dedaić 2010:2
§15. In general, what connects words that are considered discourse markers is not their form, but their guiding function in utterance processing or in interaction. This general function can be subdivided: for example, Andersen 2001:26 describes the “main aspects of marker meaning” as subjective (referring to the speaker’s attitude toward a proposition), interactional (concerning the hearer’s relation, as perceived by the speaker, to a proposition), and textual (involving coherence relations). Others posit different subcategories.  What all discourse-marker studies share, however, is a focus on functions and meanings that transcend the transfer of referential information: they rather concern aspects such as those mentioned by Andersen. In order to get a grip on this kind of functions and meanings, different approaches are available, to which we will now turn.
3.3 Different approaches in discourse-marker studies
§16. Different approaches to discourse markers illuminate their uses in different ways. This part of the chapter will introduce four: coherence approaches, Conversation Analysis, Relevance Theory, and Construction Grammar.  I will describe the theoretical substance of each approach, give a sample analysis, and discuss the advantages of each.
3.3.1 Coherence approaches
§17. Schiffrin’s 1987 monograph on discourse markers inaugurated a productive line of research in this area. Although her work utilizes several other frameworks as well, she makes it clear that her main focus is on the role of discourse markers in coherence:
The analysis of discourse markers is part of the more general analysis of discourse coherence—how speakers and hearers jointly integrate forms, meanings, and actions to make overall sense out of what is said (…).
Schiffrin (9-10) builds on Halliday and Hasan 1976, who describe how certain linguistic forms work as cohesive devices. These forms indicate how different units of discourse relate to each other, but do not themselves create the relations. On the general concept of discourse coherence she (21-22) follows Gumperz (e.g. 1982), who argues that hearers infer speakers’ intentions through situated interpretation: coherence depends on the speaker’s integration of verbal and nonverbal cues to situate a message, and the hearer’s ability to interpret these cues as a totality.  Schiffrin describes the role of several discourse markers in establishing certain cohesive ties.
§18. According to Schiffrin, discourse markers “bracket units of talk. Sometimes those units are sentences, but sometimes they are prepositions, speech acts, tone units.” (35) She defines discourse markers as verbal devices independent of sentential structure (32), “which provide contextual coordinates for ongoing talk” (41). Discourse markers, in other words, establish connections between discourse units and parts of their context. This functional definition encompasses items from different word classes, such as conjunctions, adverbs, and verb phrases (40).
§19. Schiffrin inspired many scholars to investigate discourse markers as connected to discourse coherence.  For example, Lenk 1998:1 considers the main function of discourse markers to be “to signal structural organization within discourse.” According to Maschler 1998, 2003, and 2009, discourse markers have a “metalingual” function: they communicate about language use, that is, the text, the ongoing interaction, or the speaker’s cognitive processes, rather than about some “extralingual” world.  This overarching metalingual function includes textual, interactional, and cognitive functions. Within a similar research framework, Aijmer, Foolen, and Simon-Vandenbergen 2006:105-110 list the following considerations as crucial for the understanding of what they term “pragmatic markers”: how these words relate to the utterance, to the context, and to the hearer. That is, it is not only the verbal level of communication that is relevant in coherence approaches to discourse markers, but also the nonverbal level. Aijmer et al. assume that every pragmatic marker has a core meaning to which all "pragmatic meanings" can be related. For example, they argue that English well has acceptance as its core meaning.
§20. Though the precise definition and classification of discourse markers vary among the studies, all these scholars consider such words to contribute to the coherence or structure of discourse. Their aim is usually to describe the core meanings or functions of a certain discourse marker, and to identify contextual features (both verbal and nonverbal) that play a role in their interpretation. Methodologies include classifying functions across discourse domains or discourse types, and comparing different discourse markers, both within one language and cross-linguistically. Coherence approaches have been applied to the analysis of both spoken and written discourse.
§21. Let us have a look at a sample analysis: the description of English well by Aijmer 2013. The author builds on various pragmatic approaches, thereby extending the one of Schiffrin: Aijmer considers coherence to be just one of the factors playing a role in the functions of pragmatic markers (her term). In particular, she adopts the recently developed framework of variational pragmatics, an approach that studies language use as influenced by regional as well as social differences.  In her study of well, she investigates the marker in different communicative situations, such as face-to-face conversations, courtroom interviews, and radio discussions. Discourse markers such as well, she argues (148), have specific functions depending on the discourse type. Consider the following example:Aijmer notes that this well is “lengthened and pronounced with a relatively high pitch and is accompanied by laughter.”  The marker “reflects [the speaker’s] ongoing cognitive process,” in a use belonging to unplanned, spoken discourse. Aijmer interprets the marker here as connected to word search, one of the functions connected to coherence: well “signals that the speaker has found the right expression.”
A: One’s about human brain and language, and the other’s about uh this guy called Chomsky who’s uh, well one of the world’s most important human beings if you happen to be interested in linguistics.
Fragment of a telephone conversation from Aijmer 2013:32, transcript simplified 
§22. She explains the instance in (t4) as a use of well that is typical for a broadcast discussion, specifically for the moderators of such a discussion. In this discourse type, moderators employ well for specialized functions such as introducing a controversial issue, achieving a shift of topics, or asking clarification questions. In this case the moderator invites a new speaker to take the floor:The moderator “selects a speaker by using well followed by the name of the nominated speaker” (59). Here the marker “signals both that the speaker is the moderator and that the transition is to a new stage in the broadcast discussion” (59).
A: Well Terence Hawkes as a professor, a university professor, you must disagree totally with (...)
Fragment of a broadcast discussion, from Aijmer 2013:59
§23. Aijmer points out that different discourse types lead to different interpretations of the same marker. Collocations play a role as well in selecting the intended interpretation; examples are “well I guess” for well’s function connected to word search (32), or “well here is” to introduce a new player in a radio sports commentary (69). Aijmer thus stresses that local and global contexts may simultaneously be relevant to the use of discourse markers.
§24. Investigations using a coherence-based approach provide analyses of a wide variety of discourse markers in many discourse types. In this way, they give insight into how different contexts influence the markers’ uses. Scholars who adopt an approach along these lines pay attention to linguistic, paralinguistic, and extralinguistic features, such as the immediate co-text, the prosody or punctuation of an utterance, the speaker or writer’s role in a situation, and her goal in communicating. More generally, coherence approaches show the fundamental relation of linguistic choices to language users within their environments.
3.3.2 Conversation Analysis
§25. Conversation Analysis (CA) looks at the regularities of talk-in-interaction.  It is concerned with the communicative actions performed by turns-at-talk, with the organization of turn-taking, the structuring of turns, the sequencing of actions, and the organization of turn sequences. The main focus of this approach is on the interaction between speaker and hearer.
§26. CA began with Sacks, Schegloff, and Jefferson 1974 on turn-taking practices.  These sociologists show that there is systematicity in conversation, even though it often seems otherwise. They identify certain general tendencies in turn-taking practices, such as the general avoidance of overlap between turns by different speakers (1974:700-701). Following the seminal work of Sacks et al., sociologists as well as linguists developed the theoretical framework that is now known as Conversation Analysis. 
§27. The main assumption within CA is that utterances are seen as vehicles for social action.  Examples of “actions” in CA are complaining, (dis)agreeing, evaluating, inviting, offering, requesting, summoning, and so on.  CA researchers find that for the accomplishment of such actions, the structural position of linguistic material within a turn and within a sequence or “pair” of turns is crucial.  Other contextual features, such as status differences between speaker and hearer, or the location of the interaction, are considered relevant whenever conversation participants show in their utterances or other behavior that they themselves take them into account.
§28. CA analyzes spoken language; written material is rarely used, and constructed examples never. CA scholars take into account not only linguistic, but also paralinguistic features (such as loudness and pauses) and extralinguistic features (such as laughter, and, if known, gestures, gaze, and movements). In their analyses of discourse markers they investigate, for example, for which actions a certain marker can be used, and what position it tends to occupy within the sequence and within the turn. Different positions are those in initiating versus reacting turns, and in turn-initial versus later positions. 
§29. Let us take the work of Heritage on English oh as an example. He distinguishes among the uses of this discourse marker in three different interactional environments. Consider the following example, in which speaker B receives a piece of news from speaker A.Heritage 1984 notes that turn-initial oh often occurs in responses to utterances that inform the hearer of something. Such oh receipts are often combined with assessments of the news, as in (t5). In this environment it signals that the new speaker has undergone a change in her current state of knowledge or awareness.
A: I was just ringing to say I’ll be comin’ down in a moment.
B: Oh good.
Fragment of a telephone conversation, from Heritage 1984:302
§30. However, in other interactional contexts, such as at the beginning of responses to questions (see Heritage 1998), oh works differently:Heritage interprets such oh as signaling that the preceding question was somehow inappropriate. In the interview in (t6), speaker A has already given information from which it can be inferred that he had learned Chinese; the interviewer’s question is treated as problematic, because the answer is considered self-evident.
A: Some of my students translated Eliot into Chinese. I think the very first.
B: Did you learn to speak Chinese?
A: Oh yes. You can’t live in the country without speaking the language.
Fragment of a radio interview, from Heritage 1998:294
§31. In the third type of context, responses to assessments, Heritage finds that turn-initial oh can modify an expression of agreement or disagreement with the previous speaker (see Heritage 2002). In particular, oh may signal that the observation being evaluated had already been independently arrived at by the current speaker. This may suggest her greater expertise on the topic at hand. This is illustrated in “oh it’s a great cat” in (t7).Speaker A agrees with speaker B, yet at the same time “invokes his epistemic authority as a Burmese cat owner” by using turn-initial oh, as Heritage puts it. That is, A claims that he is in a better position than B to evaluate Burmese cats, since he owns one. Heritage finds that this implication of presenting oneself as a better judge than the addressee only arises when oh occurs at the start of a turn after an assessment. 
A: I acquired a Burmese. D’you know what that breed is?
B: Oh yes indeed, uh, we had a neighbour that had a couple of Burmese. They’re nice.
A: Oh it’s a great cat. It’s the only cat I ever saw that chased dogs.
Fragment of a telephone conversation, from Heritage 2002:207
§32. In short, Heritage’s analysis shows that a speaker’s goal in speaking (receiving information, answering, agreeing), and the nature of the previous turn (news, question, assessment) are relevant to our interpretation of oh. This discourse marker, though apparently meaningless and random at first sight, turns out to be used according to clear patterns. The results serve as a reminder that when describing the functions of a discourse marker we should take into account its larger context and not just the individual sentence in which the marker appears. In particular, the action performed by the first turn in a sequence raises expectations about the form of the reacting turn. Finally, not only the utterances themselves, but everything that is happening during an interaction may play a role in utterance interpretation or understanding, e.g. what the speaker wants to accomplish in terms of negotiating the relationship with the hearer. CA sheds light on the interaction as a whole, not only on a specific linguistic item.
3.3.3 Relevance Theory
§33. Relevance Theory (RT) has been developed by Sperber and Wilson 1995 , building especially on the ideas of Grice (e.g. 1961, 1989).  Grice argues that most communication involves the expression and recognition of intentions, and that not only the decoding of linguistic signs, but also inference, is crucial for successful communication. He claims that utterances automatically create expectations that guide the hearer toward the speaker’s intended meaning. In particular, hearers expect speakers to be cooperative, that is, to produce utterances that are of proper length, truthful, relevant, and clear. 
§34. Although Sperber and Wilson criticize several aspects of Grice’s claims, they follow him in that they too assume that utterances create expectations for hearers. More specifically, Sperber and Wilson start from the cognitive assumption that the human mind tends to be geared toward the maximization of relevance, that is, toward achieving as many cognitive effects as possible with a minimum of processing effort. Cognitive effects can consist of producing a new assumption, strengthening an existing assumption, or deleting a previously held assumption. From this “cognitive principle of relevance” Sperber and Wilson develop a “communicative principle of relevance”: “[e]very act of ostensive communication communicates a presumption of its own optimal relevance” (1995 :158). According to Sperber and Wilson, hearers approach all utterances with the presumption of optimal relevance in mind. This principle enables hearers to search for the intended context and the intended interpretation of an utterance.
§35. RT makes a fundamental distinction between conceptual meaning, which refers to something in the world spoken about, and procedural meaning, which restricts the ways in which a hearer can interpret an utterance. Some linguistic items have conceptual meaning, but discourse markers mainly have procedural meaning: they encode certain constraints on the inferential processes that are needed for utterance interpretation. 
§36. Works such as Blakemore 1987, 1992, 2002; Carston and Uchida 1998; Rouchota and Jucker (eds.) 1998; Noh 2000; and Carston 2002 represent a few of the important developments made in RT.  Some scholars contrast RT with theories concerned with discourse and coherence. For example, Blakemore 2002:157 claims that hearers aim to construct representations of the speaker’s thoughts, but not representations of relationships in discourse.  RT also differs from CA in that RT’s primary focus is not on utterances as social actions, but on utterances as expressions of cognitive processes. 
§37. Another difference from CA is that RT researchers make use of constructed examples, usually in their native language, in addition to or instead of corpus data.  These constructed examples are analyzed, checked for acceptability, and compared to slightly different versions. For example, Wilson and Sperber 2012a:158 discuss the following constructed utterances:The authors explain that the sentence “Peter’s not stupid” is either interpreted as evidence for the conclusion “he can find his own way home,” as represented in (t8b); or as a conclusion derived from evidence presented in the second sentence, as made explicit in (t8c). For (t8a) both interpretations are possible. Referring to Blakemore’s account of procedural meaning (e.g. 1987, 2002), Wilson and Sperber argue that discourse connectives (in Blakemore’s terminology) such as so and after all do not encode concepts, but instead constrain the hearer’s inferential processes in different ways.
a. Peter’s not stupid. He can find his own way home.
b. Peter’s not stupid; so he can find his own way home.
c. Peter’s not stupid; after all, he can find his own way home.
Example from Wilson and Sperber 2012a:158 (highlighting added)
§38. An example of a relevance-theoretic discourse-marker study is found in Mišković-Luković 2010 on the Serbian “pragmatic particles” (her term) baš and kao. By using spoken, written, and constructed examples, the author describes both particles as semantic constraints on the explicit content of utterances. The two particles work in opposite directions: baš (“exactly,” “really”) makes an utterance stronger or more precise, whereas kao (“like,” “or something,” “kinda”) makes it weaker or looser. I will discuss two baš examples in order to illustrate Mišković-Luković’s analysis:Mišković-Luković points out that baš signals that the intended meaning of an utterance bears “literal resemblance” (75) to what is said. In RT the term “literal resemblance” is used to refer to a situation in which no conceptual adjustment is needed from the hearer. “Less-than-literal resemblance” is in fact much more common in everyday communication (74-75). In this case, the particle “serves to confirm the identity” between the concept of “consulate” that speaker A refers to in the first utterance, and the more specific consulate suggested in speaker B’s question. That is, Mišković-Luković interprets baš as “a marker of non-loose use” of language (80). She argues that the particle helps the hearer in his “conceptual adjustment” (74-75): when baš is present, the concept that a speaker intends to communicate bears literal resemblance to the concept she puts into words.
A: To je konzulat sakupljao. Ne znam zbog čega.
B: Konzulat u Štutgartu?
A: Da, baš konzulat.
Fragment of spoken conversation, from Mišković-Luković 2010:69 (her translation)
A: The consulate was collecting it [i.e. certain data]. I don’t know what for.
B: The consulate in Stuttgart?
A: Yes, baš the consulate. (“That very consulate.”)
§39. Mišković-Luković also offers one baš example from a newspaper article:At the time of publication of the newspaper, in 2003, the main referent in focus for the readers was not Borislav Milošević, but his brother Slobodan, the ex-president of Yugoslavia. According to Mišković-Luković, “[b]aš before the proper name is used to block this immediately accessible assumption [i.e. that Slobodan Milošević would be referred to] by reaffirming the identity of the referent [i.e. Borislav Milošević]” (76n12). In the processing of spoken or written baš utterances, the hearer or reader can exclude less-than-literal interpretations of the proposition as it is expressed, or of the part over which baš has scope.
Borislav Milošević slovi kao ključna figura u klanu. (…) Navodno je baš Borislav Milošević u Rusiji pohranio milionske sume, što sam osporava.
Newspaper fragment, from Mišković-Luković 2010:75-76n12 (her translation)
Borislav Milošević is considered to be a key figure in the clan. (…) Allegedly, it was baš Borislav Milošević who had deposited millions in Russia, which he himself denies.
§40. RT’s strength lies in its use of a consistent explanation for every context and discourse type. It employs general cognitive principles to explain how different uses of a discourse marker are connected.  The distinction that it makes between procedural meaning and referential (“conceptual”) meaning helps make it clear how discourse markers, which have no referential meaning, can still provide a specific contribution to the communication.
3.3.4 Construction Grammar
§41. The research field of Construction Grammar (CxG) developed as one of the alternatives to generative linguistics.  A generative approach does not describe the use of language, but the knowledge of speakers about grammatical structures in a given language; speakers are said to “generate” sentences from a set of rules.  CxG, by contrast, is usage-based, and assumes that words and other linguistic structures are learned and interpreted in context. Langacker (e.g. 1987, 1990) has been especially influential in developing the framework of Cognitive Grammar, a field of research foundational to CxG.  He argues, for example, that the grammar of a language provides speakers with pairings of phonological and semantic units. All linguistic structures are based on general cognitive processes, according to Langacker. 
§42. CxG began with Fillmore, Kay, and O’Connor 1988 on the meaning of let alone. Based on the explanation of this idiomatic construction, these authors propose (501) that we view not rules, as in generative linguistics, but constructions, as the “proper units of grammar.” CxG has been further developed by, among others, Goldberg 1995, 2006; Croft 2001; Verhagen 2005; Boas (ed.) 2010; and Traugott and Trousdale 2013.  Currently, numerous scholars are active in this field, including researchers of discourse markers. 
§43. Constructions are conventional and symbolic pairings of form and meaning. They are conventional because their use is shared among a group of speakers; and they are symbolic because they are signs, that is, potentially arbitrary associations of form and meaning.  Phonology, morphology, syntax, semantics, and pragmatics all work together in a construction: no one level of grammar is seen as autonomous in this approach.  This holistic view of grammar means, for example, that there is no sharp boundary between grammar and lexicon.  That is, individual words are considered constructions just as much as abstract syntactic structures that do not include specific lexical items, such as a transitive-verb-with-object construction.
§44. According to a CxG approach, words are not normally interpreted or learned in isolation. It is likely, therefore, that speakers and hearers use specific co-textual or contextual features to identify which forms (for speakers) or meanings (for hearers) they need. CxG does not assume specific communicative principles for this, as RT does, but assumes that speakers and hearers simply have knowledge about language use from their experience. The formal features of a construction can be morphemes, words, phrases, intonational patterns, syntactic structures, even whole text types, among others.  For a certain construction involving a discourse marker, the “form” pole would be the lexical item itself, but it possibly also includes information on co-occurring words or structures. A construction’s description should thus specify which co-textual features, if any, are crucial for an interpretation of a word. On the “meaning” pole, not only semantic information, such as the reference to a certain concept, can be represented, but also pragmatic information, such as the expression of a specific stance of the speaker, and discourse-functional information, such as a connective function.
§45. CxG is informed by both monosemy and polysemy. On the one hand, each construction is monosemous, that is, it has only one meaning, though possibly a highly general or a highly complex one, involving semantic, pragmatic, and discourse-functional dimensions at the same time. On the other hand, many constructions are connected in a polysemous network.  If one specific grammatical or lexical form has several meanings, then it participates in several constructions. In such cases the meaning pole of a construction with only that one grammatical or lexical item as form pole will be highly general, and sometimes impossible to pin down. Specific constructions are called “daughter” constructions: they inherit all aspects of form and meaning from their more general “parent” constructions, but also have specific form features that lead to a slightly different interpretation. For example, the English construction kick the bucket is a daughter construction of the more general transitive-verb-with-object construction, but its idiomatic, metaphorical meaning (“to die”) cannot be predicted from this parent construction; therefore it is considered a separate construction, that has to be learned independently from its parent constructions.  By specifying the particular features of form that make up the daughter construction, a constructional approach explains how hearers and readers arrive at a specific interpretation of a multifunctional item (that is, a word or phrase that participates in several constructions). Such a multifunctional word or phrase may form a parent construction for more specific daughter constructions.
§46. Questions concerning discourse markers within a constructional framework concern, for example, the different constructions in which a certain marker participates, and which other features exactly are relevant to the several constructions in which the marker is found. By analyzing how the interpretations of a linguistic expression change across many different contexts, CxG studies aim to map the features in the co-text and context that play a role in those changes. The data used is always drawn from actually spoken or written language; CxG does not use constructed examples.
§47. An example study is Koier 2013 on the Dutch “particle” (her term) ergens.  The author shows that readers use specific parts of the co-text when interpreting this multifunctional particle. Ergens can function as an adverb with a locative meaning (“somewhere,” “anywhere”) or one of several modal interpretations (e.g. “somehow,” “from a certain point of view”). Koier asked a group of native speakers to choose one of eight interpretations of this form for a number of corpus instances.  The results enabled her to identify “triggers,” that is, words in the co-text that are responsible for each specific interpretation (66).
§48. For instance, in (t11), ergens is generally interpreted as “in someone’s feelings or thoughts.”Koier explains (72-74) that the interpretation “in someone’s feelings or thoughts” is triggered by references to the first person, or by subjective forms such as the mental state predicate “want.” Even informants who are presented with only the shorter word string “zou dat ergens wel willen” (“would ergens want that”) tend to choose the same interpretation. In this case, one of the two triggers, namely “want,” remains explicitly present in the version with a restricted co-text.
Ik zou dat ergens wel willen maar ja, we maken keuzes in het leven hè?
Fragment of a transcribed interview, from Koier 2013:72 (her translation)
I would ergens want [to do] that but well, we make choices in life, don’t we?
§49. In contrast, the interpretation of ergens in (t12) changes when the co-text is marginalized. Ergens is interpreted as “about,” “around” when informants read the full co-text, but receives various other interpretations when only a restricted co-text is given.Koier points out (67) that numbers are a required co-textual trigger for the interpretation of ergens as “about,” “around.” Informants do not choose this interpretation when they only have access to “dat is ergens ja pff” (“that is ergens yeah pff”). The restricted version does not provide enough help on choosing an interpretation, and accordingly informants’ answers vary widely (75).
Stenen voorwerpen uit een periode die men het mesolithicum noemt. Dat is ergens ja pff laten we zeggen zesduizend, vijfduizend voor Christus.
Fragment of a transcribed interview, from Koier 2013:75 (her translation)
Stone objects from a period that is called the mesolithicum. That is ergens yeah pff let’s say six-thousand, five-thousand before Christ.
§50. The important point to remember here is that the co-textual triggers of interpretations of ergens are not just co-occurring, but necessary for the interpretations—that is, they are part of the various constructions that ergens can appear in. For example, the specific construction that has the meaning pole “about,” “around” does not only include the lexical item ergens in its form pole, but also “numbers” as a co-textual feature (see Koier 2013:67).
§51. A CxG approach, based on general cognitive principles, may explain how a hearer or reader interprets a multifunctional discourse marker, by specifying which co-textual elements are crucial for each interpretation. Since there is no distinction between grammar and lexicon in CxG, all kinds of linguistic information can be taken into account in the form pole.
3.4 Further relevant studies
§52. Modern studies on discourse markers do not only apply different approaches to language use, but also provide other useful background for investigating ancient Greek particles. Without claiming to be complete, I will in this section discuss four other areas of research that we have profited from.
§53. First, many discourse-marker studies are contrastive, that is, they compare markers across different languages, analyzing their functional similarities and differences. Weydt 1969 is an early example: he focuses mainly on differences in the particle systems of German and French, but he also briefly compares the German system to that of ancient Greek. Fleischman and Yaguello 2004 represents a more recent contrastive discourse-marker study.  These authors compare the uses and development of English like and French genre, two discourse markers that are functionally highly similar, although they developed from different lexical sources. For example, both words can be used as a hedge, as in (t13).Fleischman and Yaguello 2004 interpret this use of genre as related to the speaker’s hesitation, to tentative suggestions, or to politeness. The authors see a functional correspondence to the English discourse marker like (see their translation of the French example). Such comparisons among formally and/or functionally similar words may clarify their multifunctionality. If the compared words have the same origin, such as German doch and Dutch toch analyzed by Foolen 2006, they may still differ in use. If the forms are not etymologically related, as in Takahara’s 1998 study of English and Japanese, their uses may show similarities because of similarities in context and development. 
Je me demandais si tu pourrais genre me donner un coup de main.
Fragment from conversation, from Fleischman and Yaguello 2004:134 (their translation)
I was wondering if you could like give me a hand?
§54. Second, diachronic studies on discourse markers describe their development from words or phrases with propositional content into items with mainly pragmatic functions. These studies cut across several of the approaches mentioned, and sometimes present the diachronic focus as a theoretical perspective in its own right.  Processes that play a role in the development of discourse markers are subjectification, intersubjectification, and grammaticalization.  These terms refer to the developments of meanings that are progressively more subjective (involving speaker reference or speaker perspective), more intersubjective (involving the speaker’s attention to the addressee’s “self”), or more grammatical than the original meanings of a form. For example, Margerie 2010 analyzes kind of/kinda from this perspective.  She shows that this construction originally had only a propositional meaning, as illustrated in (t14), and acquired additional intersubjective meanings, such as in (t15).Margerie explains that in (t14) kind is a noun and carries a propositional meaning. In contrast, in (t15) kinda has a hedging force: the speaker, an undergraduate student, tries to be polite in asking an older student for help in writing a paper. Margerie identifies intermediate examples, in which kind of/kinda has both a propositional and an expressive function. All these different uses, which developed separately, co-exist from a synchronic perspective. Considerations of a certain discourse marker’s historical development are therefore illuminating for synchronic studies such as those represented in this monograph. 
They wanna look at the kind of college you’re coming from.
Fragment of an American English conversation, from Margerie 2010:317
I was just kinda hoping you’d read over and say this has to be changed or you know whatever.
Fragment of an American English conversation, from Margerie 2010:327
§55. Third, several studies analyze discourse markers in historic corpora from a synchronic perspective. These scholars can obviously only use written language as their material, but they nevertheless pay attention to the interaction between an (explicit or implied) speaker and her addressees. Usually the corpus includes dramatic texts, such as Shakespeare, or narratives with direct speech. Consider the following example of Old English hwæt, from Brinton’s 1996 monograph on “pragmatic markers” (her term) in Old and Middle English.  The cited passage, from the poem Juliana by Cynewulf, is spoken by a devil to the martyr Juliana, who had just asked who sent him.Hwæt usually accompanies old (that is, shared) information, or the speaker at least pretends that the information is old. In this use, hwæt is similar to Modern English you know. The instance in (t16) can be considered “insulting”: the information given by the devil is clearly new to Juliana, but he implies that she should have known it already, and conveys irritation at her “slowness in understanding” (188).
Hwæt, mec min fæder on þas fore to þe,/ hellwarena cyning, hider onsende (…)
Poem fragment (Old English, ca. ninth century), from Brinton 1996:188 (her translation)
What, my father, the king of the hell-dwellers, sent me hither on this journey to you (…)
§56. Finally, studies of discourse markers in narrative discourse resemble our research.  Some of the works just mentioned that deal with older corpora inevitably discuss narrative, such as Fludernik 1995 and 2000, and Brinton 1996. Other analyses, however, describe the use of discourse markers in oral narrative, and therefore use modern, spoken corpora. An example is González 2004 on English and Catalan pragmatic markers.  Among other markers, she analyzes so as a “closing segment boundary marker”: with it, a speaker “closes the last part of her story” (98).
§57. To sum up: beyond specific theoretical approaches, several general perspectives on discourse markers may inform our study of Greek particles. Cross-linguistic comparisons highlight differences and similarities across functionally or formally related markers in different languages. Diachronic studies illuminate how pragmatic meanings develop out of the lexical ones. Furthermore, synchronic research on discourse markers in historic corpora are close to our work. Whenever a narrative corpus is analyzed, finally, this may help us observe functions typical of such context. In fact many of the insights offered by these studies have already been applied to Greek and Latin. It is to an overview of this scholarship that I now turn.
3.5 Studies on particles and discourse markers in ancient Greek and Latin
§58. This section summarizes some recent analyses of ancient Greek and Latin particles and discourse markers that explicitly adopt an approach similar to the ones discussed above.  Let us first step back, however, in order to look at the general issues of terminology, definition, and classification discussed in §§6-15 from a Hellenist’s perspective. These issues are relevant to the study of ancient Greek particles as well. Hellwig 1974, for example, observes that words which are sometimes called “particle” are at other times designated as “adverb,” “interjection,” or “conjunction.”  Despite the problems surrounding the term “(Greek) particle,”  we adopt this term in the current monograph because it has been the common term in the long history of research on these words. 
§59. As for the issue of definition, Duhoux 1997a:16 posits that a defining feature of Greek particles is that they never form an utterance on their own, unlike adverbs.  This characteristic also makes them different from interjections, though Duhoux does not make this distinction explicit.  His 2006 article extends and clarifies the earlier discussion. He adds two other criteria to the provision that they never form an utterance on their own: they are morphologically invariant, and they have no referential meaning. 
§60. Bonifazi notes that Greek particles are mainly negatively defined, if at all, and proposes that we consider the pragmatic meaning of particles, that is, the interpersonal and/or procedural aspects of communication that they convey.  She argues that certain lexical items may function sometimes as adverbs, with propositional content, and sometimes as particles, without such content.  Koier 2013:28 also points out that particles form an ill-defined category, and chooses a working definition similar to Bonifazi’s: “an uninflected form with no referential function that manages the speaker-hearer interaction on a textual or social level.” 
§61. There is a clear similarity, then, between the way Greek particles are defined in modern scholarship and the way discourse markers are defined in discourse-marker studies: lack of propositional meaning is a standard criterion for both categories. Duhoux’s criterion for Greek particles, however, the feature of not forming an utterance on their own, is not used to define discourse markers in modern languages.
§62. As for the issue of classification, the list of Greek lexical items considered to belong to the category of particles varies from study to study.  Many modern scholars take Denniston’s 1950  list as their guideline, but some studies expand the range of words he includes.  Since the category is ill-defined, there can be no complete consensus about which words to include or exclude. 
§63. Studies on Greek particles adopt different approaches, just as discourse-marker studies do for the modern languages. Of the approaches described above, several Hellenists have applied insights from coherence approaches and Relevance Theory (see §§64-72 below). Construction Grammar is adopted in Koier 2013 on που (see §73 below), and now in Drummen 2019 on καί, τε, and δέ. Conversation Analysis is used by Van Emde Boas 2017a.  More important, many classicists use a generally pragmatic approach. That is, they pay attention to the linguistic and extralinguistic context of a passage: they treat text as discourse, as language in use.  Therefore their descriptions of particles are not restricted to syntax or to the level of the sentence. 
§64. Apart from discourse-marker studies on modern languages, contemporary analyses of particles in Latin have clearly influenced the research on Greek particles. The most well-known Latinist in this field is Kroon, who has published numerous articles and a monograph on the topic.  In her 1995 monograph, the author develops a specific coherence-based approach to discourse particles in Latin, building in particular on Halliday and Hasan 1976, Roulet et al. 1985,  Mann and Thompson (e.g. 1987), Schiffrin 1987, and Foolen 1993. Kroon establishes a definition of and distinction among the representational, presentational, and interactional levels of discourse.  The representational level concerns relations between states of affairs in a represented world. The presentational level involves the manner in which a language user organizes communicative units. The interactional level, finally, concerns the exchange and relationships between speaker and addressee within a particular communicative situation.
§65. Numerous other scholars have analyzed Latin particles using a pragmatic approach.  Particularly prolific scholars from the 1990s onwards include Orlandini,  Risselada,  and Rosén.  Recently, Schrickx has published several articles and a monograph on Latin particles.  In her 2011 book, this author points to problems of terminology and definition in the research on Latin particles, similar to the situation in studies of other languages. Schrickx mainly builds on Kroon 1995, and chooses Functional Discourse Grammar (see especially Hengeveld and Mackenzie 2008) as her approach. 
§66. Let us now turn to pragmatic analyses of Greek particles.  E.J. Bakker (in the other volumes referred to as Bakker; here to be distinguished from S.J. Bakker) has done important work in this area: from the 1980s onward he has published on several particles, primarily in Homer.  His approach is to a large extent based on the insights of the cognitive linguist Chafe, who argues that spoken discourse represents the speaker’s flow of consciousness.  Bakker approaches the Homeric poems in a similar way: he interprets the “intonation units” (“discourse acts” in our terminology  ) that make up the discourse as reflecting the singer’s focus of consciousness at a given moment. In other words, Bakker takes into account the wider discourse as well as the cognitive constraints on the speaker and listener.
§67. Sicking published several works on Greek particles in the 1980s and 1990s as well.  He implicitly adopts a pragmatic approach, as he emphasizes the general importance of analyzing segments larger than sentences, and of viewing language as a means for communication rather than merely a means for the expression of thoughts.  Sicking’s 1993 study on particles in Lysias is published together with Van Ophuijsen 1993 on four particles in Plato.  Like Sicking, Van Ophuijsen takes a general pragmatic approach.
§68. Wakker adopts a pragmatic approach in her various studies of (primarily classical) Greek particles, published from the 1990s onwards.  In her 1994 monograph on conditional clauses in several Greek genres she uses the framework of Functional Grammar, mainly developed by S.C. Dik in 1978 and 1989.  Adopting the framework’s distinction among several levels of a clause, Wakker divides the particles used in conditional clauses into two groups. The first group modifies conditional clauses as a whole and are labelled “scope particles”  ; the second group modifies the proposition put forward in a conditional clause. Under the “scope particles” she includes γε, περ, and καί. The second group, “modal or attitudinal particles,” marks the speaker’s attitude to the proposition presented, and includes ἄρα, δή, που, and τοι. In later publications (1997a on γε, δή, ἦ, and μήν in tragedy; 2009a on οὖν and τοίνυν in Lysias) Wakker adopts the discourse model developed by Kroon 1995 (see above, §64).
§69. Revuelta Puigdollers has investigated several Greek particles in the 1990s and 2000s.  He pays special attention to αὖ and αὖτε, and to πάλιν. In his 2006a article on πάλιν Revuelta Puigdollers uses, as Wakker does, ideas from Functional Grammar as well as from the discourse model developed by Kroon 1995. His 2009a and 2009b publications, which mainly analyze αὖ and αὖτε, explicitly apply Functional Grammar.
§70. The volume New approaches to Greek particles (1997) edited by Rijksbaron presents several applications of modern linguistic frameworks to the study of Greek particles. Apart from Wakker, also Basset 1997 on ἀλλά, Jacquinod 1997 on καίτοι, and Slings 1997a on adversative particles use specific approaches that fall under pragmatics. 
§71. Bonifazi 2008, 2009, and 2012 throw light on several particles and adverbs in Homeric Greek. The author analyzes αὖ, αὖτε, αὐτάρ, αὖτις, αὐτίκα, αὔτως, αὖθι/αὐτόθι, and αὐτοῦ from a pragmatic perspective, explicitly applying insights from coherence approaches to discourse markers in modern languages, as well as from Relevance Theory, in particular the concept of “procedural meaning.”  She also builds on E.J. Bakker’s work on particles and Homeric discourse, and on Kroon’s 1995 research on Latin particles. In addition, Bonifazi pays attention to the role of memory constraints and the visual and narrative role of the lexical items under investigation.
§72. In 2009 S.J. Bakker and Wakker published Discourse cohesion in ancient Greek, a collection of papers on several cohesive devices, including particles. In this volume I myself analyze ἀλλά in drama; S.J. Bakker discusses γάρ and οὖν in Plato; Van Erp Taalman Kip writes about καὶ μήν, καὶ δή, and ἤδη in drama; Revuelta Puigdollers investigates αὖ and αὖτε in several genres; and Wakker delves into οὖν and τοίνυν in Lysias.  Each of these authors emphasizes different aspects of the broad framework of pragmatics.
§73. Other recent particle studies are the dissertations of Koier 2013 and Zakowski 2017, and several papers in Denizot and Spevak (eds.) 2017. Koier takes a Construction Grammar approach to που in several genres.  The author identifies co-textual features that lead to a specific interpretation of this polyfunctional particle. In addition, she carries out a diachronic analysis in order to describe the functional development of που from Homer to Isocrates. Zakowski uses Relevance Theory to analyze δέ, γάρ, and οὖν in several texts from the fourth century CE. He argues that it is more fruitful to describe these particles as procedural items, rather than as connected to coherence. δέ, γάρ, and οὖν help a hearer’s interpretation process, each in their own specific way. In the 2017 volume edited by Denizot and Spevak, Allan analyzes Greek adversative particles as polysemous networks of semantic extensions, and Crespo gives a unitary description of καί. Other articles focus on Latin particles.
§74. Several pragmatic approaches have been applied to the analysis of ancient Greek and Latin particles over the past few decades. The studies make it clear that we can understand particle use better than before if we pay attention to the larger co-text—that is, beyond the single sentence in which a particle occurs—and to communicatively relevant aspects of the extralinguistic context. These insights, taken over from research on modern languages, have led to a shift of attention, from a sentence-based view to a more discourse-based view, in the study of Greek and Latin particles.
§75. In recent years much linguistic research has been carried out on words in various languages that are functionally similar to ancient Greek particles. These words are often called “discourse markers.” Three related issues surrounding them recur: terminology, definition, and classification. Despite disagreements on these points, scholars of discourse markers have developed many theoretical insights and methodological tools.
§76. Different approaches are used in discourse-marker studies, with varying assumptions, questions, and methods. This chapter has discussed four of them: coherence approaches, Conversation Analysis (CA), Relevance Theory (RT), and Construction Grammar (CxG). Beyond these specific approaches, cross-linguistic, diachronic, synchronically historical, and narrative discourse-marker studies have also been useful to us. Finally, approaches similar to the ones described here have already been applied to the study of Greek and Latin particles by numerous classicists.
§77. The study of Greek particles can profit from taking into account both the findings of previous work on the particles themselves—this monograph’s “horizontal axis”—and the theoretical and methodological insights of contemporary research on functionally similar words in other languages—our work’s “vertical axis.” This chapter is designed to inform the reader of those studies on the vertical axis that have particularly influenced the approaches and analyses in this multi-part work. Their intellectual footprint is discernible in our efforts to be sensitive to the kind of discourse being studied (as in coherence approaches), to the communicative actions performed by speakers (as in CA), to the constraints on interpretation that particles may encode (as in RT), and to the specific co-textual features that influence our interpretation (as in CxG). In other words, the insights of these and similar approaches have led us to approach Greek particles the way we do: not as syntactically irrelevant parts of sentences, but as pragmatically crucial parts of verbal interactions.
[ back ] 1. See I.2 §§90-91 for more on Devarius’ work, and the heterogeneous list of Greek words he regards as particles.
[ back ] 2. Other monographs on Greek particles from the nineteenth century are Stephens 1837 and Paley 1881. Other important works are Klotz’s edition (1835) of and commentary (1842) on Devarius.
[ back ] 3. Des Places 1929, Denniston 1950 , Labéy 1950, Thrall 1962, Blomqvist 1969, Thyresson 1977, Van Ophuijsen and Sicking 1993, and Redondo Moyano 1995. In the twenty-first century, the dissertations of Koier 2013 and Zakowski 2017 (see §73 below) have been added to this list.
[ back ] 4. The descriptions of a selection of particles found in these works are summarized in part V, our online repository of particle studies.
[ back ] 5. See §§7-10 below for discussion of terminology such as “discourse markers”; see I.1 §§12-13 on the general notion of discourse.
[ back ] 6. We have also profited from contemporary frameworks of a more general nature and their applications, that is, from research that does not (usually) deal with particles or discourse markers, but with general pragmatic and cognitive phenomena. Consider for example general analyses of narrative discourse, which illustrate that such discourse comes in many varieties, such as oral vs. written, and conversational vs. literary. The studies provide insight into e.g. the stories’ internal structure and discourse transitions, their formulaic nature, and the cognitive processing of anaphoric expressions. Oral narrative, mainly in conversational settings, is the focus of e.g. Georgakopoulou 1997 and 2007; Norrick 2000; papers in Schiffrin, De Fina, and Nylund (eds.) 2010; and Rühlemann 2013 (this book includes remarks on the functions of oh and well in English conversational narrative). In contrast, Sanford and Emmott 2012, for instance, focus on written narrative, adopting a cognitive approach. Examples of other general frameworks besides narrative studies we have utilized in our chapters are the Birmingham and Geneva schools on discourse structure and segmentation (II.2, II.3, IV.3); cognitive linguistics (II.2, II.4, II.5); register studies (III.2); dialogic syntax (III.3); Du Bois’ approach to stance (IV.4).
[ back ] 7. See also I.1 §§8-11; I.2 §1, §§14-17, §§54-55, §§61-65, §§90-91 on these problems concerning ancient Greek particles.
[ back ] 8. Brinton 1996:29 cites the following other terms: “comment clause, connective, continuer, discourse connective, discourse-deictic item, discourse operator, discourse particle, discourse-shift marker, discourse word, filler, fumble, gambit, hedge, initiator, interjection, marker, marker of pragmatic structure, parenthetic phrase, (void) pragmatic connective, pragmatic expression, pragmatic particle, and reaction signal.” Brinton is herself the first to use “pragmatic marker” instead. Taboada 2006:572 adds even more terms: “coherence discourse markers, (…) lexical discourse markers, (…) sentence connectives, cue phrases, clue words, discourse signalling devices”.
[ back ] 9. The term “discourse marker” is used by e.g. Schiffrin 1987, 2006; Brinton 1990; Redeker 1990, 2006; Jucker 1993, 1997; Fludernik 1995, 2000; Lenk 1997, 1998; Bell 1998; Jucker and Ziv 1998a, 1998b; Hansen 1998a; Risselada and Spooren 1998; Rouchota 1998; Fraser 1999; Schourup 1999; Bazzanella and Morra 2000; Archakis 2001; Norrick 2001; Waring 2003; Maschler 2003, 2009; Fleischman and Yaguello 2004; Onodera 2004; Bazzanella 2006; Bolden 2006, 2009; Fox Tree 2006; Stvan 2006; Taboada 2006; Yang 2006; Furman and Özyürek 2007; Cuenca and Marín 2009; Fairbanks 2009; Fuami 2009; Pons Bordería and Estellés Arguedas 2009; Christodoulidou 2011; Lee-Goldman 2011; Lewis 2011; Schourup 2011; Mazeland 2012; and Mišković-Luković and Dedaić 2012.
[ back ] 10. The term “pragmatic marker” is used by e.g. Brinton 1996, 2006; Andersen 1998, 2001; Andersen and Fretheim (eds.) 2000; Erman 2001; Aijmer and Simon-Vandenbergen 2004; González 2004; Aijmer, Foolen, and Simon-Vandenbergen 2006; Defour 2008; Norrick 2009a, 2009b; Feng 2010; Fischer 2010; and Aijmer 2013.
[ back ] 11. The term “discourse particle” is used by e.g. Schourup 1982; Abraham (ed.) 1991; Hansen 1998b; Hakulinen 1998, 2001; Fischer 2000, 2006a, 2006b; Aijmer 2002; Yılmaz 2004; Bolden 2008; Lam 2010; Briz and Estellés 2010; Mazeland and Plug 2010; and Mišković-Luković and Dedaić 2010.
[ back ] 12. The term “pragmatic particle” is used by e.g. Hölker 1990; Östman 1991, 1995; Foolen 1996; Kirsner and Van Heuven 1999; Fujii 2000; Beeching 2002; Fried and Östman 2005; Mišković-Luković 2009; and Denis 2015. [ back ] Note that Aijmer, Brinton, Fischer, Hansen, Mišković-Luković and Dedaić, Norrick, and Schourup use different terms in different publications: “discourse particle” in Schourup 1982, Hansen 1998b, Fischer 2000, 2006a, 2006b, Aijmer 2002, and Mišković-Luković and Dedaić 2010; “discourse marker” in Brinton 1990, Hansen 1998a, Schourup 1999, 2011, Norrick 2001, and Mišković-Luković and Dedaić 2012; “pragmatic marker” in Brinton 1996, 2006, Norrick 2009a, 2009b, Fischer 2010, and Aijmer 2013; “pragmatic particle” in Mišković-Luković 2009.
[ back ] 13. See Mišković-Luković and Dedaić 2010:4-5.
[ back ] 14. See also Degand, Cornillie, and Pietrandrea (eds.) 2013a, and the editors’ introduction to that volume, for elaborate discussion of this problem of definition. The editors conclude that “DMs [discourse markers] and MPs [modal particles] are two subclasses of the general class of pragmatic markers.” (2013b:15)
[ back ] 15. See e.g. Mišković-Luković and Dedaić 2010:2-3.
[ back ] 16. For example Schiffrin 1987, Lenk 1998, Taboada 2006. See also §§18-25 below on coherence-based approaches to discourse markers.
[ back ] 17. For example Andersen 2001.
[ back ] 18. See also Andersen 2001:40, who restates his restricted understanding of the term “discourse marker.” We adopt a more comprehensive view of the term “discourse” in this monograph: see I.1 §§12-13.
[ back ] 19. See e.g. Schourup 1999 for an elaborate overview, and Taboada 2006:572 for a concise one.
[ back ] 20. Jucker and Ziv 1998b:2-3.
[ back ] 21. See IV.2 §4 and §§138-146 on this issue. See also e.g. Georgakopoulos and Goutsos 1998, and papers in Laury (ed.) 2008.
[ back ] 22. This criterion is used by e.g. Hölker 1990:78; Brinton 1996:6; Lenk 1997; Risselada and Spooren 1998:131; Schourup 1999:232; Imo 2013:159, 180. However, Andersen 2001:40 argues that “non-propositionality is only partly a valid criterion, because some pragmatic markers can be seen to have truth-conditional implications.” Andersen’s research concerns English, but his point also holds for several Greek particles: the same word may have propositional uses as well as non-propositional ones. We do not adopt non-propositionality as a defining criterion in this monograph, despite its centrality in many discourse-marker studies. Instead, we want to explore the different uses of these polyfunctional words without being potentially restricted by a priori distinctions. See I.1 and IV.2 for discussion of our view on this.
[ back ] 23. The criterion of syntactic flexibility is also stated by e.g. Fischer 2006b:4; Lenk 1997:2; Schourup 1999:242; Imo 2013:180.
[ back ] 24. Beeching 2002 on French pragmatic particles follows Brinton’s definition apart from the criterion of sentence-initial position.
[ back ] 25. See also Fischer 2000 on discourse particles (in her terminology) in different positions of a turn.
[ back ] 26. See e.g. Ferrara 1997 on the different prosodic profiles of anyway signaling different functions; Yang 2006 on the prosody of Chinese discourse markers; and Barth-Weingarten 2012 on and. Already Weydt 1969 offers insightful remarks on the prosodic realizations of German particles (e.g. on 39, 45-47, 55-58); however, he uses constructed examples.
[ back ] 27. See e.g. Weydt 1969 and Abraham (ed.) 1991 for early studies on German particles. For comparisons of cognate German and Dutch particles, see e.g. Foolen 2006 (doch vs. toch); Van Bergen, Van Gijn, Hogeweg, and Lestrade 2011 (eigentlich vs. eigenlijk).
[ back ] 28. Wierzbicka 1986b:519 even claims that the relative neglect of (in her terminology) particles in linguistic theory until at least 1986 was partly due to the focus on English, in which the role of particles is relatively limited. On the prevalence of discourse-marker studies on English, see also Schourup 1999:261 (this prevalence was even greater at the time than it is now).
[ back ] 29. Examples of discourse-marker studies on non-Indo-European languages are R. Blass 1990 on Sissala (a language spoken in Ghana and Burkina Faso); Luke 1990 on Cantonese Chinese; Copeland 1997 on Tarahumara (a Mexican indigenous language); Fujii 2000 and Onodera 2004 on Japanese; Wouk 2001 on Indonesian; Walrod 2006 on Philippine languages; Yılmaz 2004, and Furman and Özyürek 2007 on Turkish; Keevallik 2008 and Valdmets 2013 on Estonian; Laury and Seppänen (eds.) 2008 on Finnish; Fairbanks 2009 on Ojibwe (an endangered North-American language); Maschler 2009 on Hebrew; Dér and Markó 2010 on Hungarian; Feng 2010 on Chinese; Chang and Su 2012 on Taiwanese; Masinyana 2013 on Xhosa (a South African language).
[ back ] 30. See Lenk 1997 for a similar view.
[ back ] 31. For this metaphor, see also Onodera 2011:620-623 (referred to in §12 above). Similarly, Rijksbaron 1997b:14 compares Greek particles to “policemen controlling the traffic,” following Jespersen 1933:404, who compares particles and other small words to policemen that direct the other words to their proper place in the hearer’s brain.
[ back ] 32. On the (functional) subclassification within the category of discourse markers, see e.g. Schiffrin 1987:e.g. 316-317; Brinton 1996:36-39; Fraser 1999; Louwerse and Mitchell 2003.
[ back ] 33. Aijmer 2002:7-8 mentions a different list of useful approaches to discourse particles, noting that only formal grammar seems to have less to say about such words. A formal, i.e. generative approach assumes that language users construct new utterances on the basis of subconscious rules, rather than from their knowledge of specific instances in context. Surprisingly (in view of the emphasis on non-propositional functions and context-dependence in discourse-marker studies), Urgelles-Coll 2010 does use a generative approach in her monograph on English anyway. See also e.g. Feng 2010:165-181 and Aijmer and Simon-Vandenbergen 2011 for elaborate overviews of discourse-marker approaches.
[ back ] 34. On Gumperz’ notion of “contextualization cues” (coined in Cook-Gumperz and Gumperz 1976), see also III.4 §12. On the concepts of coherence and cohesion, see e.g. Drummen 2013b and 2013c, with further literature.
[ back ] 35. Other applications of a coherence-based approach include Redeker 1990 on a subclassification of discourse markers; Maschler 1998; 2003; 2009 on several Hebrew discourse markers; Risselada and Spooren 1998 on discourse markers and coherence relations; Bazzanella and Morra 2000 on translating English discourse markers into Italian; Norrick 2001 on discourse markers in English oral narrative; González 2004 on pragmatic markers in English and Catalan oral narrative; Aijmer, Foolen, and Simon-Vandenbergen 2006 on a methodological proposal concerning pragmatic markers in translation; Taboada 2006 on discourse markers and rhetorical relations; Fairbanks 2009 on discourse markers in Ojibwe; Pons Bordería and Estellés Arguedas 2009 on Spanish digressive markers; Aijmer 2013 on several English pragmatic markers.
[ back ] 36. See Maschler 2009:1-2.
[ back ] 37. See Barron 2014 for an concise overview of variational pragmatics, and Schneider and Barron (eds.) 2008 for several papers using this approach.
[ back ] 38. This example of well is also analyzed in Aijmer 2009:7.
[ back ] 39. All quotations in this paragraph are from Aijmer 2013:32.
[ back ] 40. Helpful recent introductions to CA are Schegloff 2007 and Sidnell 2010, and in shorter form Gardner 2005 and Heritage 2010. See III.4 for elaborate discussion of CA and an application of this approach to the analysis of particle use in Greek tragedy and comedy.
[ back ] 41. Prior to this publication, Sacks had given lectures on conversation (1964-1968), posthumously published as Sacks 1995; and Schegloff had analyzed conversational openings (1967, 1968).
[ back ] 42. Examples of a CA approach to discourse markers are Jefferson 1983 on yeah and mmhm; Heritage 1984, 1998, 2002 on oh; Luke 1990 on several Cantonese Chinese particles; Hakulinen 1998 and 2001 on several Finnish particles; Mazeland and Huiskes 2001 on Dutch maar; Sorjonen 2001 on Finnish response particles; Waring 2003 on also; Raymond 2004 and Bolden 2006, 2009 on so; Bolden 2008 on Russian -to; Person 2009 on oh in Shakespeare; Mazeland and Plug 2010 on Dutch hoor; Christodoulidou 2011 on several Cypriot Greek discourse markers; Gaines 2011 on okay; Lee-Goldman 2011 on no; Mazeland 2012 on Dutch nou; Imo 2013 on German ja (combining CA with other frameworks).
[ back ] 43. See also the similar framework of Interactional Linguistics: e.g. Barth-Weingarten 2012; Ochs, Schegloff, and Thompson (eds.) 1996; Selting and Couper-Kuhlen (eds.) 2001. Couper-Kuhlen and Selting call Interactional Linguistics a research field “at the intersection of linguistics, conversation analysis and anthropology” (2001:1).
[ back ] 44. On actions in CA, see e.g. Schegloff 2006:73; 2007:xiv; Sidnell 2010:61; Sidnell and Enfield 2012:328; Enfield 2013:86-103. See also the discussion in III.4 §§19-22.
[ back ] 45. On the notion of “adjacency pair” in CA, see III.4 §§15-16.
[ back ] 46. See III.4 (passim) for the relevance of these positions for a number of Greek particles in drama.
[ back ] 47. Note, incidentally, the other turn-initial oh, spoken by speaker B: this one is in an answer to a question, which implies, according to Heritage’s analysis, that the question was somehow problematic.
[ back ] 48. See e.g. Wilson and Sperber 2004 for a compact overview of Relevance Theory. See Wilson and Sperber 2012b for an elaborate, recently updated discussion of Relevance Theory; this collection of papers also includes previously published material by the two authors.
[ back ] 49. See e.g. Grice 1989:26-27 on the corresponding “maxims” (as he calls them) that speakers are expected to conform to. These concepts have been highly influential in pragmatics.
[ back ] 50. See Andersen and Fretheim (eds.) 2000:7: “From a relevance-theoretic point of view, pragmatic markers can be seen to facilitate inferential processes.” See also Andersen 2001:33: pragmatic markers (in his terminology) “contribute to relevance by telling the hearer how an utterance is to be understood, thus reducing the processing effort that the hearer must employ in utterance comprehension.”
[ back ] 51. Concerning discourse markers, RT has been elaborated upon by e.g. Blakemore 1987 (monograph); 2000; 2002 (monograph); R. Blass 1990 (monograph); Haegeman 1993; Jucker 1993; Andersen 2001 (monograph); Ifantidou 2001 (monograph); Mišković-Luković 2009; Fielder 2010; Schourup 2011; Mišković-Luković and Dedaić 2012; several contributions in Jucker and Ziv (eds.) 1998a (e.g. Rouchota on parenthetical discourse markers); Andersen and Fretheim 2000; and Dedaić and Mišković-Luković 2010; Schourup 2011.
[ back ] 52. See Blakemore 2002:149-185 for a full discussion. See also Schourup 2011:2110 on RT as different from discourse- or coherence-oriented approaches. Interestingly, Lenk 1998 combines a coherence-based and a relevance-theoretic approach in her study of “pragmatic markers”.
[ back ] 53. See e.g. Blakemore 2002:155.
[ back ] 54. Exceptions are, for example, Zakowski 2017 on Ancient Greek (see §73 below) and Fielder 2010, who applies RT to the connective ama in Bulgarian, which is not her native language. Both studies use only corpus material and no constructed examples..
[ back ] 55. The prominent concept in RT of (minimizing) the hearer’s processing effort also plays a role in cognitive analyses of, for example, anaphoric expressions; see e.g. II.5 on this topic.
[ back ] 56. Other frameworks developed in the same vein are for example Functional Grammar (see e.g. Dik 1968; 1978; 1989), Cognitive Grammar (see below within §41, with note 58), and Emergent Grammar (see e.g. Hopper 1988; 2011; Bybee and Hopper (eds.) 2001).
[ back ] 57. The framework of generative linguistics originated in the work of Chomsky (e.g. 1965). Current overviews of generative linguistics are e.g. Philippi 2008; Ludlow 2011; Chomsky and McGilvray 2012; Den Dikken (ed.) 2013.
[ back ] 58. See Langacker 2008 and 2010 for recent overviews of Cognitive Grammar.
[ back ] 59. See also Bybee 2010 on general cognitive principles as an argument for a usage-based theory of grammar, such as CxG.
[ back ] 60. Östman and Fried 2004 give an overview of the historical and intellectual background of Construction Grammar. E.g. Traugott and Trousdale 2013:2-8 give a short overview of different constructional approaches to language. See e.g. the recent Oxford Handbook of Construction Grammar (Hoffmann and Trousdale [eds.] 2013) for a detailed overview of the theoretical substance, development, and sub-approaches within Construction Grammar.
[ back ] 61. Applications of Construction Grammar to discourse-marker studies include Fujii 2000 on Japanese mono; Fried and Östman 2005 on pragmatic particles in Czech and Solv (a Swedish dialect spoken in Finland); Imo 2005 on I mean (interestingly combining CxG and CA); Imo 2007 on several discourse markers developed from verbs in spoken German; Diewald 2008 on German ruhig; Imo 2008 on German halt; Fried 2009a on Czech jestli; Fischer 2010 on several pragmatic markers in spoken English; Masini and Pietrandrea 2010 on Italian magari; Lewis 2011 on the diachronic development of instead and rather; Fischer and Alm 2013 on German also and Swedish alltså; Van der Wouden and Foolen 2015  on several Dutch utterance-final particles; Koier 2013 on ancient Greek που and Dutch ergens.
[ back ] 62. See e.g. Traugott and Trousdale 2013:1.
[ back ] 63. See e.g. Fried and Östman 2004:24; Traugott and Trousdale 2013:3.
[ back ] 64. Traugott and Trousdale 2013:31 note that outside of CxG the term “construction” has usually been associated with syntax only; however, in CxG it is explicitly extended to include lexicon as well.
[ back ] 65. See III.2 for examples and discussion of large-scale constructions, applied to the analysis of Greek drama.
[ back ] 66. On the relationship between CxG and both monosemy and polysemy, see e.g. Boogaart 2009; Drummen 2013a; Koier 2013. On networks and CxG, see e.g. Traugott and Trousdale 2013:8-11. For clarification about how the term “polysemy” has been used in CxG, see Traugott and Trousdale 2013:200-202.
[ back ] 67. See Croft and Cruse 2004:263-264 for this example.
[ back ] 68. Koier also analyzes ancient Greek που in the same monograph; see §73 below.
[ back ] 69. A similar online-survey method is used by e.g. Van Bergen, Van Gijn, Hogeweg, and Lestrade 2011 in their investigation of the Dutch discourse marker eigenlijk.
[ back ] 70. Other examples of cross-linguistic discourse-marker studies are Fraser and Malamud-Makowski 1996 on English and Spanish; Takahara 1998 on English and Japanese; Bazzanella and Morra 2000 on Italian and English; Aijmer and Simon-Vandenbergen 2003 on English, Swedish, and Dutch; González 2004 on English and Catalan; Foolen 2006 on Dutch and German; Bouma, Hendriks, and Hoeksema 2007 on Dutch, English, and German; Altenberg 2010 on English and Swedish; Sudhoff 2012 on German and Dutch; Bruijnen and Sudhoff 2013 on German and Dutch; Fischer and Alm 2013 on German and Swedish; Izutsu and Izutsu 2013 on German, French, Japanese, and English; Koier 2013 on Dutch and Ancient Greek (see §73 below); Squartini 2013 on Italian and French.
[ back ] 71. In this monograph we draw comparisons to English discourse markers in II.3 §26 and §72 (because and for compared to γάρ), II.5 §§53-56 (of course compared to ἄρα), III.4 (§§51-52 okay compared to καὶ δή and ἰδού, §55 well compared to μέν), IV.2 (and compared to δέ, τε, and καί).
[ back ] 72. Examples of diachronic discourse-marker studies (with or without a specific approach) are Traugott 1986 on but and and; Brinton 1990 and 2006 on English; Fludernik 1995 on Middle English þo; Jucker 1997 on well; Traugott and Dasher 2002:152-189; Onodera 2004 on Japanese; Diewald 2008 on German ruhig; Defour 2008 on well and now; Stvan 2006 on why and say; Visconti 2009 on Italian mica; Meurman-Solin 2012 on and, for, but, and only; Bolly and Degand 2013 on French discourse markers derived from voir (“to see”); Denis 2015 on general extenders (such as and stuff) and epistemic parentheticals (such as I think) in Canadian English; and papers in Davidse, Vandelanotte, and Cuyckens (eds.) 2010 and in Degand and Simon-Vandenbergen (eds.) 2011 on discourse markers in several languages. Heine 2013 discusses approaches to the genesis and development of discourse markers in general.
[ back ] 73. See e.g. Davidse, Vandelanotte, and Cuyckens 2010 on these three processes. See e.g. Traugott and Trousdale 2013 for elaborate discussion of several kinds of language change, including grammaticalization and pragmaticalization.
[ back ] 74. On the use of this marker, see also Mišković-Luković 2009.
[ back ] 75. We refer to (suggestions about) particles’ development in II.2 §§46-48 (μέν) and II.3 (§§40-42 ἦ, §57 δή).
[ back ] 76. Other examples of synchronic studies on historic corpora are Fludernik 1995 and 2000 on discourse markers in Middle English; Blake 1996 on why and what in Shakespeare; Fuami 2009 on well in Shakespeare; Person 2009 on oh in Shakespeare; Lutzky 2012 on three discourse markers in Early Modern English; Lutzky and Demmen 2013 on pray in Early Modern English; Jonker 2014 on several discourse markers in Jane Austen.
[ back ] 77. In this monograph, these studies are especially relevant to the research on Homer and Pindar (part II) and Herodotus and Thucydides (part IV).
[ back ] 78. Other examples are Koike 1996 on Mexican Spanish ya; Fox Tree 2006 on like; Fairbanks 2009 on discourse markers in Ojibwe (an endangered North American language); Furman and Özyürek 2007 on Turkish discourse markers.
[ back ] 79. For studies on Modern Greek discourse markers, see e.g. Brewster 1992 on lipon; Georgakopoulos and Goutsos 1998 on the distinction between conjunctions and discourse markers; Ifantidou 2000 on taha; Archakis 2001 on diladi, m ala loja, thelo na po, and i malon; Christodoulidou 2011 on lipón, ára, and oréa in Cypriot Greek.
[ back ] 80. See also I.2 §1, §§14-15, §§18-19 on this issue.
[ back ] 81. See e.g. Revuelta Puigdollers 2006a:468-469; Bonifazi 2012:201.
[ back ] 82. See I.1 §§8-11.
[ back ] 83. Duhoux 1997a:16: “Comme les adverbes, les particules sont invariables, mais elles diffèrent d’eux en ceci qu’elles ne peuvent pas être employées de façon autonome: elles doivent obligatoirement être utilisées avec d’autres mots, alors qu’un adverbe peut, à lui seul, constituer un énoncé (ainsi, Καλῶς, ‘Bien’).” He adds in a footnote that this is why he excludes οὐ and μή from his study on particles.
[ back ] 84. Hölker 1990 also mentions this criterion, concerning French, in order to distinguish connecting markers from interjectional markers: the former never occur as independent utterances, he claims.
[ back ] 85. Ancient scholars also used these criteria: see I.2 §16, §§54-55, §80, §83.
[ back ] 86. See Bonifazi 2008:36-37; 2009:30.
[ back ] 87. Bonifazi 2012:10, 186.
[ back ] 88. She here refers to the presentational and interactional levels of discourse as used by Kroon 1995 (see §64 below).
[ back ] 89. See e.g. Páez Martínez 2012 §4 on this issue.
[ back ] 90. See e.g. Blomqvist 1969 on several other particles in Hellenistic Greek; Muchnová 1993, 2000, 2003, 2006, 2009 and 2011 on ἐπεί; Sicking 1993:17-18 on εἶτα and ἔπειτα; Duhoux 1997a:16-17, who adds αὖ, ἤ, νυν, and ὅμως; e.g. Basset 1988 and Wathelet 1997 on ἄν and κε(ν) as particles; Muchnová 2004 on ὡς, and 2007 on ὅτι; Revuelta Puigdollers 2009b and Bonifazi 2012:185-292 on several αὖ-particles.
[ back ] 91. On the (functional) subclassification within the category of ancient Greek particles, see e.g. Stephens 1837; Denniston 1950 ; Blomqvist 1969; Hellwig 1974; Sicking 1986; George 2009; E.J. Bakker 2011.
[ back ] 92. See III.4 for the application of CA to the study of particles in ancient Greek drama. See also Van Emde Boas 2017b for a CA approach to Greek tragedy; particles are however not his main focus here.
[ back ] 93. See I.1 §§12-17 on the notions of discourse and pragmatics. On pragmatics in general, see e.g. Verschueren 1999; Grundy 2000; Mey 2001; Horn and Ward (eds.) 2006; Huang 2007; Cutting 2008. On applying a pragmatic approach to ancient Greek, see Collinge 1988; Bonifazi 2001, 2012; and E.J. Bakker 2010b.
[ back ] 94. S.C. Dik’s framework of Functional Grammar (see e.g. S.C. Dik 1978; 1989) has been highly influential and is sometimes even seen as equivalent to “pragmatics.” Examples of Functional-Grammar applications to ancient Greek are Wakker 1994 (see §68 below), H.J. Dik 1995 (see note 112 below), Revuelta Puigdollers 2009a (see §69 below), and Polo Arrondo 2007; 2011; forthcoming. In this monograph we subscribe to a broader view of pragmatics, of which Functional Grammar is only one of several sub-branches.
[ back ] 95. Kroon 1989 on nam, enim, igitur, and ergo; 1992 on several particles; 1994a on at; 1994b on Latin equivalents of Dutch maar; 1995 (monograph) on nam, enim, autem, vero and at; 1998 on a framework for Latin discourse markers; 2004, 2005, and 2009 on quidem; Kroon and Risselada 1998 and 2002 on iam.
[ back ] 96. II.2 and II.3 are to a large extent informed by the work of Roulet and colleagues (the so-called “Geneva school”).
[ back ] 97. This approach is explicitly adopted by e.g. Wakker 1997a on γε, δή, ἦ, and μήν in tragedy; 2009a on οὖν and τοίνυν in Lysias; Schrickx 2011 on Latin particles; Bonifazi 2012:185-291 on several discourse markers in Homer.
[ back ] 98. Examples are Revuelta Puigdollers 1998 on Latin “focusing particles”; Langslow 2000 on the diachronic development of several Latin particles, and the influence of genre differences; Moore 2006 on videlicet; Tarriño Ruiz 2009 on several Latin adverbs and particles; Molinelli 2010 on the pragmaticalization of certain Latin verb forms into discourse markers; Pinkster 2010 on quia and quoniam; Holmes 2011 on nam; Goldstein 2013a on nedum. See also the elaborate online bibliography on Latin particles at http://latinparticles.userweb.mwn.de (last accessed on 11 July, 2015), maintained by Josine Schrickx.
[ back ] 99. Her particle publications include Orlandini 1994 on at; 1995 on atqui and immo; 1999 on autem and ceterum; Orlandini and Poccetti 2007 and 2009 on several particles.
[ back ] 100. Her particle publications include Risselada 1994 on modo and sane in directives; 1996 on nunc; 1998a on sane; 1998b on tandem and postremo; 2005 on particles in questions; Kroon and Risselada 1998 and 2002 on iam.
[ back ] 101. Her particle publications include Rosén 1989 on several Latin particles; 1993 on demum; 2002 on several Latin particles as cohesive devices; 2003 on immo; 1999 on several Latin connecting particles; 2007 on sed; 2009 on several Latin particles.
[ back ] 102. Her particle publications include Schrickx 2009 on nam and namque; 2010 on nempe; 2011 (monograph) on nempe, quippe, scilicet, videlicet and nimirum; 2014 on scilicet and videlicet.
[ back ] 103. See II.2 and II.3 for discussion of ideas from Functional Discourse Grammar. See also note 6 above.
[ back ] 104. In the following overview, I will restrict myself to monographs on particles and scholars who have published repeatedly on this topic. The numerous other articles on Greek particles that adopt a modern linguistic approach are mentioned and summarized in part V, our database ORPS (Online Repository of Particle Studies).
[ back ] 105. See E.J. Bakker 1986, 1988, and 1993c on περ; 1993a on δέ; 1993b on ἄρα; 1997a on many aspects of the Homeric language, including the use of ἄρα, αὐτάρ, γάρ, δέ, δή, καί, μέν, and οὖν; 2005 on γάρ, γε, δέ, καί, περ, and τε; 2011 for a short overview of Homeric particles. See especially part II of this monograph, on particle use in Homer and Pindar, for discussion of E.J. Bakker’s particle analyses.
[ back ] 106. See e.g. Chafe (ed.) 1980, 1994. His research is discussed in II.2.
[ back ] 107. On discourse acts, see especially II.2 (with extensive discussion of E.J. Bakker’s work) and IV.3.
[ back ] 108. See Sicking 1986 on classical Greek; 1993 on Lysias; 1996b and 1997 on Plato.
[ back ] 109. Sicking 1986:125-126, 138-139; 1993:7-8; 1997:157, 173-174. On 1986:136 and 1993:48 he cites the framework of Functional Grammar by S.C. Dik 1978 (see note 94 above).
[ back ] 110. Van Ophuijsen and Stork 1999 also pay considerable attention to particle use in their commentary on parts of book 7 of Herodotus’ Histories.
[ back ] 111. See Wakker 1994 on ἄρα, γε, δή, καί, περ, που, and τοι in conditional clauses in several authors; 1995 on several particles in Sophocles; 1996 on μάν and μήν in Theocritus; 1997a on μήν in tragedy; 1997b on several particles in Herodotus and Thucydides; 2001 on μέν in Xenophon; 2002 on ἤδη in Xenophon; 2009a on οὖν and τοίνυν in Lysias; 2009b on several particles in several authors.
[ back ] 112. Similarly to Wakker, H.J. Dik uses Functional Grammar in her 1995 monograph on word order in Herodotus, and includes some observations on particle use. Polo Arrondo 2007 and 2011 on πλήν also uses the framework.
[ back ] 113. See I.1 §20 for a brief discussion of “scope” as used in this monograph; we do not adopt the term “scope particles.”
[ back ] 114. His particle research includes Revuelta Puigdollers 1996 (the author’s unpublished dissertation) on αὖ, αὖτε, αὖθις, ἄψ, πάλιν and ὀπίσω; 2000 on focus particles; 2006a on πάλιν; 2006b is an overview (in Spanish) of Greek particles; 2009a and 2009b on αὖ, αὖτε, and other particles as topicalizing devices (2009b is a more elaborate version of 2009a); 2013 is an encyclopedia entry on Greek particles in general.
[ back ] 115. Basset and Jacquinod mainly build on the ideas of Ducrot 1972, 1975, 1980, 1984; Slings on those of Polanyi and Scha 1983, and Roulet et al. 1985.
[ back ] 116. Bonifazi 2012 drops “procedural meaning” in favor of the terms proposed by Kroon 1995 (see §64 above).
[ back ] 117. R.J. Allan 2009 on narrative modes in Euripides in the same volume also provides many remarks on particles. Additionally, George 2009 in the same volume is highly relevant to the study of particles: he qualifies the findings of Duhoux 1997a; 1997b; 1998; 2006 on the distribution of particles across dialogic versus non-dialogic texts.
[ back ] 118. See §§47-50 above on Koier’s analysis of the Dutch particle ergens, published in the same monograph.