Inferential Processes in English and the Question whether English has Modal Particles

Abstract In this paper, we ask whether English pragmatic markers may evoke similar inferential processes in discourse as German modal particles, studying alright/all right, already and then in more detail. Moreover, we investigate whether specific formal features are associated with these uses and thus whether there is any evidence for a productive modal particle category that can serve as a guideline for the creation and interpretation of modal particle uses of English pragmatic markers. Our analysis shows that even though evidence for a schematic modal particle construction is not conclusive, modal particle uses of pragmatic markers may be potentially widespread in English, and the inferential processes involved may be similar across languages.


Introduction
Various suggestions have been made in the literature concerning the equivalents of modal particles in English, and more recently, it has been suggested that English may be developing a modal particle category itself (e.g. Haselow 2013). Indeed, it seems strange that modal particles should play such an important role in some languages and not at all in English, especially if viewed from the perspective of the inferences made in interaction. That is, if modal particles involve certain inferential processes in interaction in some languages, then similar processes might arise in English as well. Nevertheless, the widely held belief is that English does not have modal particles, or if so, then much less frequently; for instance, Fillmore (1984: 133-134) suggests that speech formulas like after all are translation equivalents, but since they are not syntactically integrated, they occur in much lower frequencies than German modal particles.1 Furthermore, such a category does not appear in standard reference grammars such as Quirk et al. (1985), Biber et al. (1999), Huddleston and Pullum (2002) in the relevant sense.
However, recent attention to the pragmatics of argumentative structure and of dialogal interaction has suggested that a modal particle category may indeed be developing, starting in the 19 th century (Lenker In order to determine whether English pragmatic markers fulfill functions similar to those suggested for German modal particles (Diewald & Fischer 1998;Fischer 2000Fischer , 2007Diewald 2006; in press), we will first discuss utterance-final then (cf. Haselow 2011Haselow , 2012Haselow , 2013. We will then study certain utterancefinal uses of already, which have emerged recently and so far only in American English. Given that the modal particle use of already is such a recent (and regionally restricted) phenomenon, it will provide indirect evidence concerning the question whether there is an emerging modal particle category in English. The third example we will discuss is alright in order to address to what extent final particles in English have a contradictory flavor to them, as, for instance,  suggests. Finally, in order to investigate to what extent the phenomenon is restricted to final particles in English, we briefly discuss some other possible examples, namely after all, indeed and just.
Our analysis will thus address the following questions: 1. May English pragmatic markers fulfill the functions of German modal particles? That is, can English pragmatic markers be used in similar functions as German modal particles at all, i.e. are there uses of pragmatic markers that can be analyzed in similar ways as German modal particle? 2. How common are such uses? I.e., is it only a small group of specific pragmatic markers that fulfills functions similar to German modal particles, or does the phenomenon concern a larger group of items? Are such uses of English pragmatic markers rather exceptional, or do they constitute a considerable proportion of the polysemous pragmatic marker tokens? 3. Are there specific formal features associated with these uses? I.e. is the potential group of modal particle candidates in English restricted to specific positions (for instance, utterance-finally) or are they spread out over many different positions? 4. Is there any evidence for a productive modal particle category that can serve as a guideline for the creation and interpretation of modal particle uses of English pragmatic markers? That is, how are the type and token frequencies of these uses distributed?
Before we can address these questions, we need to define modal particles and specify their meanings and functions.

Inferential Processes in German Modal Particles
For German, the language we focus on here as a starting point for the comparison with English, only few features of modal particles are accepted unanimously; these include (see, for instance, Helbig 1988;Thurmair 1989Thurmair , 2013: 1. Modal particles are non-inflecting. 2. They have scope over the entire clause. 3. They cannot function as sentence constituents. 4. With very few exceptions, they occur inside the clause in a particular sentence-medial position, the so-called syntactic middle field. 5. The occurrence of most modal particles is restricted to certain sentence types/moods.

The Functions of German Modal Particles
Much work on modal particles focuses on the pragmatic effects modal particles can have in the different sentence types in which they occur (cf., for instance, Kwon 2005;Thurmair 2013;Weydt et al. 1983). Because their pragmatic interpretations may differ according to the sentence types in which they occur, they are often conceived of as illocutionary operators (e.g. Altmann 1987;Helbig 1988;Thurmair 1989Thurmair , 2013Detges & Waltereit 2009), where they are taken to modify or even specify the utterance's speech act meaning. For example, nur (roughly: 'only') can turn a directive into an encouragement, permission or threat (Zifonun et al. 1997(Zifonun et al. : 1218. Furthermore, modal particles are often taken to have different readings depending on the sentence types in which they occur; for instance, Weydt et al. (1983: 122) assume the modal particle ja to have four different functions: -in exclamations, it may express marvel and astonishment, for instance, Du hast ja ein neues Auto! ('You have PTC a new car!') -in imperatives, it can be used as a warning or threat, for instance, Mach das ja nicht noch mal! ('Don't you dare PTC do this again!') -in assertions, it appeals to common knowledge, for instance, Du weißt ja, dass ich morgen Geburtstag habe ('You know PTC that I have a new car.') -in responses in combination with nicht ('not'), it enforces the rejection, for instance in ja nicht (lit. 'yes not').
Irrespective of whether modal particles are best thought of as illocutionary operators or not, their functions thus clearly interact with the sentence types in which they occur (cf. Alm et al. in press), and some approaches focus on the pragmatic features of modal particles in context, such as to express marvel or astonishment, a warning or threat or to appeal to common knowledge, as Weydt et al. (1983: 122) suggested for ja, for instance. Other approaches focus on more general functions of modal particles; one such function that has commonly been proposed for German modal particles is to connect the current utterance to common ground. This function is essential to the models by Lindner (1991), Haselow (2011) and Bross (2012), to name but a few, and it is at the heart of the model we are going to take as a starting point in this investigation (e.g. Diewald & Fischer 1998;Diewald 2006Diewald , 2015Fischer 2000Fischer , 2007Alm et al. in press). It is also at the core of models that assume polyphony, where the voice taken to be evoked by the modal particle corresponds to the pragmatically presupposed proposition suggested to be part of common ground (Bross 2012). In accordance with this previous work, in the model we build on here, modal particles are taken to refer to given but not usually verbalized3 propositions, which are suggested to be shared, and thus to mark the utterances in which they occur as non-initial, i.e. as related to, and coherent with, previous discourse. The function we are going to look for in English is thus the function to connect the current utterance to an aspect of common ground (a pragmatically given unit). Thus, instances of modal particles give rise to a paraphrase in terms of common ground, for instance, as you and I both know; as we have discussed before; as we have been thinking all along; as is generally assumed; as we both can see; one might have expected that…, but, depending on what kind of common ground is alluded to (shared perception, shared cultural background, shared dialog history, etc., cf. Clark 1996: 100-120) and depending on whether the modal particle affirms or rejects the pragmatically given unit. This proposed function of modal particles accounts for the fact that modal particles are inherently dialogical and significantly related to dialogical text types (cf. Diewald 2015). Diewald and Fischer (1998) present a basic semantic schema that suggests a three-part relationship; in particular, the schema comprises 1) a pragmatically given unit, a non-verbalized proposition, which is in most cases a logical variant of the host utterance4 (Foolen 1989); 2) the speaker's stance towards this pragmatically given unit (the current situation, i.e. what the speaker wants to express); and 3) the host utterance with the modal particle (the current utterance):5 Basic Schema for the Description of Modal Particles pragmatically given unit: proposition at hand (logical variant of current utterance) current situation: speaker evaluation of proposition at hand → utterance: modified proposition with particle The relationship between the pragmatically given proposition and the speaker's attitude towards this proposition, i.e. what the speaker wants to express, determines (speaker's perspective) the choice of the modal particle, and in turn the choice of modal particle in the current utterance determines the pragmatically given unit (hearer's perspective). For instance, some modal particles evoke a relationship of agreement with the pragmatically given unit, in which case the host utterance is identical with the pragmatically given unit, which is just re-iterated and thus strengthened by the modal particle. Unstressed ja (roughly: 'yes') is such a particle. In contrast, if the speaker disagrees with the pragmatically given unit, i.e. a proposition taken to be common ground, the pragmatically given unit is the negation of the host utterance, for instance, when the modal particle is aber (roughly: 'but'). 'Pragmatically given' does not mean that the speakers have to have agreed on it or that it has been explicitly discussed and is part of the discourse record; instead, by using the modal particle, the speaker suggests that the proposition evoked is shared.
The modal particle thus presents the current utterance as non-initial, such that it is oriented to a proposition that is presented as shared, but since the current utterance with the modal particle specifies exactly what this proposition is, it does not really matter whether this proposition is discourse-old, hearerold or even activated by the hearer; the current proposal is thus neutral with respect to the specific model of common ground assumed (we are using Clark's (1996) typology of common ground because it is the most comprehensive). Moreover, since the proposition indexed is retrievable from the host utterance with the particle (it's a logical variant of the host utterance), it can also be imposed onto the hearer, who is then suggested to have shared the information (cf. Alm & Larsen 2015).6 The pragmatically given units comprise cultural facts, assumptions about human nature, shared perception, as well as the discourse record and all the other aspects of common ground listed in Clark (1996). Alm & Larsen (2015) show that the kinds of propositions evoked by modal particles may differ based on the respective communication partner and topic of conversation; for instance, in child-directed speech, modal particles evoke mostly perceptually given information, whereas when teenagers discuss music styles, the most commonly evoked common ground concerns cultural facts.
By providing a schema for describing the pragmatically given unit as a "logical variant" (Foolen 1989) of the current utterance (for instance, its negation), the model provides a systematic method for retrieving the pragmatically given unit referred to, which is in contrast to other models that also rely on evoking aspects of common ground (e.g. Blakemore 2010;Haselow 2011;Bross 2012).
Our approach to identifying the pragmatically given unit is semantic; in particular, we assume that it is the meaning of the respective modal particle lexeme (Diewald & Fischer 1998;Fischer 2000;Diewald & Kresic 2010) and the grammatical meaning of the word class (Diewald 2006) that characterize the work modal particles do. The different pragmatic functions in the different speech acts are therefore taken to be epiphenomenal. For instance, regarding Weydt et al.'s (1983) four meanings of ja, they are all accounted for using the same model in our approach. Weydt et al. argue  That is, the modal particle links the current utterance to the (possibly perceptually available) common ground, reiterating what the participants can both see, and the fact that the current utterance is an exclamation causes the utterance to receive the interpretation of astonishment. That is, we take the meaning encoded in the modal particle construction to be only to evoke a pragmatically given unit to which the current utterance is anchored, while the sentence type construction with the given modal particle is responsible for encoding the pragmatic effects, including effects at the illocutionary level (see Alm et al. in press).
Similarly, we analyze the use of (stressed) ja in imperatives, where it can be used as a warning or threat, e.g. mach das ja nicht noch mal! 'don't you dare do this again!', using the same schema, where the difference in meaning is due to the difference in sentence type, which is part of the current situation: Pragmatically given unit You know as much as I do: don't do this again -take it into account. current situation: I demand: don't do this again.
We take the prosodic stress on ja to add to its meaning that not only the information is shared and agreed upon, but that the listener should take it into account (see also Abraham 2012: 39, who takes stressed ja to communicate a high commissive expectation). The imperative contributes the directive interpretation (I demand that you don't do this again). Also ja in assertions, for instance in Du weißt ja, dass ich morgen Geburtstag habe (you know that my birthday is tomorrow), is analyzed with the same schema: Pragmatically given unit we both know: you know that my birthday is tomorrow. current situation: you know that my birthday is tomorrow. → utterance Du weißt ja, dass ich morgen Geburtstag habe.
In our model, we thus analyze all of these examples in the same way, arguing that ja has the same meaning in all of them, namely to express agreement between the current situation and the pragmatically given aspect of common ground to which it refers. That the examples differ in pragmatic function, i.e. that they can convey astonishment or make a threat or an assertion, is the result of the modal particle occurring in exclamations, imperatives and declaratives respectively. These functions thus arise as contextual effects, which however differ from the grammatical function of the modal particle word class construction (see Diewald 2006). In our analyses, these differences are accounted for in the paraphrases of the illocutionary force of the host utterance, for instance I wonder for questions and I demand/I expect for directives (cf. Alm et al. in press).7 At the same time, the relation introduced by the respective particle (e.g. agreement in the case of ja) also contributes to the interpretation of the modal particle in its host utterance. In order to illustrate the contribution of each modal particle, we present an analysis of the sentence above, just with the modal particle aber; aber is a particle that evokes the negation of the current utterance: Pragmatically given unit at hand is: You don't have a new car. current situation: You have a new car. → utterance Du hast aber ein neues Auto 'you have PRT a new car'.
7 In a construction grammatical model (cf. Alm et al. in press), the interaction is specified in lexeme-specific sentence type constructions; that is, we distinguish here between the contribution of the modal particle itself, the modal particle word class construction and the sentence type constructions in order to account for these different paraphrases (for more detail, see Alm et al. in press).
Brought to you by | University Library of Southern Denmark -Syddansk Universitetsbibliotek Authenticated Download Date | 1/9/19 3:03 PM Such an utterance is plausible, for instance, in a situation in which the communication partner has just complained that nothing is changing in his life, thus behaving as if he did not have a new car. In the model presented, the grammatical function of the modal particle use is to anchor the current utterance to the shared background. Thus, even though both interactants know in principle that the partner has a new car, evoking this information from common ground serves to make it relevant for the current line of argumentation. For this reason, we have argued the main function of German modal particles to be to mark the current utterance as non-initial, i.e. as related to some pragmatically given proposition (cf. Diewald & Fischer 1998;Diewald 2006) and to anchor the current utterance to argumentative common ground (Fischer 2007). Now, before we set out to identify modal particle uses of a set of pragmatic markers in English, let us consider Haselow's recent suggestions concerning modal particles in English. He bases his analyses on the model proposed by Diewald & Fischer (1998) as well, not only by citing Diewald & Fischer (1998) and Diewald (1999Diewald ( , 2006Diewald ( , 2008 as the framework for his analysis (e.g. Haselow 2011: 3620), but also by using similar formulations concerning the functions of modal particles, such as "they imply the existence of a prior turn and thus specify the communicative context" (Haselow 2013: 389) and mark the unit they accompany as "non-initial" (Haselow 2013: 378). The question thus arises whether the issues addressed in the current study have already been sufficiently addressed by Haselow, rendering the current study superfluous. This is, however, not the case. First, Haselow's starting point is final particles in English, for which he aims to identify their functions, whereas our starting point is the identification of modal particle functions in English pragmatic markers, which does not restrict us to the final position. Second, even though Haselow claims to base his analyses on the Diewald/Fischer model, he does in fact not make use of its key tool for identifying the pragmatically given proposition to which the utterance containing the modal particle is anchored. That is, although the Diewald/Fischer model provides a systematic, mechanistic approach to identifying the pragmatically given unit to which the utterance containing a modal particle refers, Haselow does not apply the schema. While the results are often similar -it would in fact be a bad sign if intuitions and plausibility yielded different results than those afforded by the model proposed by Diewald & Fischer -they may lead to different interpretations. Third, Haselow (e.g. 2013: 387) distinguishes categorically between verbalized and non-verbalized pragmatically given units, counting only the latter as modal particles, which leads to different numbers of instances of modal particle interpretations (for detail see the discussion of then below). Consequently, a quantification of the modal particle readings of English pragmatic markers based on the current model is still missing. In the following we therefore shall have to see how widespread the phenomenon really is.

Corpus Study
In order to determine whether for English pragmatic markers similar inferential processes can be assumed as for German modal particles, a corpus study was carried out with the purpose to determine a) whether similar uses can be identified, b) how frequent these uses are, c) to what extent these uses are connected to certain formal characteristics, and d) to what extent these uses are restricted to particular sentence type constructions.

Modal Particle Candidates in English
The English pragmatic markers we investigate in this study are alright, already and then. Of these, only then has previously been suggested to function as a modal particle (cf. Haselow 2011), whereas the others are mentioned in König et al. (1990) as translation equivalents of German modal particles, yet they have not been investigated as such in their own right.

The Corpora
Three corpora (The British National Corpus, The Corpus of Contemporary American English, and The Corpus of American Soap Operas) were chosen for the present study to ensure a broad overview with regard to the usage patterns of the possible modal particles. Both the BNC and the COCA comprise many different registers and represent British and American English respectively. However, both corpora have a very limited amount of informal conversational data compared to the amount of written material. At the same time, modal particles are characteristic of spoken interaction (Diewald 2015). For this reason, the Corpus of American Soap Operas was included in the database. All three corpora can be searched via the same interface (offered by BYU) and display occurrences in a similar fashion, which increases the comparability of the results.
The British National Corpus (BNC) is a 100 million word corpus of British English, 10 percent of which are orthographic transcriptions of unscripted informal conversations and spoken language collected in various contexts. At the time of access, all data had been collected between the 1980s and 1993.
The Corpus of Contemporary American English (COCA) is a large and balanced corpus of American English, containing more than 450 million words of spoken discourse, fiction, popular magazines, newspapers, and academic texts. It includes 20 million words each year from 1990-2012 and is updated regularly (the most recent texts, at time of access, were from summer 2012). About 20% are spoken language data.
The Corpus of American Soap Operas (SOAP) contains 100 million words in more than 22,000 transcripts of ten American soap operas from between 2001 and 2012. According to the composers of the corpus, the language used in soap operas, though scripted, can be taken to be a good representation of naturally occurring spoken interaction: "Even though the dialogue in the soap operas is scripted, we believe that it provides very useful insight into informal, colloquial American speech, and that it complements other similar corpora. For example, there are many informal phrases and words that are much more common in this corpus of soap operas than in the spoken portion of COCA and the BNC" (SOAP, 2015).

Corpus Methods and Procedure
The current corpus study was carried out by retrieving a number of instances from the corpora and by annotating these instances for possible modal particle functions.

Search Parameters
Given that the potential modal particles have homonyms in other lexical categories, there was high risk that an unrestricted search would yield a very low percentage of instances with modal particle functions in the data. Since recent work (e.g. Haselow 2011Haselow , 2013 suspects that modal particle meanings occur only in sentence final position, to increase the likelihood of encountering these uses, we investigated the pragmatic markers under consideration in final position, using punctuation marks in the search parameters: full stop/period, exclamation mark and question mark. For each of these punctuation marks, our aim was to elicit 200 occurrences per word and corpus. To sum up, we targeted 600 occurrences for each of the modal particle candidates alright/all right, already and then in each corpus; for all right/alright, we elicited 600 occurrences for each of the two spelling variants alright and all right. However, sometimes the corpus search yielded fewer than 600 instances; specifically, the corpora returned 9,785 occurrences in total, due to limited occurrences of some pragmatic markers.
In addition, in order to evaluate the role of the final position, we investigated the three different pragmatic markers also in other positions on the basis of 100 additional occurrences per modal particle candidate from the corpus in which the modal particle uses occurred most frequently. Here, the search was not restricted by syntactic position.

Data Analysis
The data were coded first for their function, in particular, whether they allow the interpretation as a modal particle according to the Diewald/Fischer model or not. Some cases could not be clearly decided, as they allowed both the conventional and the modal particle interpretation; these occurrences were annotated as undecided. Furthermore, some occurrences did not allow for any analysis because there was not enough context available to make any decision; these occurrences were discarded. The four categories annotated are thus modal particle use, other use, undecided, and discarded.
Finally, the distribution across sentence types was analysed for all instances that allow a modal particle interpretation.

Modal Particle Uses in English
In this section, we present the results from our corpus analysis, which suggest that there are uses of pragmatic markers (to use the potentially most neutral term, cf. Fraser 1999) in English that can be analysed in accordance with the model suggested above, and which may therefore provide evidence for similar inferential processes as those modal particles evoke in German. For these pragmatic markers, uses can be identified that can be assumed to evoke a pragmatically given proposition that is suggested to be common ground.

Then
The Oxford Advanced Learner's Dictionary (OALD, Hornby & Wehmeier 2005) classifies the current uses of then as either adverbial or adjectival. Since the adjectival uses A-F are very far from a possible modal particle function (e.g. "That decision was taken by the then president"), the modal particle uses will be found among the adverbial uses of then. Several adverbial uses can be distinguished: G. Used to refer to a particular time in the past or in the future a. Life was harder then because neither of us had a job; b. I've been invited too, so I'll see you then H. Used to introduce the next item in a series of actions a. He drank a glass of whisky, then another and then another I. Used to show the logical result of a particular statement or situation a. If you miss that train then you'll have to take the next one; b. S1: You haven't done anything to upset me. S2: So what's wrong then c. S3: My wife's got a job in Glasgow. S4: I take it you'll be moving then. J. Used to introduce additional information a. She's been very busy at work and then there was all that trouble with her son. K. (formal) Used to introduce a summary of something that has just been said a. These, then, are the main areas of concern L. Used to show the beginning or end of a conversation, statement, etc. a. S5: I really have to go. S6: Ok. Bye then b. OK then, I think we've just about covered everything on the agenda.
The modal particle use is generally the one under I (in final position, when it is not part of an if-then construction); using our model, example I b. can be analysed as follows: S1: You haven't done anything to upset me. We thus take the function of final then in this example to indicate that the question "What's wrong?" was already pragmatically given and thus obvious to ask. One may argue that then in this example conveys a certain attitude, like annoyance. While this may be true, we consider such meanings to be pragmatic effects of the exchange. In contrast, the grammatical function of modal particles in the Diewald/Fischer model is to anchor the current utterance in argumentative common ground, i.e. to evoke a pragmatically given unit to which the current utterance responds (see Diewald 2006). The pragmatic effects are thus secondary to the primary, grammatical function (and rather due to the interactions between sentence type and particle meaning, see section 2.1 and below).
As the above examples illustrate, besides in interrogatives, then can also be found in declarative sentences in the modal particle function. Whereas in questions the role of then is to anchor the current question to the common ground by indicating that the question follows from the common ground, in statements we take then to present the current utterance as a logical conclusion from prior discourse or common knowledge and therefore to be pragmatically given. See, for instance, the analysis of example I c.: S3: My wife's got a job in Glasgow. Pragmatically given unit: if one has a job in Glasgow, then one may want to move. Current situation: I think: you may want to move. → utterance (S4): I take it you'll be moving then.
So while final then in questions serves to indicate that asking the question is motivated by the common ground, in declaratives it presents the current utterance as following from common ground and thus as communicatively given. The analyses thus differ depending on whether then occurs in a question or in an assertion; as argued above, modal particles are sensitive to the sentence types and their illocutionary potential (see, for instance, Gast 2008; Alm et al. in press; among many others). To sum up, the function of then in general is to indicate that the question or assertion follows naturally from the interlocutors' common ground. In this way, then marks the question or assertion as pragmatically given. König et al. (1990: 52) argue that then is a possible translation for the German modal particle denn when it refers back to a statement or an action of the interlocutor that provides the reason for the speaker's question and shows that the respective interaction has a previous history;8 examples for this use are (König et al. 1990:52): (1) German: Wo ist es denn? English: Where is it then?
German: Was willst du denn nun tun? English: What are you going to do, then?
Furthermore, then may function as a translation equivalent of stressed denn if it refers to a previously discussed but negatively answered possibility and requests delivery of the missing information.9 An example for this use is (König et al. 1990: 54): (2) German: A: Warst du in Berlin? B: Nein. A: Wo warst du dénn? English: A: Were you in Berlin? B: No, I wasn't. A: Where were you, then?
Thus, the stressed and the unstressed versions of denn in German fulfil the same grammatical function, namely to mark the current utterance as non-initial (cf. Diewald 2006), but the stressed version stresses the fact that the question is open and in need of an answer, and English then covers both interpretations.
Corresponding to the fact that in German, denn is restricted to interrogative sentence types (e.g. Kwon 2005), all of König et al.'s (1990: 52-54) examples containing then are questions, and also Haselow (2011) 8 German original: "verweist zurück auf eine Aussage oder Handlung des Gesprächspartners, die Anlass zur Frage des Sprechers gibt; zeigt an, dass die jeweilige Interaktion eine Vorgeschichte hat" (König et al. 1990: 52) 9 German original: "bezieht sich auf eine vorangegangene verneinte Möglichkeit und fordert auf, die fehlende Information zu liefern" (König et al. 1990: 54) Brought to you by | University Library of Southern Denmark -Syddansk Universitetsbibliotek Authenticated Download Date | 1/9/19 3:03 PM argues that then is used as a modal particle only if it occurs finally in "information seeking directives of the wh-interrogative type;" however, as the examples in the OALD above suggest, the modal particle interpretation is also possible in declarative sentences. Haselow (2011Haselow ( , 2012Haselow ( , 2013 analyzes then as a modal particle in the International Corpus of English (ICE) based on the semantic schema introduced by Diewald and Fischer (1998). He presents, for instance, the following example of then as a modal particle (Haselow 2011: 3614 I think uh people are tired (.) Haselow (2011: 3614) argues that then in this example is a modal particle because the propositional content of the directive does not refer to any state of affairs related to Matt or Fiona that serves as the background for the utterance with final then. Rather, speaker B introduces a new subtopic with the directive, marking it as fitting into the context by means of creating a contrastive relation between a pragmatic pretext, i.e. non verbalized proposition (here: 'I expected Lee to be involved as well.'), and the situation (here: 'Lee has not even been mentioned thus far.').
By evoking a topic-organizing function and by freely paraphrasing the proposition that is assumed to be common ground, Haselow passes the opportunity to use the basic semantic schema introduced by Diewald & Fischer (1998), in spite of his claims to base his analyses on this model (Haselow 2011). Our analysis of this example of then relies on the application of the schema:

Semantic schema applied to what's up with Lee at the moment then:
Pragmatically given unit: it's open what is up with Lee current situation: I wonder: what's up with Lee → Utterance: What's up with Lee at the moment then The work done by then is thus to present the current question as a consequence of something communicatively given, in particular that the question at hand, which concerns a member of the same category as the other possible candidates discussed before, has not been sufficiently addressed yet -which is largely equivalent to Haselow's paraphrase and can be considered as an instance of data triangulation. However, while our analysis rests on the application of a semantic schema, Haselow's procedure is generally more prone to error. Another example from Haselow (2011) shows that the fact that he does not use the semantic schema provided leads to different analyses. In particular, Haselow (2011: 3616) suggests that example (4) is not an instance of a modal particle because the reference unit (the implied relationship between being happy and not writing) is in fact verbalized: According to the Diewald/Fischer model, the expectation that someone who is happy is likely to write is perhaps implicitly alluded to by the fact that the question why he does not write is presented as open. However, for the reconstruction of the relationship between happiness and writing, other mechanisms than the workings of the pragmatic marker are responsible. For example, such inferences might be accounted for by the topoï developed in the Argumentation in Language Theory (Anscombre and Ducrot 1983) or in a frame semantic representation (e.g. Fillmore 1982). The model of the inferences involved in modal particles and their translation equivalents presented here allows only to reconstruct that the failure to write constitutes an open issue that is taken to follow from the common ground, including the discourse record.
To sum up, while Haselow (2011Haselow ( , 2012Haselow ( , 2013 takes the same theoretical starting point by alluding to the Diewald/Fischer model, his results only overlap partially with our own. Therefore, given the differences identified, in order to be on the safe side concerning the number of modal particle readings he identifies (34 out of 57 instances of then in final position in interrogative clauses (Haselow 2011: 3616)), we take a fresh look at the corpus for then, as well as for the other pragmatic markers.
In our corpus study, we coded between 504 and 601 instances of final then in the BNC, COCA and the SOAP corpora (depending on availability), and find between 10.67% (COCA) and 30.45% (BNC) of modal particle uses of then, using the Diewald/Fischer model. So especially for British English, the modal particle function seems to be a very common use, see Table 1. We distinguish between clear cases to which the model applies, cases with clear temporal uses, unclear, undecided cases and uses that do not allow any decision because of lack of context. Table 2 illustrates the kinds of corpus instances classified as undecided; these examples oscillate between a possible temporal reading (e.g. where was Roscoe the guard dog at that time?) and a modal particle interpretation (the question is open: where was Roscoe the guard dog). Example (5) below illustrates how then can serve to mark the upcoming question as non-initial and as following from what previously was said (pragmatically given is the question whether I can ask a question), whereas example (6) illustrates final then fulfils in an imperative clause (Aidan suggests that it is pragmatically given that Anita should tell him): (5) F7FPS001: And I think kids should be encouraged (pause) to erm (pause) think about themselves rather than just tick boxes (pause) and I think er, er er a group of sentences down here written by the form teacher, by the (pause) subject teacher and the pupil (pause) erm, as it's been done for about, for a few years, a good (pause) habit to get the kids into, erm (pause) especially as a record of achievement should start from year seven.
10 The number of discarded occurrences for then is particularly high since because its meaning is consecutive, the context is particularly important to identify the presence of the modal particle function.
Brought to you by | University Library of Southern Denmark -Syddansk Universitetsbibliotek Authenticated Download Date | 1/9/19 3:03 PM F7FPSUNK Mm. Can I just ask then, if you're saying to be it should be a joint statement, but you F7FPS001: Mm. F7FPSUNK: think that format's alright, is that not a contradiction because (pause) if it's a joint statement and this format's alright you've got staff comment (pause) divorced from pupil comment? (BNC, Teachers' Conference 1992) (6) Aidan: Shooting star. Quick, make a wish.
Anita: I see it, I see it! Aidan: Did you make a wish? Anita: Of course. Aidan: Well, tell me, then. What did you wish for? (SOAP, All my children 2005) The modal particle uses of final then were subsequently analyzed further according to their occurrence in different sentence types. We distinguish here between declarative, interrogative and imperative sentence types, as well as between their occurrence in quoted (direct) and reported (indirect) speech. Table 3 provides an overview of the grammatical contexts of modal particle uses of final then across the corpora. The analyses confirm that the modal particle uses of final then are mostly associated with interrogatives, but they also show that other sentence types are possible, as also the examples from the OALD suggested. The analysis of 100 additional instances in all kinds of positions shows that the modal particle reading can generally only be found in final position, as suspected by Haselow (2011Haselow ( , 2013. Of the 10 final instances of then in our sample of 100, there were three undecided cases, six modal particle uses and four adverbial uses.
To sum up, we have seen that then in final position is used with modal particle meanings between 10% and 30% of the time, predominantly in interrogative clauses, but also in some other sentence types.

Already
The Oxford Advanced Learner's Dictionary classifies already as an adverb and lists the following uses: A. Before now or before a particular time in the past a. S1:'Lunch?' S2:'No thanks I have eaten already' B. Used to express surprise that sth has happened so soon or so early a. S3: Is it 10 o'clock already?
Brought to you by | University Library of Southern Denmark -Syddansk Universitetsbibliotek Authenticated Download Date | 1/9/19 3:03 PM C. Used to emphasize that a situation or problem exists a. There are far too many people already As a temporal adverb, already denotes a relation between a certain pragmatically given, i.e. expected, transition point (ETP) and a real transition point (TP). Together with the tense, already indicates that the TP has occurred prior to the time of utterance (T u ), as in (7) (original already often occurs with present perfect): (7) _________________TP_____T u ____ETP____ Now, the English adverb already can also be found in clause-final position in ways that deviate from its previous temporal use,11 as illustrated in (8) and (9) This use is described in the Oxford English Dictionary as originally and chiefly American English, especially in Jewish usage, "as an intensifier after a word or phrase, esp. to express impatience, exasperation, etc. frequently in enough, already!" The first example of this usage they provide is from 1903. In the modal particle analysis of already, the function of the novel use of already is taken to anchor the current utterance in common ground. If our utterance is, for instance, 'Move those sheep already!' (example 8 above), a perceptually available aspect of common ground is that the addressees, here the shepherds, appear to assume that they may move the sheep later, so that the analysis is: Semantic schema applied to Move those sheep already! Pragmatically given unit: you seem to think you may move those sheep later current situation: I think you should move those sheep now. → Utterance: Move those sheep already! This analysis assumes that already evokes an inference as to the existence of a pragmatically given proposition. In this example, the perceptually available situation (Clark 1996) may be taken to indicate that the addressees are behaving as if they were thinking that they could move those sheep later. Already serves here to evoke this contextual background and to relate the current utterance to that background, thus indicating 'why that utterance here now' (Fischer 2007). In line with other contextualization cues, already is thus taken to contribute to the interactive constitution of the shared situation (see also Gumperz 1992). Our analyses of the distributions of this novel use of already in the BNC, COCA and SOAP corpora show that the proportion of modal particle uses of already is relatively high in the COCA corpus and highest in the SOAP corpus (almost 22%), whereas it is virtually non-existent in British English; in fact, we found a single example, which however was a quote from a speaker of US English (Madonna): (10) '…to live in the public eye, I mean, I would slit my wrists already. I can't understand how this poor girl can take it.' (The Daily Mirror, London 1992, BNC) Table 4 shows the distribution of occurrences for already across corpora. We thus have to assume that this use of already is a) recent and b) regionally restricted. This makes already particularly interesting for our analysis since it can shed light on the question whether an abstract constructional schema can be assumed in English that might license the new uses. Table 5 shows examples of undecided occurrences, which allow both a temporal and a modal particle reading. While this use of already is most common in the imperative, it is not restricted to it (even though a directive meaning is implicit in all of the examples, cf. Diewald 2008 for similarities with the German modal particle ruhig (roughly: 'calm')): (11) My husband -well, he says I just need to relax already. Thus, even though already as a modal particle is most common with the imperative sentence type, previous work on German has shown that such restrictions tend to disappear over time (Diewald 2008), and as the examples below illustrate, it is used also in other constructions:  Table 6 shows the quantitative distribution of the modal particle uses of already across sentence types. The additional analysis of already in all kinds of positions shows that the modal particle function is restricted to the final position only, where the three modal particle uses identified among the 100 occurrences in the SOAP corpus occurred in final position only.
To sum up, our suggestion is that already in final position indicates a discrepancy between a pragmatically given expectation (concerning a certain temporal order in which things are expected to happen) and the current situation and thus anchors the current utterance to some kind of common ground, which may comprise a number of different resources, such as the shared perception, shared cultural facts, or dialog history (see Clark 1996;Alm & Larsen 2015). The modal particle provides a clue by indicating what the kinds of contents are to which to the current utterance is oriented and how to identify these.

Alright/all right
The Oxford Advanced Learner's Dictionary describes all right/alright as an adjective, adverb, and an exclamation and lists the following major uses (alright is described as an informal version of all right): In example E, all right modifies the entire sentence, which is a typical feature of modal particles (Diewald & Fischer 1998). Furthermore, in König et al. (1990) this usage of all right is suggested to serve as a translation equivalent for the German modal particle schon and to express emphatic affirmation and confirmation of a circumstance or situation, often in contrast with the view of a conversational partner (König et al. 1990: 205).12 König et al. (1990: 205)  Just as in example E from the OALD, in these instances all right modifies entire clauses, and the paraphrase offered by König et al. (1990: 205), that it affirms a given proposition, indicates the relational nature of alright, where the proposition in question is part of the suggested common ground. Accordingly, our analysis of example E Oh, it's her all right is as follows: The pragmatic marker serves here to evoke a proposition in common ground that presents the proposition expressed as in question. While König et al.'s paraphrase of schon suggests that the communication partner is taken to expect the contrary, we believe that the contrast evoked results from the fact that the speaker's confirmation may only be partial; all right/alright serves to indicate what can be agreed on before the part that cannot be taken to be settled is presented. Thus, alright marks a particular information as uncontroversial -potentially in contrast with information that is controversial. Accordingly, in the analysis of example (iv.) in König et al.'s collection, the current utterance affirms one of the alternatives in the common ground; however, the concession is already alluded to by but: Our quantitative analysis indicates rather few uses of final alright/all right with modal particle function, yet these are relatively equally distributed across corpora: Without exception, all modal particle uses of alright and all right in all corpora were found in declarative utterances. Further examples from our corpus analysis include the following: (17) (SP:KPJPSUNK) And he sa--and he said to me one day, are you fucking deaf! And I heard it. I says, I haven't got any, and he came down here and I said to him, I'll give you fucking deaf alright! I said, when you press that button upstairs it muffles it and there's nothing you can do. (BNC) Examples for undecided cases in the BNC, which allow both the modal particle interpretation and the literal, adverbial interpretation, are listed in Table 8. (SP:D95PS001) Erm, the first B. ones that was growing in the shelter of the oak trees, they were surviving alright. Patch here and there. (unclear) But as soon as you cut the the C. the unwashed masses demand this. HOLDEN (off comic) This is easy, alright! And right now it pays the bills. Just don't forget that we The investigation of 100 position-independent instances of all right in the corpus with the most frequent occurrences, the BNC, revealed six uses with modal particle interpretation, all of which occurred in final position. It seems thus that the modal particle uses are restricted to the final position.

Some Other Potential English Modal Particles
In fact, many other pragmatic markers are likely to fulfil functions similar to those discussed above, see, for instance, Mulder & Thompson (2008) for but, Haselow (2013) for actually, anyway, even, so, though as well as but, Schwenter & Waltereit (2010) for too, among many others.

after all
Another pragmatic marker that may be a candidate for a modal particle is after all (see Fillmore 1984). The OED offers several paraphrases for this adverb that indeed suggest expectation management and therefore the marking of a relationship between the current utterance and a pragmatically given unit: in spite of any indications or expectations to the contrary; when all is said and done, nevertheless. The OALD treats it as an idiom with two uses: A. despite what has been said or expected a. So you made it after all! B. used when you are explaining something, or giving a reason a. He should have paid. He suggested it, after all.
The modal particle function is obvious in use A; our analysis of this example is: Pragmatically given unit: you might not have made it. current situation: you made it. → utterance you made it after all.
The violation of expectation is even more obvious in example (20), which presents a rather common type of example from the BNC from a meeting between seven career counselling staff; at issue is the usefulness of discussing with the participants their experiences with the courses they take: In this example, the speaker makes it explicit that initially, the participants' expectations were different. It is to these negative expectations that the host utterance orients: Pragmatically given unit: the course was bad (awful even). current situation: the course was not bad, they got something out of it. → utterance it wasn't so bad after all.
Thus, after all signals a contrast between the discourse common ground, that the course was awful, with the evaluation after the discussion, that it was actually not that bad. That is, with after all the speaker orients the current utterance to a pragmatically given proposition that made the opposite expectable. Based on uses like this, Fillmore (1984) suggested that English speech routines ('formulas') correspond to German modal particles.

indeed
As an example for a confirming, iterative modal particle use, let us consider indeed. According to the OALD, indeed is an adverb with five different readings: A. used to emphasize a positive statement or answer (e.g. 'You said you'd help?' 'I did indeed-yes.' 'It is indeed a remarkable achievement'.) B. (especially British English) used after very and an adjective or adverb to emphasize a statement, description, etc. (e.g. 'Thank you very much indeed!') C. (formal, especially British English) used to add information to a statement ('I don't mind at all.' 'Indeed, I would be delighted to help.') D. (informal, especially British English) used to show that you are surprised at something or that you find something ridiculous ('A ghost indeed! I've never heard anything so silly.') E. (informal) used when you are repeating a question that somebody has just asked and showing that you do not know the answer ('Why did he do it?' 'Why indeed?') Here, the modal particle reading is the one under A, with the alleged function to emphasize a positive statement or answer. While indeed can be a focus particle in 'It is indeed a remarkable achievement' and have scope only over 'a remarkable achievement', it can also be analyzed as having sentence scope and relating to a proposition in common ground stating that it is a remarkable achievement, which is confirmed by the current utterance: Pragmatically given unit It is a remarkable achievement. current situation: I think: it is a remarkable achievement. → utterance it is indeed a remarkable achievement.
As a modal particle, indeed serves to iterate (and hence to emphasize) a proposition which is suggested to be part of the speakers' common ground. In the following example from an advertising text from the BNC, we find indeed in a statement that retaining control on a wet road can save your life: (21) CFT W_advert: The S2's ABS brakes have internally ventilated discs that won't lock. So even with the brake pedal to the floor on wet roads, you can still retain steering control. Which may indeed save your life.
This use is most similar to reading A, given that it is used to emphasize a positive statement; however, our analysis can shed light on how it may achieve this function and why it is persuasive (used in an ad): Pragmatically given unit retaining steering control may save your life. current situation: retaining steering control may save your life. → utterance it may indeed save your life.
In this example, indeed points the reader to information, which is presented as given, that retaining control can save lives. Thus, the ad does not only present the conclusion, it pretends that this conclusion is common knowledge and that the reader has known this all along -which can be taken to make the ad more effective: the reader has reached the conclusion him-or herself. Thus, with indeed the writer asserts that the conclusion is uncontroversial and based on the reader's shared knowledge. In these uses, indeed is therefore a confirming, re-iterating pragmatic marker that connects the current utterance to a pragmatically given unit -and thus it is very similar to modal ja (see also König et al. 1990: 145-148;Fischer 2000).

just
In König et al. (1990: 80), the meaning of einfach13 is described as either characterizing a solution or explanation as self-evident and obvious, or as underlining that the solution the speaker offers to a problem is unproblematic (yet essential) compared to other options. The above examples of just (H,K) match one of these two meanings described by König et al. (1990): In example H, the speaker characterizes his explanation that 'he can't just drop all his commitments' as obvious, and dropping all commitments is furthermore presented as not simple. In K, the listening can be viewed as the speaker's unproblematic solution to a problem.
Brought to you by | University Library of Southern Denmark -Syddansk Universitetsbibliotek Authenticated Download Date | 1/9/19 3:03 PM Pragmatically given unit: one may think that there might be problems doing it. Current situation: I suggest: you should do it → Utterance: Just do it! Our analysis reveals between 27% (BNC) and 63% (SOAP) modal particle uses of just (see table 9). While just receives an interpretation as modal particle predominantly in imperative clauses, this reading is also possible in interrogatives, especially in expansion questions, for instance: To sum up, all of the uses discussed can be analysed using the model by Diewald and Fischer, suggesting that they fulfil similar functions as German modal particles and evoke similar inferential processes in interaction.

General Discussion
Our analyses above show that there are pragmatic markers in English that can be analysed in the same way as German modal particles; concerning question 1), whether English pragmatic markers may fulfill the functions of German modal particles and thus give rise to similar inferences in conversation, we can now state that yes, it is possible that they do. Note however that in each case, other interpretations of the respective functions of the pragmatic markers under consideration are possible, as obvious from the paraphrases provided by the OALD, for instance; thus, while it is possible that English pragmatic markers lead speakers and listeners to similar inferences as German speakers, other kinds of inferences are also possible. For instance,  takes pragmatic markers at the right periphery to be used first as "subjective metatextual comments on what precedes" (2016: section 4.5), some of which may then develop into retrospective pragmatic markers, which, in Lenker's (2010: 198) words, "force a re-processing or even reinterpretation of the preceding assertions," such as like then, though, however, after all, and actually. In a recent talk, Fischer & Traugott (2016) argue that the novel functions of already can be explained not only as modal particle uses, but also by independent sense-extension mechanisms. The novel readings of the pragmatic markers considered could build on each marker's individual semantics, but following general sense extension mechanisms, such as metaphor and metonymy. Thus, while it is possible that the items under consideration fulfil modal particle functions, it cannot be excluded that each pragmatic marker has developed these functions independently, and that possibly other pragmatic functions (such as signalling annoyance) describe their uses better than anchoring the current utterance in common ground.
Regarding question 2), how frequent such uses are, we have provided frequency data for all right/alright, already, just and then and found that at least some show considerable proportions of modal particle uses Brought to you by | University Library of Southern Denmark -Syddansk Universitetsbibliotek Authenticated Download Date | 1/9/19 3:03 PM (e.g. 30.45% for then in the BNC and 63.26% for just in the SOAP corpus), whereas for others, especially for alright/all right, modal particle interpretations are rather rare. While this does of not answer the question how frequent modal particle meanings are in English in general, we can estimate their frequency once we know the general frequency of those pragmatic markers in English that may be used in this way. In any case, we can conclude that potential modal particle uses in English are too common to be ignored.
As for formal criteria (question 3), while there are clear positional features that identify modal particles in German, in English, position seems to be flexible for some particles and relatively fixed for others: While some of the modal particle candidates investigated were found to be restricted to the utterance-final position, indeed may receive modal particle interpretations in medial position while just cannot occur in final position at all. Thus, there is no clear association between form and meaning for the whole range of items that may evoke similar inferences as German modal particles.
This has consequences for question 4, what our findings may mean for postulating the existence of a schematic construction for modal particles in English since the first type of evidence for constructional status is whether there are form-function correspondences at all, since a construction is a conventional association of form and meaning (e.g. Goldberg 1995). Taking a function-oriented starting point for the cross-linguistic comparison revealed that items in different positions may fulfil modal particle functions and thus that the function cannot be uniquely assigned to a single grammatical position. Furthermore, the meaning side of a possible construction, to relate the current utterance to an aspect of common ground, may not be specific to the pragmatic markers under consideration, either. For instance, the context evoking function of German modal particles has also been suggested for negation (e.g. Ducrot 1996). For example, the utterance I'm not coming evokes a pragmatically given proposition that it was expectable that the speaker might have been coming. Conversely, items in the same position, for instance, in utterance-final position, do not necessarily take a modal particle meaning. In German, for some modal particles, this is very different; for instance, when ja or aber occur in the typical middle field position, they automatically receive a modal particle interpretation. However, for other modal particles in German, the situation is not much different from the situation we have observed for English: The interpretation of the particle word auch in middle field position, for instance, depends crucially on the placement of the sentence accent (e.g. 'Kannst Du das auch?' may mean either 'can you do that, too?' (accent on auch) or 'are you at all able do that?' (accent on the verb)). Hence, the characteristic middle field position alone does not automatically cause all particles to receive a modal particle interpretation. This is what we have observed in our corpus study: Modal particle interpretations can be assigned to some instances of pragmatic markers in the positions studied but not to all, and some cases were undecidable based on the transcripts only. Most likely, prosodic information would facilitate the disambiguation, which was unfortunately beyond the scope of the current study. In any case, we can conclude that a one-to-one correspondence between form and function could not be identified; no single grammatical position is associated with this use, not even within the same sentence type constructions. For instance, in imperative sentence type constructions, already, alright and then occur in utterance-final position, whereas just occurs utterance-initially. Nevertheless, since the function to evoke a pragmatically given unit in the common ground depends on the respective item having sentence scope, the modal particle candidates investigated occur predominantly in positions that favour sentence scope readings, most of them occurring in utterance final position only. Since pragmatic markers in English usually fulfil discourse marker functions in initial position (e.g. Schiffrin 1987), it is likely that they develop other functions in other positions, and that these positions need to indicate scope over the whole clause. Thus, while no single position seems to be reserved for English modal particle candidates, their occurrences are restricted to a few positions.
Besides form-function correlations, as evidence for a schematic category may count if a development concerns various different lexical types (e.g. Bybee 2010). The large number of different pragmatic markers with modal particle functions suggests that there may be a general construction that licenses the modal particle interpretation across items. That a schematic category may license modal particle interpretations is supported by the fact that the newcomer already, which is so new that it has not even spread to British English yet, is probably understood there, since Madonna's quote was not edited by the Daily Mirror and hence was not seen as problematic for the British audience. While our study analysed only six modal particle candidates, it still seems that the final position is the one that allows generalization to new instances best (see also Haselow 2013;. Another example that may provide evidence for a schematic category is a recent final use of a contrastive pragmatic marker in Australian English, namely but, which, according to Mulder, Thompson and Williams (2009), fulfils sociolinguistic functions, signalling Australianness, as well as contrast marking and turnyielding functions. The authors argue that "there is no implication of semantically contrastive material left 'hanging'" (Mulder et al. 2009:351), for instance: (25) Coach: That'll do it, lads. Good work but.
This example can be perfectly well analysed using our schema: Pragmatically given unit One may think it is not good work. current situation: I think: it is good work. → utterance good work but.
The possible pragmatically given unit, that the lads had not been doing good work, may either be an inference from the coach's own suggestion that they stop doing what they are doing, or may have another basis, which is suggested to be shared. The fact that but in this position is used as a signal of community membership and has thus developed into a sociolinguistic marker is of course unaccounted for in our model. However, the fact that it is interpretable and has spread so readily in Australia suggests that speakers and hearers have certain resources at their disposal that facilitate this interpretation. This could be a schematic construction with a modal particle meaning, or it could be the knowledge about final though, in analogy with which but is interpreted. We can conclude that no clear form-function pairing can be established, but that the final position may be developing in this direction. At the same time, it may not be necessary to assume a modal particle construction in English since the inferential processes peculiar to modal particles may exist without a specific category. As Langacker (1991Langacker ( , 2008 suggests, one exemplar suffices to give rise to similar creations by means of analogy, and the case of the recent Australian use of but may be a case in point since it is possible that it is interpreted in analogy with final though. Thus, a schematic category may license novel uses, and exemplars may cluster and reinforce each other; in the usage-based model, exemplars, subschematic and schematic constructions may co-exist and encode linguistic information redundantly (cf. also Alm et al. in press). From that perspective, the common ground anchoring functions proposed for the pragmatic markers under considerations could be related so that they inspire each other and thus form clusters of instances rather than a fully-fledged category yet.

Conclusion and Future Work
We conclude from our analysis that even though a schematic modal particle construction could not be demonstrated, it still became apparent that modal particle uses of pragmatic markers are in fact potentially very wide spread in English. From the perspective of inferential processes involved in English interaction, the suggested interpretations for the particles under consideration and thus the underlying inferential processes that anchor utterances in speakers' common ground, are comparable to those involved in German interaction with its pervasive use of modal particles. Thus, despite of the potential lack of a category of modal particles, the underlying inferential processes may actually be quite similar after all.
One big question to be addressed in future work concerns the role of other formal features than position. For instance, our corpus study has shown that monosyllabic modal particle candidates have larger proportions of modal particle uses than those with two or more syllables. Given that our quantitative Brought to you by | University Library of Southern Denmark -Syddansk Universitetsbibliotek Authenticated Download Date | 1/9/19 3:03 PM analysis covers only four modal particle candidates, this observation serves only as a starting point for future research, but it is in line with Fillmore's suggestion that what keeps speech routines in English from being used as modal particles as often as their German translation equivalents is their length and lack of integration. In addition, prosody should be included in the analysis, as it also plays a big role in the definition of German modal particles (e.g. Helbig 1988;Thurmair 1989). Moreover, while we have restricted our investigation to pragmatic markers here only, in the long run also other lexical and grammatical means that may give rise to the same inferential processes as German modal particles should be investigated further (e.g. Fischer 2007).
Furthermore, future work will have to clarify the temporal relationship between the final and the other uses in order to determine whether it is the final position that gives rise to the new uses. This will then provide further evidence for the development of a schematic modal particle construction in English. Similarly, such a diachronic study may also want to address the relationship between the number of different grammatical contexts in which a particular pragmatic marker may fulfil modal particle functions and their general frequency since higher context variability may suggest increasing grammaticalization (cf. Diewald 2008). In our corpus study, we found that two markers (then and already) with relatively high proportions of possible modal particle readings were common in several sentence type constructions, whereas just, in spite of its high proportions of possible modal particle interpretations, and all right/alright were very restricted with respect to the sentence type constructions in which they can occur. Thus, future work should address whether the variability concerning the different sentence type constructions observed for some modal particle candidates corresponds to increasing grammaticalization.
Future work may also want to address the role of regional differences; concerning already, considerable differences between corpora could be found such that the modal particle use was hardly present in British English, while then was more commonly used with modal particle functions in British than in American English. Tracing these developments over time and for more modal particle candidates (for instance, for indeed) will inform us about the development of a possible schematic category as well.
Finally, the study of potential modal particles in English needs to be broadened to even more modal particle candidates in order to get a better understanding of how widespread the phenomenon really is.