2.1 Content-level and context-level uses of linguistic expressions

Discourse-pragmatic elements are known for their polyfunctionality (Cuenca & Degand, 2022; Fischer, 2014).¹ On the one hand, this can be attributed to the fact that they can simultaneously operate in different functional domains of discourse (Schiffrin, 1987). On the other hand, the polyfunctionality of many discourse-pragmatic elements may reflect the fact that they have homophonous counterparts that operate at the propositional level (Sweetser, 1990, pp. 76–86). A profitable approach to frame these issues revolves around the distinction between content- vs. context-level uses of linguistic expressions, as formulated by Hansen (2008, pp. 14–17, 2012, pp. 592–594).

The first label refers to any use of a linguistic expression in which the meaning of that expression bears saliently either on a state-of-affairs/proposition referred to in its host clause or on the relation between that state-of-affairs/proposition and other states-of-affairs/propositions. The second label, instead, refers to any use of a linguistic expression in which that expression primarily expresses the speaker’s comment either on the relation between a described state-of-affairs/proposition and the discourse itself, or on the relation between a described state-of-affairs/proposition and the wider speech situation (including contextual features, the subjective attitudes and mental states of the interlocutors, the interactional common ground). Some linguistic elements display a type of polyfunctionality that crosses this divide: if this is the case, they show a coexistence of content-level and context-level uses. The uses pertaining to the first group operate at the propositional level, while those pertaining to the second group operate at the discourse-pragmatic level.

Both elements under scrutiny in this paper show a coexistence of content-level and context-level uses. Example (1) features two occurrences of the Italian adverbial expression un po’ ‘a bit’. The first one represents a use of un po’ as a speech-act modifier operating at the discourse-pragmatic level (context-level use), while the second one represents the prototypical use of this expression as an adverbial degree modifier operating at the propositional level (content-level use).²

In a similar fashion, example (2) features several occurrences of the European Portuguese adverb lá ‘there’. The first occurrence represents the prototypical use of lá as a deictic locative operating at the propositional level (content-level use), while the four occurrences of lá in the last chunk of conversation represent the use of lá as a speech-act modifier operating at the discourse-pragmatic level (context-level use).

2.2 Illocutionary force and information management

In this paper, the discourse-pragmatic functions expressed by the context-level uses of un po’ and lá will be analyzed as related to two functional domains, namely illocution and information management. The first domain relates to the performance of speech acts, while the second domain relates to the strategies adopted by the speaker to present and highlight the information conveyed by the speech act.³ To better outline these domains as well as to frame the functional similarities between un po’ and lá, I will refer to the concepts of illocution and communicated content as elaborated by Functional Discourse Grammar (FDG), a typologically based theory of language structure (Hengeveld & Mackenzie, 2008). FDG relies on the idea that grammatical categories are organized in layers, connected to each other by scope relations: “in Functional Discourse Grammar scope relations are defined in terms of different pragmatic and semantic layers. Pragmatic layers together constitute the interpersonal level in this model, while semantic layers together constitute the representational level” (Hengeveld, 2017, p. 15).⁴

The layers that constitute the interpersonal level are the most relevant here. This level of analysis deals with all the formal aspects of a linguistic unit that reflect its role in the interaction between speaker and addressee. The highest layer is the move, which represents an entire segment of discourse. The move consists of one or more (sequentially ordered) discourse acts, which are the basic units of communication. Each act in turn consists of an illocution, which specifies the communicative intention of the speaker, the speech participants, and the communicated content, which represents the message transmitted in an utterance. Finally, within the communicated content, one or more subacts of reference and ascription are executed by the speaker, by means of which they refer to entities and ascribe properties to these entities. Moreover, the basic content of each layer may be further specified by operators and modifiers: operators capture specification by grammatical means, while modifiers capture specification by lexical means (see Hengeveld, 2017, pp. 14–18).

The attention given by FDG to the distinction between the illocution and the communicated content represents a key element in the present paper. The notion of illocution refers to the grammatical domain where communicative intentions are coded onto conventionalized linguistic expressions: “The basic illocution of a sentence can be defined as the conversational use conventionally associated with the formal properties of that sentence which together constitute a sentence type” (Hengeveld, 2004, pp. 1190–1191). Basic illocutions are represented by categories such as declarative, interrogative, and imperative, while the notion of illocutionary force is used to distinguish among specific types of speech acts, such as statements, requests, orders, warnings, permissions (see Levinson, 2017; Searle, 1969). Grammatical operators associated to this layer express illocutionary modification:

Basic illocutions may be further modified by markers of what I here call illocutionary modification. Like basic illocution, illocutionary modification should be interpreted in terms of the conversational use of sentences. But unlike basic illocution, markers of illocutionary modification do not identify sentences as speech acts of certain types, but rather mark much more general communicative strategies on the part of the speaker: they reinforce or mitigate the force of the speech act. (Hengeveld, 2004, p. 1192)

In this respect, the notion of illocutionary modification refers to the various grammatical means that modify the illocutionary force of a speech act and further differentiate between communicative intentions (to wit, the marking of specific illocutionary forces such as making statements and requests, giving orders, warnings and permissions).

Conversely, the communicated content captures the information that the speaker wants to bring across and the different strategies for molding it:

Whereas the Illocution indicates the conventionalized conversational use of a Discourse Act, and the Participants represent the essential Speaker-Addressee dyad, the Communicated Content contains the totality of what the Speaker wishes to evoke in his/her communication with the Addressee. In actional terms it corresponds to what Searle (1969) calls the ‘representational act’ and corresponds to the choices the Speaker makes in order to evoke a picture of the external world s/he wants to talk about. (Hengeveld & Mackenzie, 2008, p. 87)⁵

Overall, the communicated content is concerned with information management, as it deals with the way in which the speaker presents information and organizes the information flow.⁶ The pragmatic relations that determine the information structure of an utterance (focus, topic and contrast) are expressed at this layer (Hengeveld & Mackenzie 2008, pp. 89–99). Further interpersonal strategies expressed at this layer deal with the marking of the source of information (Hengeveld & Hattnher, 2015), the marking of the speaker’s reaction and attitude towards the information flow (Hengeveld & Olbertz, 2012), and the degree of straightforwardness of the message (Hengeveld & Keizer, 2011). In the next sections, different features related to the domains of illocutionary force and information management will be used to describe the functions expressed by context-level uses of Italian un po’ and European Portuguese lá.⁷

3.1 Directive speech acts

The use of un po’ as an illocutionary operator is particularly clear in directive speech acts in the form of imperative sentences, where the adverb occurs in the postverbal position. Examples of un po’ in directive speech acts can be grouped in three subsets according to the function expressed and the type of conversational context. In a first subset of examples, un po’ operates as a mitigating particle: it downtones the illocutionary force of the speech act, specifying directives as soft requests or invitations. In example (4), in the context of an oral exam at school, un po’ downtones the force of the professor’s directive (raccontami un po’ ‘tell me ptc’). Obviously, the professor is not asking the student to give a partial answer to the question, they rather perform the speech act so that it does not not sound too overbearing.

In a second subset of examples, un po’ brings about a different conversational effect: it enriches requests with an incitement flavor or a reinforcing overtone. In example (5), the speaker is complaining about the quality of a photo he received from a friend (it is crooked and the screen in the picture has dust on it): imagining an exchange of words with him, he utters a marked directive (togli un po’ quel dito di polvere ‘take off ptc that inch of dust’). In this case, along with a specific flavor of casualness, the utterance sounds like a stressed request.¹⁰

The last subset includes expressions like senti un po’ ‘listen a bit’, guarda un po’ ‘look a bit’ or pensa un po’ ‘think a bit’, which are composed of high-frequency imperatives marked by un po’ (see Table 1 below). They represent routinized directives which often display a non-compositional meaning: rather than expressing a directive speech act, they are used as attention-getters, to highlight specific points in the interaction or to segment discourse chunks, as in example (6). In other words, they have reached a discourse-marker status (see Waltereit, 2002, on guarda ‘look’; see also Oliveira, 2023).

In many cases, the contribution of un po’ to this kind of imperatives is more complex. In fact, it does not only contribute to identifying the conventionalized sequence as a discourse marker, but it also specifies the speaker’s attitude towards the information evoked. Specifically, in imperatives like (6) un po’ also gives a mirative flavor to the utterance (on mirativity, see Delancey, 1997; Hengeveld & Olbertz, 2012). When un po’ conveys a mirative reading, it has always scope over a directive (ma pensa un po’! ‘but guess what/imagine that!’, guarda un po’ chi arriva! ‘look who’s coming!’), so that the mirative value seems to develop as a sort of by-product of the context-level use of un po’ in this kind of directives.¹¹

Finally, un po’ can occur in hortative speech acts, which can be considered as closely related to directives (see König & Siemund, 2007, p. 313). In this kind of speech acts, un po’ combines with first person plural subjunctives. The most typical case is vediamo un po’ ‘let’s see’, which represents a routinized expression used as a discourse marker.

3.2 Assertive speech acts

Context-level uses of un po’ also appear in another illocutionary context, namely assertions. This context of use shows greater continuity with the use of un po’ as an adverbial degree modifier. However, in that case un po’ has scope over gradable expressions (adjectives, adverbs, verbal phrases), while when it is used as an operator on assertive speech acts it has scope over the illocution. In example (8) below, un po’ operates on a non-gradable predicate, so that the value of degree modifier seems to be excluded. Thus, the particle does not express a degree in relation to (parts of) the propositional content of the sentence, but rather modifies the degree of assertivity of the utterance, that is, the illocutionary force expressed by the speech act. This construction is typically used when speakers assert something, but they are not completely sure about (or confident with) what they are saying, or when they want to limit the conversational impact of their utterance. This way, the presence of un po’ specifies the assertion as a suggestion, giving the utterance a non-assertive tone:

In example (8), different contextual factors suggest that the speaker might show a weak commitment towards the performance of the speech act: the reported information (the fact that Leo is dating someone) could not have been verified, or the speaker does not feel entirely entitled to share it with the addressee. This being the situation, the speaker uses un po’ to lower the degree of assertivity of their speech act.¹²

Other examples of un po’ in assertive speech acts seem to perform a different function. In particular, it is not always easy to assess to what extent this particle operates on the illocutionary features of the speech act or on the message conveyed by it. To better elaborate on this point, it might be useful to recall the notion of communicated content as elaborated by FDG. As outlined above, the communicated content represents the grammatical layer responsible for information management, that is, the relevant layer for strategies that allow the speaker to structure the information conveyed by their speech act, and to express subjective attitudes towards it. Several examples of un po’ in assertive speech acts express functions related to this domain:

In example (9), the two occurrences of un po’ have scope over verbal phrases (sono un po’ ‘they are ptc’, vanno un po’ ‘they go ptc’) but do not modify the illocutionary force of the speech act. Rather, the use of un po’ in this example reminds the function of linguistic elements that express non-straightforwardness, that is “grammatical and lexical strategies that are available to speakers to convey that the message they intend to communicate is not straightforwardly covered by the basic elements contained in their utterance” (Hengeveld & Keizer, 2011, p. 1962). Specifically, un po’ can be considered a marker of approximation. By using approximation markers, the speaker makes it clear that the selected entity (a lexeme or a whole communicated content) is not a prototypical member of the class to which they belong, or that they are not fully able or willing to provide the exact amount of information needed for felicitous communication.¹³ Consider the following example:

In example (10), several contextual clues point to a scarce degree of speaker’s confidence with regard to the message being conveyed, namely two pause-fillers (mh) and a reformulation marker (nel senso); moreover, the process of reformulation itself is indexed explicitly by a question that the speaker perhaps asks themselves in passing (come spiegare). All these clues point to difficulties in online processing, and this explains the presence of un po’ marking approximation on a referential subact (l’età) and on a whole communicated content (abbandoni l’infanzia).¹⁴ Finally, in example (11) the function expressed by un po’ is different again:

In this case, rather than approximating the communicated content, un po’ contributes to structuring the information evoked. More precisely, un po’ seems to operate as a focus marker (here, in a pseudo-cleft syntactic structure: la cosa è un po’ quella ‘the thing is ptc that’), separating the focused part of the utterance (quella) from the backgrounded one (la cosa è). From an FDG perspective, this is not surprising since pragmatic functions such as focus and topic are expressed exactly at the layer of the communicated content, and some overlap between functions is to be expected (Hengeveld & Keizer, 2011, p. 1974).

3.3 Common features of the context-level uses of un po’

Overall, there is a strong link between the adverbial semantics of un po’ and its use as an illocutionary operator and as a marker of approximation. More precisely, the notion of small degree/quantity expressed by the content-level uses of un po’ is reflected by its context-level uses. In directive speech acts, un po’ is decisive in specifying them as requests. In this respect, the core meaning contribution of un po’, which is directly derived from its adverbial semantics, is to mark speech acts where little effort from the addressee is required to perform the action. In addition to this, other contextual features that vary according to the conversational setting (urgency, unexpectedness, casualness, and so on) contribute to further differentiating between soft and stressed requests, sudden proposals, and incitements. In assertive speech acts, un po’ is decisive in reducing the degree of commitment of the speaker towards their own speech act. In this respect, the function of un po’ is to mark a speech act characterized by low assertivity (if the low commitment targets the illocutionary force of the speech act) or by an approximated transmission of information (if it targets the communicated content conveyed by the speech act).

This evidence shows that the same element can express different functions at different grammatical layers (illocution and communicated content). In the case of un po’, it is important to highlight that both layers can be targeted by the adverb, regardless of the illocutionary context of occurrence. In directive speech acts, the adverb mark requests (illocutionary modification) but can also add a mirative flavor to the utterance (information management). In assertive speech acts, the adverb can mark low assertivity (illocutionary modification), but in many cases can be analyzed as a marker of approximation and focus structure (information management). This way, the functional distribution of un po’ shows the interplay between these two grammatical layers.

4.1 Directive speech acts

Like un po’, the use of lá as an illocutionary operator is well represented by its use in directive speech acts in the form of imperative sentences. The adverb is postponed to the verb. Imperatives with lá can express stressed requests, suggestions, exhortations and warnings.¹⁷ Overall, they represent different kinds of marked directives, and lá contributes to explicitly highlighting the fact that the imperative should not be understood as a plain order. Consider the following example:

In example (13), the presence of the adverb modifies the illocutionary force of the speech act, which sounds softened, and contributes to specify the directive as an exhortation. This seems to be related to the deictic value of lá, which can be contextually exploited by speakers to express distancing from their own speech act: thus, it can be reanalyzed as a form that signals a decreased illocutionary force and a cooperative attitude towards the interlocutor. This way, directives with lá often appear in codified interactional patterns such as interviews, teacher-student talks and similar (see Marques & Duarte, 2017, pp. 30–32).

More in general, lá marks directives where some component of the speaker’s point of view with respect to the uttered proposition is made visible or salient, and is thus responsible for activating inferences that would not be explicitly available without it. As Martins (2012, p. 221) puts it: “The presence of non-argumental lá in an imperative sentence adds imperative force in a polite and cooperative manner, thus signaling the speaker’s attitude towards the prejacent proposition”. This way, imperatives with lá represent directives that do not come out of the blue but are related to a specific conversational background: the adverb explicitly signals that the speech act must be interpreted against these contextual conditions. Consider the following example:

In example (14) a girl tells of a conversation that she had with her mother and a friend of her mother. The topic concerns the choice of university studies: she would like to study psychology, but her mother is concerned about the difficulties of finding a job after that degree. The reported conversation features two imperatives with lá. The first one (vê lá ‘make sure/be careful; lit. ‘see there’) sounds like a challenge or warning, while the second one is an exhortation or encouragement (põe lá a psicologia ‘put ptc psychology)’. Interestingly, in this example the contextual conditions that allow the correct interpretation of the directives are made explicit right after them (não te esqueças daquilo que lhe estás a dizer after the warning; que é o que tu queres after the exhortation). This supports the idea that imperatives with lá represent a marked class of directives, whereby the adverb relates the speech act to a specific conversational background (which may or may not be made explicit by the speaker).

In the dataset used for this study I identified imperative forms of fourteen verbs modified by lá, for a total of 51 occurrences (see Table 2 below). Similarly to what has been noticed about un po’, some of these forms appear to be much more frequent than others: they represent highly conventionalized sequences that express specific communicative routines (desculpe lá) or operate as discourse markers (diz lá, olha lá), as in example (15) below.

Finally, just as in the case of un po’, the adverb lá can contribute to marking a specific type of illocutionary force, namely hortatives:

4.2 Assertive speech acts

Context-level uses of lá also appear in another illocutionary context, namely assertions, which are commonly expressed by declarative sentences. While in imperative sentences the adverb appears in the postverbal position, it occupies the preverbal position in declarative sentences, a fact that possibly points to separate development paths.¹⁸ At a general level, the presence of preverbal lá marks the declarative sentence as a specific type of assertion, one that presupposes some kind of predictability in the performance of the speech act and some degree of expectedness about the proposition conveyed by it (see Duarte, 2009, p. 191). This can be interpreted as a modification of the illocutionary features of the speech act. This way, the utterance must not be understood as a plain assertion, but rather as an assertion that conveys additional subjective and inferential values: “Declarative sentences with emphatic lá denote the speaker’s attitude towards the prejacent proposition either by reinforcing the assertive force or by adding a comment on top of the mere assertion of the proposition” (Martins, 2012, p. 222).¹⁹

The speaker’s comment signaled by the presence of lá allows for different conversational effects (at times very elusive), and its interpretation heavily relies on contextual factors. This way, assertions with preverbal lá can express a wide range of subjective attitudes, such as annoyance, disappointment, relief or inevitability with respect to the proposition conveyed by the speech act. Consider the two following examples:

In example (17), a professor is talking about the well-being of school pupils and mentions conflict among them; in example (18), someone talks about a strike where separate groups of workers have different opinions on how to proceed. In both examples, the assertion with lá stands out within the discourse flow: lá vão atrás dos colegas ‘they go ptc after their classmates’ in (17); lá andaram nas conversações ‘they have been ptc negotiating’ in (18). In particular, the presence of lá signals a subjective evaluation on the part of the speaker towards the proposition conveyed by the speech act: in (17), it adds a sense of inevitability; in (18), it conveys some kind of disappointment at the fact that the negotiations have lasted longer than expected.²⁰

As in the case of directives, some assertions with lá have reached the status of routinized expressions (for instance: lá está, lit. ‘there it is’; já lá vai, lit. ‘there it goes already’). These conventionalized sequences can express broader discursive functions (confirmation, justification) in argumentative patterns. In example (19) below, the fact that the presence of lá enriches the utterance with an overtone of inevitability is exploited by the speaker for argumentative purposes. In this case, lá está operates as a discourse-marking element that scopes over a whole discourse chunk. The speaker affirms that she likes Provence and, shortly after, that she likes Southern French people: the second utterance is preceded by lá está, and thus acquires the function of a justification of what has been said before. Thus, the fact that she likes southern French people is presented as a prior argument that inevitably leads to the fact that she likes Provence.

Going back to example (18), another occurrence of lá should be noticed (lá as reivindicações ‘the demands there’): in this case, lá has scope over a noun phrase (NP). Functionally, although the adverb partially keeps its locative meaning, the distancing value should be understood as a way of expressing a subjective stance rather than a spatial reference. Consider the following example, where the adverb also has syntactic scope over a NP:

In example (20), the presence of lá marks the fact that the NP a senhora only approximately matches the referent that the speaker would like to evoke: when used with this function, lá can be interpreted as a marker of approximation.²¹ This function probably develops when the distancing value of locative lá is transferred from the representational to the interpersonal level: this way, the presence of lá can express the distancing enacted by the speaker with respect to their own assertion.²² Depending on the target, the approximation value can target parts of the communicated content as in (20) above, or the whole of it as in (21):

4.3 Common features of the context-level uses of lá

Overall, the distinction between locative and non-locative uses of lá should be understood as a continuum of content-level and context-level uses, which in some cases leaves space for underdetermination between the two interpretations (see Marques & Duarte, 2015, pp. 119–121). Looking at the discourse-pragmatic functions of lá examined so far, the connection with the prototypical locative value can be explained in (at least) two ways.

On the one hand, the deictic value of the adverb, which can express both spatial and textual deixis (Marques & Duarte, 2017, pp. 21–23), must be considered. In particular, the deictic potential of lá can be exploited to point to propositions which do not explicitly appear in the text, but are rather implicitly active in the common ground shared by the interlocutors. Thus, the adverb does not refer to textual chunks, but rather to contextual inferences, and it can be locally reanalyzed as a marker of interactional salience. This process seems to be particularly relevant for the illocutionary uses of lá, as soon as the deictic potential of the adverb is transferred to the interpersonal level and paired with the subjective attitude of the speaker. Directive and assertive speech acts with lá are thus marked as non-initial speech acts, as the presence of lá signals that the speech act must be interpreted with reference to some element of the common ground or a codified interactional frame. This way, the presence of lá contributes to specifying how the speech act should be interpreted in the relevant interactional context.

On the other hand, the distancing value related to locative use of the adverb (‘there’) must be considered. In particular, the expression of spatial distance can be paired with a subjective perspective so that lá marks “some sort of distancing between the enunciator and the situation described (‘nothing to do with me’)” (Lejeune & Mendes forthc., p. 43; see also Marques & Duarte, 2017, pp. 28–30). In particular, the speaker can use lá to mark distance towards their own speech act or distance towards the communicated content of their own utterance: this can lead to the marking of approximation on parts of the communicated content, or the expression of a low degree of confidence with respect to the whole message conveyed.²³

Importantly, as already mentioned, the functional variation of lá is partly reflected by formal distinctions. As a marker of illocutionary modification, the adverb occurs in the postverbal position in directive speech acts (imperative sentences), while it occurs in the preverbal position in assertive speech acts (declarative sentences). As a marker of approximation, it shows more flexibility, as it can occur both immediately before and after the phrase over which it has scope, as well as in other positions of the sentence. In particular, when it takes scope over the whole utterance, it mainly occurs after the finite verb form, thus showing a formal difference from the illocutionary use of lá in assertive speech acts. This way, the fact that the various context-level uses of lá show formal differences between them reflects the fact that they express different functions at different grammatical layers.

5.1 A comparison of the discourse-pragmatic functions of un po’ and lá

Having described some of the context-level uses of un po’ and lá, I will now highlight a few similarities and differences between them, and resume thereby the discussion on specific theoretical points. As it became clear through the description of the examples, the comparison between the two adverbs is not based on an exact equivalence of their meanings, but rather on their grammatical status and broader functional issues. These can be summarized in three main points.

First, both adverbs are polyfunctional: in particular, they show a coexistence of content-level and context-level uses. The first group is represented by uses of un po’ and lá that modify the propositional content of a sentence, while the second group is represented by uses of un po’ and lá that target the pragmatic features of the utterance. Considering content-level uses, un po’ operates as an adverbial degree modifier and as a quantifier in partitive constructions, while lá operates as a deictic locative adverb. Considering context-level uses, both adverbs can perform a wide range of functions: from discourse marking to illocutionary modification and several information-managing functions. This kind of polyfunctionality has been observed cross-linguistically for many adverbial elements, and has been discussed by several studies about semantic change and the semantics/pragmatics divide (see Ariel 2008; Hansen, 2008; Traugott & Dasher, 2002). Moreover, looking at the frequencies of content-level vs. context-level uses of the two adverbs in the dataset used for this study (see Table 3 below), it can be noticed that the first group largely outnumber the second one: in this perspective, this distinction should also be understood as a distinction between prototypical vs. non-prototypical uses.

Second, some of the context-level uses of un po’ and lá show a comparable speech-act distribution, namely they appear in similar illocutionary contexts. In this study, I presented the most important contexts of occurrence that are common to both adverbs: directive, hortative, and assertive speech acts.²⁴ This way, the illocutionary context can provide a baseline both for the classification of the context-level uses of these adverbs and for their functional description. Moreover, in hortatives and directives, un po’ and lá appear in a set of constructions that perform equivalent discursive functions: particularly noteworthy are a few highly conventionalized verb-particle sequences (vediamo un po’/vamos lá ver ‘let’s see ptc’; guarda un po’/olha lá ‘look ptc’; dimmi un po’/diz lá ‘tell me ptc’) that are used in the same interactional patterns.²⁵

Finally, the context-level uses of both adverbs are related to the same functional domains: illocutionary modification and information management. The first domain accounts for the actual illocutionary uses (those expressing modification of the illocutionary force and speech-act specification in an interpersonal perspective), while the second domain accounts for the uses that relate to the speaker’s handling of the information flow (those expressing approximation, focus marking, mirativity, and metalinguistic negation). This fact represents the most striking functional similarity between the context-level uses of un po’ and lá, and leads to relevant theoretical issues. On the one hand, this data can be used to discuss the respective features of these functional domains and the interplay between them. On the other hand, the theoretical discussion can lead to a better definition of the grammatical status of the context-level uses of un po’ and lá.

5.2 The interplay between illocutionary force and information management

In Section 2, I referred to the framework of Functional Discourse Grammar to discuss the notions of illocutionary force and information management, linking them respectively to the grammatical layers of the illocution and the communicated content. The distinction between these two layers is treated in FDG as a distinction pertaining to the interpersonal level, which deals with the interaction between the speaker and the addressee (Hengeveld & Mackenzie, 2008, pp. 46–48). Linguistic expressions operating at this level have no designating force: they relate to the attitude of the speaker towards the information that they are transmitting rather than to the propositional content of the sentences in which they occur. In Table 4, the functions related to four different grammatical categories at the interpersonal level (mood, negation, evidentiality, mirativity) are shown.

As regards this study, illocutionary modification and approximation (and to a lesser extent, mirative and metalinguistic negation) are the functions that played a major role in the analysis of un po’ and lá. The first element can express illocutionary modification at the layer of the illocution, as well as approximation and mirativity at the layer of the communicated content. The second element can express illocutionary modification at the layer of the illocution, as well as approximation and metalinguistic negation at the layer of the communicated content.

Moreover, the description of the examples also showed that while some context-level uses of these adverbs clearly relate to a single grammatical layer, others are fuzzy with respect to the functions expressed and, consequently, the relevant grammatical layer. In directive speech acts, due to their distinctive illocutionary features, context-level uses of un po’ and lá are mostly connected to illocutionary force modification. In assertive speech acts a greater interplay between the two layers can be observed.²⁶ In particular, several examples of un po’ and lá in assertive speech acts are ambiguous as to whether they express approximation at the layer of the communicated content or low degree of assertivity at the layer of the illocution. This is not surprising, as Hengeveld & Keizer (2011) point out:

It would be worthwhile to see to what extent these distinctions are relevant at other layers of the Interpersonal Level (the Illocution or the Discourse Act) as well. However, since the same expressions are often used to mark non-straightforwardness at the different layers, it is not always easy to determine to which level these expressions apply. For the same reason, it turns out to be difficult to distinguish between approximation/exactness on the one hand and mitigation/reinforcement on the other. (Hengeveld & Keizer, 2011, p. 1975)²⁷

Overall, the interplay between illocutionary force and information management reflects the contiguity of the illocutionary and the locutionary features that together constitute a speech act. It is therefore reasonable to expect that, according to the specificities of the context of occurrence, the same operator can target one layer or the other. Moreover, the expression of the speaker’s subjective attitude plays a role both in the marking of the illocutionary force and in the managing of the information flow. From this perspective, in the absence of clear formal features that distinguish the use of the same element at different layers, some degree of scope and functional vagueness can be assumed.²⁸

5.3 un po’ and lá as Romance modal particles

One last issue to be considered concerns the grammatical status of the context-level uses of un po’ and lá, and the question of whether they can be assigned to a specific grammatical category or word class. In the FDG line of research, only a few works have specifically addressed the cross-linguistic distribution and the specific features of grammatical operators at the layers of the illocution and the communicated content, and more work is needed on this topic (see however Fang & Hengeveld, 2022; Hengeveld & Keizer, 2011; Vismans, 1994). Moreover, the distinction between the illocution and the communicated content is not always comparable with the categories adopted by other frameworks. Nevertheless, a few similarities with other approaches can be highlighted. In particular, the functions expressed by the context-level uses of un po’ and lá closely resemble the functions of modal particles in German, Dutch and other languages (see among others Abraham, 1991; Artiagoitia, Elordieta & Monforte, 2022; Waltereit, 2001). Modal particles are syntactically integrated elements that have scope over sentences. Functionally, they fine-tune speech acts by specifying the illocutionary force. Modal particles mark a relationship between a speech act and some element in the common ground, usually a belief (‘a proposition’) attributed to the addressee. This way, they contribute to accommodating a speech act to the relevant interactional context (Detges, 2015, p. 131).

From this perspective, the functions of un po’ and lá in directive speech acts can be considered in all respects modal-particle uses of these adverbs: they specify the illocutionary force of the speech act, and relate it to specific common ground structures and interactional patterns. Moreover, since they always occur in a fixed position of the sentence, close to the main verb, they are syntactically integrated elements. Similarly, the particle-verb construction with lá in assertive speech acts also represents a modal-particle use of this element, which marks a specific class of assertions.²⁹ Given this, the illocutionary uses of un po’ and lá can be included in the inventory of Romance modal particles (see Remberger, 2021; Squartini, 2014, 2017; Waltereit, 2001, 2006). By contrast, the uses of un po’ and lá that express approximation should be probably better kept apart, as they are related to more general hedging strategies that have scope on the communicated content (see Kaltenböck, Mihatsch & Schneider, 2010).