2.1Discourse markers

Studies on DMs suffer from a lack of consensus at the level of definition. The question of how to define DMs has been a point of debate in research in this field (e.g., Fraser 1996Fraser, Bruce 1996 “Pragmatic Markers.” Pragmatics 6 (2): 167–190. ; Schourup 1999Schourup, Lawrence 1999 “Discourse Markers.” Lingua, 107 (3–4): 227–265. ; Schiffrin 2001Schiffrin, Deborah 2001 “Discourse Markers: Language, Meaning, and Context.” In The Handbook of Discourse Analysis, ed. by Deborah Schiffrin, Deborah Tannen, and Heidi E. Hamilton, 54–74. Malden. Oxford: Blackwell.; Crible 2017aCrible, Ludivine 2017a “Discourse Markers and (Dis)fluency in English and French.” International Journal of Corpus Linguistics 22 (2): 242–269. ; Heine et al. 2021Heine, Bernd, Gunther Kaltenböck, Tania Kuteva, and Haiping Long 2021 The Rise of Discourse Markers. Cambridge: Cambridge University Press. ). Accordingly, DMs can be viewed through the lens of two main criteria, viz. syntactic (integration and scope) and pragmatic (multifunctionality) (see Crible 2017aCrible, Ludivine 2017a “Discourse Markers and (Dis)fluency in English and French.” International Journal of Corpus Linguistics 22 (2): 242–269. ). Prototypically, they are syntactically optional elements and not an integrated part of the core syntax. They are relatively mobile and grammatically heterogeneous. DMs convey procedural rather than propositional meaning. Besides, what is more specific to DMs is their multifunctionality. This feature of DMs can be depicted in two forms: (1) they may have different functions on different occasions of use (see Section 4); or (2) they may simultaneously have different functions on a single occasion of use (see Section 5.1; cf. Dér 2010Dér, Csilla 2010 “On the Status of Discourse Markers.” Acta Linguistica Hungarica 57 (1): 3–28. ; Aijmer and Simon-Vandenbergen 2011Aijmer, Karin, and Anne-Marie Simon-Vandenbergen 2011 “Pragmatic Markers.” In Discursive Pragmatics 8, ed. by Jan Zienkowski, Jan-Ola Östman, and Jef Verschueren, 223–247. Amsterdam: John Benjamins. ).

The general thrust of a pragmatic approach is concerned with the meaning of utterances, particularly with how a DM in an utterance relates the message to that of a prior utterance. It presumes a separation between sentence (conceptual) and utterance (procedural) meaning. Thus, we start with a classification of pragmatic meanings. To this end, we mainly use Fraser’s (forthcoming)Fraser, Bruce forthcoming. “Canonical Sequences of Discourse Markers in English.” model to classify the DMs in our data. The reason for employing his model for this study is threefold. First, his model and method of classification provide a versatile analytical tool11.The versatility of this model is due mainly to investigating DM classifications and co-occurrences simultaneously, which is hardly seen in other studies. As the current study investigates both these issues, this model would be a firm foundation and fit this study neatly. that greatly helps to establish a firm footing for this study. Second, his classification is deeply rooted in a pragmatic approach, which holds a dominant position throughout this paper. Moreover, Kassaei and Amouzadeh (2020)Kassaei, Gholamreza, and Mohammad Amouzadeh 2020 “The Combination of Discourse Markers in Persian.” International Review of Pragmatics 12 (1): 135–163. have already found his model well-suited for the classification of Persian DMs. Note that we do not follow Fraser’s model to the letter, so certain slight modifications will be made in order to meet our objectives. Specifically, we will employ his original model with two qualifications: (a) while Fraser’s model is quite qualitative, the current study will be a combination of qualitative and quantitative analyses; and (b) since væ is strongly multifunctional and the notion of simultaneous multifunctionality of DMs is not considered in Fraser (forthcoming)Fraser, Bruce forthcoming. “Canonical Sequences of Discourse Markers in English.”, this study tries to address the multifunctionality in question in terms of meaning potential.

Fraser (forthcoming)Fraser, Bruce forthcoming. “Canonical Sequences of Discourse Markers in English.” divides DMs into three major classes: retroactive (RDMs), linking (LDMs), and proactive (PDMs). The first class (RDMs) signal the speaker’s perception of a prior utterance. For example, the bold parts in (1) below are treated as RDMs because they reflect the speaker’s view of a former utterance. Yet, they designate different types of RDMs: ‘oh’, ‘I see’, ‘well’ and ‘ok’ are expressions, respectively, of surprise, recognition, and decision.

Fraser (forthcoming)Fraser, Bruce forthcoming. “Canonical Sequences of Discourse Markers in English.” takes the view that LDMs reflect the speaker’s perspective of the relationship between earlier and forthcoming utterances. This class can further be divided into three main subclasses, each subclass being comprised of primary and secondary DMs (see Figure 1).

Finally, the third class of DMs are PDMs. As it is shown by Example (3), PDMs are the converse of RDMs as they signal the speaker’s view of the following utterance rather than a preceding one. This class, similar to the other ones, comprises certain subclasses; namely, summarizing PDMs, attention-getting PDMs, commentary PDMs, illustrative PDM, and topic PDMs.

2.2Discourse marker co-occurrences

DM co-occurrences are simply defined as contiguity of two or more DMs in an adjacency or non-adjacency order. This phenomenon has received increasing attention over the last decade since it is directly germane to the mobility and polyfunctionality of DMs. More recent studies on DM co-occurrences can be found in Oates (2000Oates, Sarah Louise 2000 “Multiple Discourse Marker Occurrence: Creating Hierarchical for Natural Language.” In Procedding of the 3rd CLUK Colloquium, 41–45. Brighton., 2001Oates, Sarah Louise 2001 Multiple Discourse Occurrence: Creating Hierarchicies for Natural Languages Generation. MA dissertation. University of Brighton.), Lohmann and Koops (2016)Lohmann, Arne, and Christian Koops 2016 “Aspects of Discourse Marker Sequencing – Emprical Challenges and Theoretical Implications.” In Outside the Clause: Forms and Functions of Extra-clausal Constituents, ed. by Gunther Kaltenbock, Evelien Keizer, and Arne Lohmann, 417–446. Amsterdam: John Benjamins. , and Haselow (2019)Haselow, Alexander 2019 “Discourse Marker Sequences: Insights into the Serial Order of Communicative Tasks in Real-time Turn Production.” Journal of Pragmatics 146: 1–18. . Almost all of these studies examined the sequencing behavior and functional motivations underlying DM sequences from different perspectives. Certain motivations were proposed in earlier studies for such co-occurrences: (i) floor holding, (ii) functional specification, (iii) functional complementation (see Aijmer 2002Aijmer, Karin 2002 English Discourse Particles: Evidence from a Corpus. Amsterdam: John Benjamins. ; Haselow 2019Haselow, Alexander 2019 “Discourse Marker Sequences: Insights into the Serial Order of Communicative Tasks in Real-time Turn Production.” Journal of Pragmatics 146: 1–18. ).

The study of DM co-occurrences in Persian, however, seems to be limited, except for a few studies done by Ghaderi (2019)Ghaderi, Soleiman 2019 Baresi Mo’tarezeh Are/Na dar Zabane Farsi [The Thetical Aspects of Are/Na (Yes/No), in Persian]. PhD Thesis, University of Isfahan., Kassaei and Amouzadeh (2020)Kassaei, Gholamreza, and Mohammad Amouzadeh 2020 “The Combination of Discourse Markers in Persian.” International Review of Pragmatics 12 (1): 135–163. , and Ghaderi and Amouzadeh (2021)Ghaderi, Soleiman, and Mohammad Amouzadeh 2021 “Aspects of Are (Yes) in Persian Discourse: Its Functions, Positions, and Evolution.” Studia Linguistica 75 (3): 623–658. . The more relevant study by Kassaei and Amouzadeh (2020)Kassaei, Gholamreza, and Mohammad Amouzadeh 2020 “The Combination of Discourse Markers in Persian.” International Review of Pragmatics 12 (1): 135–163. systematically investigated the combinations of Persian DMs. The authors (2020) examined all possible combinations of thirty Persian DMs (i.e., elaborative, contrastive, and inferential). They also argued that the order of DMs can, to a great extent, be predictable through certain frequent patterns. Moreover, they found that contrastive DMs are apt to combine with those from their own category, while elaborative and inferential ones are liable to take part in intra-category combinations. One of the main concerns in their work was the analysis of the combinatory behavior of væ DM. They claimed that væ appears in an initial position in all its combinations with other DMs, and, consequently, conformed mainly to the findings of Oates (2000Oates, Sarah Louise 2000 “Multiple Discourse Marker Occurrence: Creating Hierarchical for Natural Language.” In Procedding of the 3rd CLUK Colloquium, 41–45. Brighton., 2001Oates, Sarah Louise 2001 Multiple Discourse Occurrence: Creating Hierarchicies for Natural Languages Generation. MA dissertation. University of Brighton.) and Fraser’s (2009Fraser, Bruce 2009 “An Account of Discourse Markers.” International Review of Pragmatics 1 (2): 293–320. , forthcomingFraser, Bruce forthcoming. “Canonical Sequences of Discourse Markers in English.”).

2.3 væ in Persian

Persian has three types of monosyndetic and bisyndetic coordinate conjunctions, namely conjunctive, adversative, and disjunctive (Stilo 2004Stilo, Donald 2004 “Coordination in Three Western Iranian Languages: Vafsi, Persian and Gilaki.” In Coordinating Constructions, ed. by Martin Haspelmath, 269–330. Amsterdam: John Benjamins. , 271), and væ is treated as a monosyndetic coordinating conjunction. Although the coordinating conjunction væ and the connective clitic -o ‘and’22.The clitic-o is more common than the conjunction in informal speech (Mahootian and Gebhardt 1997Mahootian, Shahrzad, and Lewis Gebhardt 1997 Persian. London: Routledge.). are very close, and most people believe that they are two modes (spoken/written) of a single word, they vary etymologically. The former is derived from Arabic while the latter is a survival from middle Persian u (see Stilo 2004Stilo, Donald 2004 “Coordination in Three Western Iranian Languages: Vafsi, Persian and Gilaki.” In Coordinating Constructions, ed. by Martin Haspelmath, 269–330. Amsterdam: John Benjamins. ; Lambton 1953Lambton, Ann KS. 1953 Persian Grammar: Including Key. Cambridge: Cambridge University Press.).

Old Persian: utā > ud > uδ > Middle Persian: u > =o, ‘and’(Kent 1953Kent, Roland G. 1953 Old Persian: Grammar. Texts. Lexicon. New Haven: American Oriental Society., 175; Horn 1893Horn, Paul 1893 Grundriss der Neupersischen Etymologie. Strassburg: Karl J. Trübner. , 240)

This means that væ and o can conjoin any number of sentences, which are in a coordinating, causal or temporal relationship. In other words, they can be used to coordinate different elements (Mahootian and Gebhardt 1997Mahootian, Shahrzad, and Lewis Gebhardt 1997 Persian. London: Routledge.):

• NP subjects/objects (two or more than two)

• VPs

• Attributive/predicate adjectives

• Attributive/predicate adverbs

• Adverbials (adverbs of manner and participial adverb constructions)

Almost all earlier studies on væ (e.g., Lazard 1992Lazard, Gilbert 1992 A Grammar of Contemporary Persian. Cosa Mesa: Mazda Publishers.; Mahootian and Gebhardt 1997Mahootian, Shahrzad, and Lewis Gebhardt 1997 Persian. London: Routledge.; Stilo 2004Stilo, Donald 2004 “Coordination in Three Western Iranian Languages: Vafsi, Persian and Gilaki.” In Coordinating Constructions, ed. by Martin Haspelmath, 269–330. Amsterdam: John Benjamins. ), have been confined to its syntactic and coordinative aspects, and heretofore none of them has been advertent to its discoursal use.

In this paper, we argue that væ cannot be simply and merely restricted to its coordinative nature. In most occurrences, it is semantically reduced and lacks propositional content. This gives rise to the emergence of pragmatic functions operating at higher levels, particularly at the discourse level. The functional properties of væ are by no means arbitrary, they are ascertained according to multiple features (e.g., prosodic, syntactic, pragmatic, etc.).33.These factors are also determinative of the order of DMs in sequences (see Crible and Degand 2021Crible, Ludivine, and Liesbeth Degand 2021 “Co-occurrence and Ordering of Discourse Markers in Sequences: A Multifactorial Study in Spoken French.” Journal of Pragmatics 177: 18–28. ). Therefore, the current study is the first of its kind that brings up væ, its multifunctionality and co-occurrences in terms of pragmatics, and it predominantly favors the pragmatic aspect of væ, which is rather overlooked in earlier studies.

4.1Elaborative function

The main function of væ can be described in terms of elaboration (see Kassaei and Amouzadeh 2020Kassaei, Gholamreza, and Mohammad Amouzadeh 2020 “The Combination of Discourse Markers in Persian.” International Review of Pragmatics 12 (1): 135–163. ). It is considered to be a Persian Primary Elaborative Linking Discourse Marker (PELDM). As its name suggests, it signals that the following utterance provides an elaboration of a preceding one which can be viewed in different forms (e.g., addition, specification, temporal); in (4) the underlined utterance is regarded as an elaboration on the utterance preceding væ. This utterance is about the status of the U.S dollar currency in Iran and the elaboration provided by the next utterance falls within the same topic.

væ does not necessarily elaborate on a preceding utterance. In some cases, væ with an elaborative function operates beyond the sentence level and at the discourse level. However, not all elaborative instances of væ can be seen as a single function since this includes some subfunctions, namely temporal, addition and specification. Yet, these subfunctions have been treated by some scholars as separate main functions of and (e.g., Halliday and Hasan 1976Halliday, Michael Alexander Kirkwood, and Ruqaiya Hasan 1976 Cohesion in English. London: Longman.; Crible 2017). As Examples (5)–(9) show, all elaborative functions carried out by væ can be classified under one of these subfunctions as follows:

4.1.1Addition

This is the prototypical function of væ that signals a simple addition to a previous utterance of information within the same topic. As (5) shows, the underlined utterance is believed to add some information to the preceding utterance. The utterance dæst tænhɑ hæstæm (‘I have no assistant’) adds information regarding the status of mæn kɑr dɑræm (‘I am busy’).

(5)

væqti

when

dærbɑreye

about-ez

æli

Ali

porsid-æm(.)

ask.pst-1sg (.)

goft(.)

say.pst.3sg (.)

mæn

I

kɑr

work

dɑr-æm

have.prs-1sg

væ

and

dæst

hand

tænhɑ

alone

hæst-æm.

be.prs-1sg

When I asked about Ali (.) he said (.): “I am busy and have no assistant”. (spoken data)

4.1.2Specification

The specification function of væ resembles that of addition to some extent, particularly when it does not co-occur with specification DMs such as beviʒe (‘especially’) and mæsælan (‘for example’). As a specificational marker, væ provides more details and examples in the segment following it. It applies when its following utterance describes the early situation in detail. Information stated in this utterance falls within the scope of the prior one. In (6) below, the specification reading of væ specifies one person (Javad) among many others. This specification reading utters in the process of pragmatic interpretation if there is an implied question (see Onea and Volodina 2011Onea, Edgar, and Anna Volodina 2011 “Between Specification and Explanation: About a German Discourse Particle.” International Review of Pragmatics 3 (1): 3–32. ). In (6), we also observe that when the speaker introduces the proposition that “many people came and went” into the common ground, gives rise to an implied question, “who exactly posed the problem?”. If such a question does not arise in the discourse, or if the speaker cannot assume that this question is present in the discourse, we may not see specificational væ in use.

(6)

xeili-ya

many.pl

umæd-æn

come.pst-3pl

o

and

ræft-æn

go.pst-3pl

væ

and

Jævɑd

Javad

bud-e

be.pst.3sg

ke

that

hæmiše

always

moškel

problem

dorost

correct

kærd-e

do.pst-3sg

Many people came and went, and it was Javad who always caused the problem. (spoken data)

4.1.3Temporal

Elaborative sub-functions are quite similar to each other in such a way that making a distinction among them requires meticulous attention to available distinctive cues. Apropos of the temporal function, the most important cue to make the distinction is the specific stress carried by væ concerning a chronological order. Without this distinctive cue, one may not be able to discern the correct type of elaboration. As can be induced from (7), væ signals slight chronological stress on the order of the preceding and following segments. In this example, it is impossible to replace the utterance preceding væ with the one following it since chronological order would be halted in a real sense and, consequently, the sentence would be pragmatically inappropriate. In (7), it is quite clear that the action of using the mobile camera is a prerequisite for seeing digital content on the mobile screen. Cases like (7), in which changing the order of clauses would lead to a change in the interpretation of the sentence, have been called the asymmetric use of væ (see Sweetser 1990Sweetser, Eve 1990 From Etymology to Pragmatics: Metaphorical and Cultural Aspects of Semantic Structure. Cambridge: Cambridge University Press. ).

(7)

be

to

ʤɑy-e

place-ez

ɑn

that

bærčæsb-ha,

label-pl

mɑikrosɑft

Microsoft

æz

from

yek

one

durbin-e

camera-ez

telephone-hamrɑh

phone-accompany

bærɑye

for

eskæn

scan

kærdæn-e

do.inf-ez

bɑrkod

barcode

estefɑde

use

mi-kon-æd

ipf-do.prs-3sg

væ

and

mohtævɑ-ye

content-ez

diʤitɑl

digital

ru-ye

on-ez

telephone-hamrah

phone-accompany

næmɑyeš

show

dɑde

give.pst.prf

mi-šæv-æd.

ipf-get.prs-3sg

Microsoft uses the mobile camera instead of those labels in order to scan the barcode and then the digital content is shown on the mobile screen. (written data)

4.2Contrastive function

As mentioned above, it is difficult to draw a clear-cut distinction between addition and specification or addition and temporal. The same is true about contrast and concession. However, a distinction between contrast and concession is deemed to be inevitable, and we make such a distinction whenever it is necessary, despite the fact that they are highly related to each other and that the distinction will be challenging. In the contrastive function, we merely compare two things in dissidence, while in a concession-based one, the observed situation is contrary to expectations. Furthermore, there is some counter-expectation in concessions, and the implied causal relation is canceled (see Webber et al. 2019).

In Example (8), væ functions as a contrastive marker since the segment following it is in contrast with the one preceding it; it signals a direct contrast between the preceding segment and the following one. In this example, the speaker believes everyone is making progress, however, her brother is doing worse than before.

As noted above, the concession is conveyed when a causal relation based on one argument is canceled or denied. Correspondingly, Example (9) resembles a similar situation, which can be interpreted as follows: ‘because I was sick, I was supposed not to come to your wedding party, and yet I came’. Here væ signals the segment following it as well as the fact that is the result of a canceled causal relation.

4.3Inferential function

væ can also be used as an inferential marker. For example, in (10a) the utterance preceding væ is considered to be a reason for the one following it, while the following utterance is regarded as a result of the one before it. In Example (10a) people prefer Ash (‘soup’) and this preference is because of the cold weather:

Akin to the other functions noted before, here the change in the order of the present segments occurring before and after væ is almost impossible. It stems from the fact that the segment following væ is regarded as a result of the segment preceding it; hence the result is not placed before the cause. By contrast, (10b) can be regarded as a typical example of the asymmetric use of væ. Although (10b) seems syntactically correct, it is semantically peculiar since the result has been placed before the cause, and the relationship between the two segments (following and preceding væ) is nonsensical.

4.4Alternative function

This function has been regarded as an elaborative function for a long time (cf. Fraser 2009Fraser, Bruce 2009 “An Account of Discourse Markers.” International Review of Pragmatics 1 (2): 293–320. ; Kassaei and Amouzadeh 2020Kassaei, Gholamreza, and Mohammad Amouzadeh 2020 “The Combination of Discourse Markers in Persian.” International Review of Pragmatics 12 (1): 135–163. ). It is here recognized as a single function, and not an elaborative sub-function. As the name of this function suggests, here væ is adopted to present alternatives. Unlike functions mentioned above, the utterances placed before and after væ can be used interchangeably with no alteration to the unmarked meaning. This may affect the level of emphasis or stress, as the first utterance is generally of primary importance. As is understood from (11a), the utterance following væ is an alternative to the previous one, both of them are underlined, and this does not present any elaboration, but only an alternation. In other words, væ in (11a) can be replaced by either væ ya (‘and or’) or ya (‘or’). (11b) can display the palpable interchangeability of utterances in this function, which is quite impossible in previous ones.

4.5Topic-shifting function

The functions of væ are not merely limited to LDM ones. In some cases, væ can also signal a topic PDM function. As Example (12) illustrates, while speaker (A) is talking about her life in Turkey, the second speaker shifts the topic from her life to her brother’s. væ is used as a topic-shifter to change/start a new topic, where none of the other functions applies. Here, in contrast to the earlier functions denoting connectivity, væ makes a break with the previous utterance and introduces a new topic.

5.1Simultaneous multifunctionality of væ

In addition to the remarks above, there are different LDMs classified by their functions. The question then is whether one LDM can fulfill other functions. This leads us to the notion of multifunctionality, a perpetual feature of DMs; however, this important concept did not receive due attention by Fraser (forthcoming)Fraser, Bruce forthcoming. “Canonical Sequences of Discourse Markers in English.”.

We view multifunctionality through the lens of meaning potential (see Norén and Linell 2007Norén, Kerstin, and Per Linell 2007 “Meaning Potentials and the Interaction between Lexis and Contexts: An Empirical Substantiation.” Pragmatics 17 (3): 387–416. ). Thus, we do not take a maximalist semantic approach to væ, signifying that væ does not characterize different stable lexical meanings performing different functions. However, according to the theory of meaning potential, væ is treated as a potentially polysemous word whose meanings are determined by the interface between meaning potential and contextual factors (e.g., co-text and situational conditions). In other words, the multiple meanings of væ are not considered to be its constant features.

It must be noted that væ as a multifunctional DM can have additional, specific, temporal, inferential, contrastive, alternative, and topic-shifting uses in different contexts. Its functions can be grouped into three domains of discourse represented in Figure 2. These domains were inspired mainly from Halliday and Hasan (1976)Halliday, Michael Alexander Kirkwood, and Ruqaiya Hasan 1976 Cohesion in English. London: Longman., Redeker (1990)Redeker, Gisela 1990 “Ideational and Pragmatic Markers of Discourse Structure.” Journal of Pragmatics 14 (3): 367–381. , Sweetser (1990)Sweetser, Eve 1990 From Etymology to Pragmatics: Metaphorical and Cultural Aspects of Semantic Structure. Cambridge: Cambridge University Press. , and Gonzalez (2005)González, Montserrat 2005 “Pragmatic Markers and Discourse Coherence Relations in English and Catalan Oral Narrative.” Discourse Studies 7 (1): 53–86. ; then they were revised and redefined by Crible (2017bCrible, Ludivine 2017b “Towards an Operational Category of Discourse Marker: A Definition and Its Model.” In Pragmatic Markers, Discourse Markers and Modal Particles, ed. by Chiara Fedriani, and Andrea Sansó, 99–124. Amsterdam: John Benjamins. , 107):

1. ideational: discourse relations between real-world events;

2. rhetorical: discourse relations between epistemic, speech-act events and metadiscursive functions;

3. sequential: structuration of discourse segments, both for local management of small units and macro-level organization.

If we regard ‘addition’ as the core meaning of væ (see the previous section), we can present its functional scopes covering from its core sequential domain down to less pragmatic uses in the ideational domain (see Figure 2).

Now the question is whether væ can serve different functions simultaneously. A positive answer can be provided by the analysis of our data. The observed patterns of multifunctionality in our data are as follows:

i.Additional-temporal function

væ can designate the succession of events that helps discourse forward. It simultaneously adds more information to a preceding utterance. In other words, væ signals a sort of temporal relation between utterances, as well as adding more information within the same topic. In (13) below, not only does væ 77.One of the reviewers kindly pointed out that the second væ could also function as a consequential marker. This might be true if we assume that the killing of the leopard has led to her losing her cubs. show the sequential order of events (poisoning, killing, losing), but it also can signal addition to the current topic, which is the leopard.

(13)

do

two

hæfte

week

piš

before

yek

one

mɑde

female

pælæng

leopard

dær

in

hævɑli-ye

around-ez

rustɑy-i

village-indf

dær

in

hævɑli-ye

around-ez

Poldoxtær

Poldoxtar

tævæsot-e

by-ez

šekɑrči-ɑn

hunter-pl

mæsmum

poisonous

šod

get.pst.3sg

væ

and

be

to

qætl

kill

res-id

reach.pst-3sg

væ

and

tule-hɑ-ye

cub-pl-poss

in

this

pælæng

leopard

nɑpædid

disappear

šod

get.pst.3sg

`

 

Two weeks ago, a female leopard was poisoned and killed by hunters around a village in Poldoxtær; then her cubs got lost. (written data)

ii.Additional-contrastive function

A close analysis of the instances of addition indicates that, in some cases, væ proves not to be restricted to the addition function and signals contrast concurrently. For example, in (14), while væ signals addition, it also reflects a sense of contrast with the previous utterance. Despite the fact that people are putting effort to solve the problem, the problem has not been eradicated yet.

(14)

in

this

æfrɑd

people

besyɑr

very

tælɑš

attempt

mi-kon-ænd

ipf-do.prs-3pl

væ

and

be

to

sefr

zero

resid-æn-e

reach-inf-ez

in

this

mozu’

issue

kɑr-e

activity-ez

došvɑri

difficult

æst.

be.prs.3sg

These people try a lot, but removing this issue is difficult. (written data)

iii.Additional-inferential function

An inferential function can also simultaneously co-occur with the core meaning of væ. That is to say, væ not only signals additional information to a preceding utterance or discourse, but also signals the causal role of such an utterance that could be inferred from the following one. For example, væ in (15) below shows that the utterance following væ adds more information to the current message as well as indicating the causal relation between the two utterances. Here, the fact that galaxies get far from each other is added information within the same topic. It is also the result of the previous utterance. Such binary functional behavior of væ is indicative of its simultaneous multifunctionality.

(15)

enerži-ye

energy-ez

tɑrik

dark

be

to

enbesɑt-e

expansion-ez

ʤæhɑn

world

komæk

help

mi-kon-æd

ipf-do.prs-3sg

væ

and

kæhkešɑn-hɑ

galaxy-pl

æz

from

hæm

each other

dur

far

mi-šæv-ænd

ipf-get-3pl

Dark energy helps world expansion and galaxies get far from each other. (written data)

iv.Specific-contrastive function

All instances of the simultaneous multifunctionality of væ are not always subject to constraints of addition. In some cases, we can see other functions signaled by væ concurrently. For example, in (16) below, væ in specificaiton reading takes a certain number out of a total amount (i.e., 30 chairs out of 130 chairs). As already mentioned, this kind of pragmatic interpretation is an answer given to an implied question underlying the discourse. Here, the implied question is ‘how many seats did they expect to win?’. By focusing on (16), we observe that væ simultaneously establishes a contrastive relation between the following and preceding utterances. This sense of contrast can be felt in a way that the political movement nominated one hundred and thirty candidates while their expectation was limited to only thirty seats.

(16)

in

this

ʤæriyɑn

stream

sæd

hundred

o

and

si

thirty

kɑndid-ɑ

candidate-pl

rɑ

objm

mo’ærefi

nomination

kærd

do.pst.3sg

væ

and

tænhɑ

only

be

to

si

thirty

dærsæd-e

percent-ez

korsi-hɑ

seat-pl

češm

eye

duxte

sew.ptcp

bud

be.pst.3sg

This (political) movement nominated one hundred and thirty candidates, but they only expected to win thirty seats. (written data)

The presence of multifunctionality in væ was delineated above. Interestingly enough, the two simultaneous functions of it come from different domains of discourse. As illustrated in Figure 3, at least two functional domains are required to be involved. In other words, it seems that functions from the same domain would not be simultaneously present.

Figure 3.Domains of simultaneous functions

5.2væ co-occurrences

As noted earlier, Fraser (forthcoming)Fraser, Bruce forthcoming. “Canonical Sequences of Discourse Markers in English.” provided a list of principles corresponding to DM co-occurrences. Here our aim is to revisit the ones pertinent to væ co-occurrences and to provide a detailed overview of its co-occurrences with other DMs. To the best of our knowledge and based on analyzed data, the general tendency towards væ co-occurrences can be listed as follows:

1. It can co-occur with all the secondary DMs from the same class. They can be additive, specific, or temporal DMs. For instance, væ in (17) co-occurs with an additive DM thereby the information, which is ‘this winning could put our mind at ease’, is added to the results of Sepahan (an Iranian football club) win.

   (17)

   A:

   sepɑhɑn

   Sepahan

   bɑ

   with

   in

   this

   bord

   winning

   43

   43

   emtyɑz-i

   score-inf

   šod

   get.pst.3sg

   o

   and

   um-æd

   come.prs-3sg

   sædr-e

   top-ez

   ʤædvæl.

   table

   B:

   væ

   and

   hæmčenin

   also

   xyal-e

   mind-poss

   mɑ

   we

   rɑ

   objm

   hæm

   also

   rɑhæt

   easy

   kærd.

   do.pst.3sg

   A:

   After winning this match, Sepahan came out at the top of the table with 43 scores.

   B:

   And also it put our mind at ease. (spoken data)

   However, væ may seem redundant when it co-occurs with other DMs that are nearly synonymous (e.g., væ hæmčenin ‘and also’); this is almost identical to the mechanism proposed by Oates (2000)Oates, Sarah Louise 2000 “Multiple Discourse Marker Occurrence: Creating Hierarchical for Natural Language.” In Procedding of the 3rd CLUK Colloquium, 41–45. Brighton., who pointed out that the impression of redundancy has not been made in this case. This stems from the fact that væ is bleached semantically in these co-occurrences and its potential meaning is directed by the DM, which follows or precedes it. The central intriguing fact about these co-occurrences is that væ is bleached to the extent that it can be deleted without creating any flaw in discourse flow. Another point concerns the fact that deleting the second DM will not lead to any discourse gap if væ is used as a single DM either.

2. As can be seen in Table 3, væ may co-occur with almost all secondary LDMs. These co-occurrences can function either following or preceding væ.

   Table 3. væ DM co-occurrences with Persian SLDMs

   væ	Elaborative DMs	væ	Contrastive DMs	væ	Inferential DMs	væ	Alternative DMs
   ✓	hæmčenin (also)	✓	væli (but)	✓	bænɑbærin (therefore)	✓	yɑ (or)
   ✓	be viẑe (especially)	✓	æz suj-e digær (on the other hand)	✓	lezɑ (thus)	✓	yɑ inke (or that)
   ✓	mæsælæn (for example)	✓	æz tæræf-e digær (on the other hand)	✓	dær nætiʤe (therefore)
   ✓	mæxsusæn (particularly)	✓	dær moqɑbel (by contrast)	✓	be hæmin ellat (because of this)
   ✓	xosusæn (particularly)	✓	bɑ in hɑl (nevertheless)	✓	be hæmin dælil (because of this)
   ✓	bexosus (particularly)	✓	bɑ in voʤud (nonetheless)	✓	æz in ru (thus)
   ✓	be ælɑve (in addition)	✓	bɑ voʤud-e in (in spite of this)	✓	æz in su (because of this)
   ✓	be ezɑfe (in addition)	✓	bær æks (on the contrary)	✓	natiʤætæn (therefore)
   ✓	æ’lɑve bær in (moreover)			✓	be hæmin xɑter (because of this)
   				✓	æz in xɑter (because of this)

3. As noted above, væ can occur either before or after SELDM. However, Fraser (forthcoming)Fraser, Bruce forthcoming. “Canonical Sequences of Discourse Markers in English.”, and Oates (2000)Oates, Sarah Louise 2000 “Multiple Discourse Marker Occurrence: Creating Hierarchical for Natural Language.” In Procedding of the 3rd CLUK Colloquium, 41–45. Brighton., in English, assume that PLDMs are followed by SLDMs and that a reverse sequence is impossible. In terms of the linear order of markers, they argue that weak markers (e.g., væ ‘and’, æmmɑ ‘but’, and pæs ‘so’) are followed by strong makers (hæmčenin ‘also’, væli ‘but’, and bænɑbærin ‘thus’) while in Persian this order is not perfectly genuine. In this connection, it can be argued that in Persian not only væ but also æmmɑ (‘but’) and pæs (‘so’), as PLDMs, can be either preceded or followed by SLDMs. This possibility can be observed in Examples (18a) and (18b) below, where væ and hæmčenin can be used interchangeably.88.This might have different reasons, but Persian’s flexible word order and free structuring constituent (see Faghiri and Samvelian 2020Faghiri, Pegah, and Pollet Samvelian 2020 “Word Order Preferences and the Effect of Phrasal Length in SOV Languages: Evidence from Sentence Production in Persian.” Glossa: A Journal of General Linguistics 5 (1): 86. , 8) seem a compelling reason for the mobility of DMs in sequences. Here we merely intend to show the possibility of change in the order of DMs, but it should be noted that it is certainly the case that, in some sequences, changes in position can lead to different pragmatic meanings.

   (18)

   1. hæmčenin

      also

      væ

      and

      be

      to

      gozɑreš-e

      report-ez

      isna

      isna

      moʤtæbɑ

      Mojtaba

      šærifi

      Sharifi

      æz

      from

      dæ’væt-e

      invitation-ez

      æli

      Ali

      dɑyi

      Dai

      væ

      and

      sɑdeq

      Sadeq

      dorudgær

      Dorudgar

      xæbær

      news

      dɑd

      give.pst.3sg
   2. væ

      and

      hæmčenin

      also

      be

      to

      gozɑreš-e

      report-ez

      isna

      isna

      moʤtæbɑ

      Mojtaba

      šærifi

      Sharifi

      æz

      from

      dæ’væte

      invitation.ez

      æli

      Ali

      dɑyi

      Dai

      væ

      and

      sɑdeq

      Sadeq

      dorudgær

      Dorudgar

      xæbær

      news

      dɑd.

      give.pst.3sg

      And also, based on the report provided by ISNA, Mojtaba Sharifi announced that Ali Dayi and Sadeq Droudgar would be invited. (written data)

4. Fraser (forthcoming)Fraser, Bruce forthcoming. “Canonical Sequences of Discourse Markers in English.” argues that in sequences of PLDMS, particularly those involving and, only the sequence of and so is highly probable, but he controverts other alternative forms of co-occurrence. However, our examination of Persian data reveals that væ can also occur with all Persian PLDMs, which can be either in the form of following or preceding PLDMs. Examples (19a) and (19b) below are typical of the PELDM and PCLDM case of co-occurrences, where væ is following the PCLDM (19a) and is followed by that (19b). Note that the two DMs in (19a) do not form a single unit in terms of combination like in (19b). In other words, the co-occurrence of æmma væ (‘but and’) in (19a) does not constitute a combined discourse marker conveying a similar function.

   (19)

   1. dæqiqæn

      exactly

      ne-mi-dun-am

      neg-ipf-know-1sg

      če

      what

      zæmɑn-i

      time-indf

      rɑ

      objm

      ruy-e

      on-ez

      dæstgɑh

      system

      tɑnzim

      setting

      kærd

      do.pst.3sg

      æmmɑ

      but

      væ

      and

      be

      to

      mæn

      me

      goft

      say.pst.3sg

      ke

      that

      qæsd

      intention

      dɑšte

      have.pst.3sg.prf

      ɑb

      water

      be

      to

      dʒuš

      boiling

      beres-e.

      reach.prs-3sg
   2. dæqiqæn

      exactly

      ne-mi-dun-am

      neg-ipf-know-1sg

      če

      what

      zæmɑn-i

      time-indf

      rɑ

      objm

      ruy-e

      on-ez

      dæstgɑh

      system

      tɑnzim

      setting

      kærd

      do.pst.3sg

      væ

      and

      æmmɑ

      but

      be

      to

      mæn

      me

      goft

      say.pst.3sg

      ke

      that

      qæsd

      intention

      dɑšte

      have.pst.3sg.prf

      ɑb

      water

      be

      to

      dʒuš

      boiling

      beres-e.

      reach.prs-3sg

      I don’t know exactly what time he set the device, but he told me that he wanted to boil the water. (spoken data)

It is important to note that these two DMs can undergo incoordination99.Here, the notion of incoordination for væ æmmɑ traces back to what Kuteva et al. (2017)Kuteva, Tania, Bernd Heine, Peter Austin, Seongha Rhee, Marine Vuilermet, and Domenico Niclot 2017 “The ‘Mirror’ of Insubordination.” Linguistics Departmental Seminar Series, SOAS University of London. Available at: https://​www​.youtube​.com​/watch​?v​=Jfq1KcoRens introduced as incoordinate sentences. In these sentences, the connectors ‘and’ and ‘but’ occur at the beginning of sentences while losing their original meaning and status as coordinative conjunctions. Kuteva et al. (2017)Kuteva, Tania, Bernd Heine, Peter Austin, Seongha Rhee, Marine Vuilermet, and Domenico Niclot 2017 “The ‘Mirror’ of Insubordination.” Linguistics Departmental Seminar Series, SOAS University of London. Available at: https://​www​.youtube​.com​/watch​?v​=Jfq1KcoRens believe that when these conjunctions go through the incoordination process, they often transform into sentence particles with mirative values, namely, sudden discovery, surprise, unprepared mind, counter expectation, new information (for details on the category of mirativity, see Aikhenvald 2012Aikhenvald, Alexandra Y. 2012 “The Essence of Mirativity.” Linguistic Typology 16 (3): 435–485. , 437). in some cases, when they function as a topic-shifting, turn-shifting, or turn-taking DM. As it can be gathered from (20), væ æmmɑ (‘and but’) is employed to take a turn. This function can be seen in roundtable discussions by moderators. Here, two guests are discussing with each other, but the moderator halts the discussion to take the turn and deliver it to another speaker.

This incoordination can be used in terms of topic shifting as well. As (21a) below shows, this co-occurrence acts as a device to change the topic. Speaker C has two pieces of news, a good and a bad one. First she tells the bad one, which provokes hearers’ (speaker A and B’s) emotional reactions. To change the topic and lighten the hearers’ somber mood, she benefits from the topic shifting nature of væ æmmɑ.

Moreover, an interesting fact about this co-occurrence with regard to the different functions it undergoes is that its linearization order can be flexible; that is, the function remains constant by the omission of væ or æmmɑ (‘but’). This possibility can be seen in Examples (21b) and (21c) below.

Another mode of PLDM co-occurrence would be PELDM + PILDM. As was the case with the previous one, in this sequence, væ can be followed or preceded. However, the occurrence of væ in a preceding position is by far more frequent. In Examples (22a) and (22b) below, although the original order belongs to the first one, the second one is possible as well.

An intriguing feature of PELDM + PILDM, not the reverse order, is that they can be nonadjacent and signal the same function, but they are mainly used in an informal context as illustrated in Example (22c) below.

With regard to the last sequence of væ with PLDM, note that it co-occurs with PALDM yɑ (‘or’). As Examples (23a) and (23b) show, the two above-mentioned features of PLDMs (linear order flexibility and optionality) can be seen in the case of the last sequence.

One should note that the results of corpus-based studies (see Kassaei and Amouzadeh 2020Kassaei, Gholamreza, and Mohammad Amouzadeh 2020 “The Combination of Discourse Markers in Persian.” International Review of Pragmatics 12 (1): 135–163. ) on sequences of væ with other PLDMs indicate that conventional patterns, in which vae occurs earlier, are predominant. This might be due to the weak and ambiguous nature of væ; the weaker the DM for a given relation, the more it will be compensated by other DMs (Crible 2020Crible, Ludivine 2020 “Weak and Strong Discourse Markers in Speech, Chat, and Writing: Do Signals Compensate for Ambiguity in Explicit Relations?” Discourse Processes 57 (9): 793–807. ).

DM co-occurrences do not always merely consist of two parts. There are instances of three-DM co-occurrences as well. As noted earlier, Fraser (forthcoming)Fraser, Bruce forthcoming. “Canonical Sequences of Discourse Markers in English.” assumed that PLDMs could be followed, but not preceded, by two SLDMs from the same class. However, væ in multi-part co-occurrences acts quite differently. It can be followed, preceded, or even be placed between DMs; such a possibility is evident below.1111.The point that should be raised here is that although all these three co-occurrences signal a single function, which is addition, there is a kind of variation in their prosodic structure when each combination is uttered. All instances below (a-f) seem theoretically possible in Persian and can be substituted with væ hæmčenin be’ælɑve (‘and also in addition’) in (24).

1. væ hæmčenin be’ælɑve (and also in addition)

2. væ be’ælɑve hæmčenin (and in addition also)

3. hæmčenin væ be’ælɑve (also and in addition)

4. be’ælɑve væ hæmčenin (in addition and also)

5. hæmčenin be’ælɑve væ (also in addition and)

6. be’ælɑve hæmčenin væ (in addition also and)

With respect to multi-part co-occurrences, not only does væ co-occur with SELDMs, but it also might co-occur with heterogeneous SLDMs. This can be observed in different instances of (25a, b, c) below. Sequences of væ with other SLDMs bear a close resemblance to the linearization order that it follows in contiguity with SELDMs. To state it explicitly, væ might occupy all the three different positions in multi-part sequences with SLDMs.1212.Although all multi-part instances of væ co-occurrences might bear the same meaning, diverse pragmatic functions of væ make it necessary to note that we cannot afford to overlook the possibility of change in meaning in some cases, when it shifts from one position to the other one. Note that (25b) is the original example from our data and the two others show the mobility of væ when it co-occurs. Here, the speaker is worried as he and his colleagues quarreled with their boss, but on the other hand, he is happy that they could finally voice their concerns. The multi-part sequence of ‘væli væ æz tæræfi’ was used to demonstrate this contrast.

As regards the motivations underlying væ co-occurrences with other DMs, some of the proposals made by earlier studies about DMs (see Section 2.2) appear to be fairly reasonable. The first underlying motivation, which was explained above, is known as functional specification. It suggests that in sequences of DMs, one DM, usually the stronger one, specifies the function of one with which it co-occurs (see Oates 2000Oates, Sarah Louise 2000 “Multiple Discourse Marker Occurrence: Creating Hierarchical for Natural Language.” In Procedding of the 3rd CLUK Colloquium, 41–45. Brighton.; Haselow 2019Haselow, Alexander 2019 “Discourse Marker Sequences: Insights into the Serial Order of Communicative Tasks in Real-time Turn Production.” Journal of Pragmatics 146: 1–18. ). Sequences like væ sepæs (‘and then’), can be explained as being motivated by restricting and specifying the vague meaning of væ. One could note that such motivation also confirms Gricean view of production and comprehension processes1313.On the basis of the Gricean model, speakers/authors try to be as informative as required and produce informative utterances, such that the listeners or readers can understand the message (Grice 1975Grice, Herbert P. 1975 “Logic and Conversation.” In Syntax and Semantics Vol. 3: Speech Acts, ed. by Peter Cole, and Jerry L. Morgan, 41–58. Brill.). Regarding his maxim of quantity, DM occurrences are subject to ambiguity and redundancy. To avoid redundancy, and as high-informative DMs can achieve an adequate level of interpretation, there is no need for another DM. In contrast, low-informative and ambiguous DMs such as væ are expected to be compensated by other DMs to be more transparent. (see Crible 2020Crible, Ludivine 2020 “Weak and Strong Discourse Markers in Speech, Chat, and Writing: Do Signals Compensate for Ambiguity in Explicit Relations?” Discourse Processes 57 (9): 793–807. ). Alternatively stated, væ is of little informative value and does need to be enriched by other signals, which expedite the process of comprehension in discourse. The presence of compensating signals like sepæs (‘then’) might be mostly established when they co-occur with weak DMs, rather than with stronger ones. This potential factor is consistent with the Uniform Information Density Hypothesis (Levy and Jaeger 2007Levy, Roger, and T. Florian Jaeger 2007 “Speakers Optimize Information Density through Syntactic Reduction.” In Advances in Neural Information Processing Systems (NIPS), ed. by Bernhard Schölkopf, John Platt, and Thomas Hofmann, 849–856. MA: MIT Press.). According to this hypothesis, when an accurate interpretation can be plausibly deduced from one part of a sentence, the need for extra markers will be obviated.

In addition to functional specification, floor-holding can be viewed as another motivation underlying DM co-occurrences. From this point of view, væ is indicative of one of the disfluencies occurring in online communication, as it demands instant cognitive processing and interlocutors’ undivided attention (see Bortfeld et al. 2001Bortfeld, Heather, Silvia D. Leon, Jonathan E. Bloom, Michael F. Schober, and Susan E. Brennan 2001 “Disfluency Rates in Conversation: Effects of Age, Relationship, Topic, Role, and Gender.” Language and Speech 44 (2): 123–147. ; De Klerk 2005De Klerk, Vivian 2005 “Procedural Meanings of ‘Well’ in a Corpus of Xhosa English.” Journal of Pragmatics 37 (8): 1183–1205. ). In this contiguity, væ serves to bridge the gap during a cognitive planning pause. For example, in (26) below, væ within a lengthier span of production is used to increase the processing time of comprehension, and it is plainly void of semantic meaning. Here, the second speaker is thinking about the first speaker’s ideas about new philosophical concepts, correspondingly væ is lengthened to fulfill this goal.

Cognitive planning is not the only reason for floor holding motivations. In some cases, the speaker’s cognitive orientation in discourse processing accompanies some strategies of online communication, such as turn-holding. As it is understood from (27), not only does speaker (A) use væ to buy time for his cognitive process, but he also attempts to keep the floor by repetition and sudden stress via a high pitch on the first væ. In (27), the second speaker wants to start vindicating himself immediately after hearing the first speaker’s accusatory remarks, but the first speaker does not let him by repetitive use of væ.