PROSODY, RELEVANCE AND MANIPULATION IN POLITICAL DISCOURSE

: The aim of this paper is to examine the role of intonation in political discourse in River Plate Spanish as a means to manipulate audiences into accepting questionable assumptions. The analysis of intonation is carried out in the Autosegmental-Metrical model applied to Argentinian Spanish and its interpretation is made in the framework of Relevance Theory. Some extracts from three interviews of two politicians and a presidential address are analysed, using PRAAT (BOERSMA; WEENINCK, 2020), a software for speech analysis, and interpreted using the Sp_ToBI transcription system. The analysis shows that level and rising intonation, often associated with background information, can be used to indicate, together with other linguistic devices, that information is to be processed as forming part of the common ground shared with the audience, and thus beyond questioning. In this way, information favourable to the speaker is made more accessible and attention is diverted from critical information.


INTRODUCTION
It is a fact that the different ways in which speakers can say an utterance makes a major contribution to the way they intend that utterance to be interpreted.The continuous variations of pitch, length, loudness and voice quality which are superimposed on words affect the way they are processed, and the same utterance with different prosodic features may be interpreted in different ways.Prosodic features provide different types of input to the comprehension process: some of them are 'natural' and may express states of mind like anger, agitation, etc.; others are properly linguistic, such as lexical stress or tone (WILSON; WHARTON, 2006).In other words, prosody has a dual nature: it is partly iconic, conveying paralinguistic meanings, and partly arbitrary, with language-specific features of a phonological nature (HOUSE, 2006).Both types of meaning make an effective contribution to the speaker's intended meaning.
For a long time, most of the investigation into the relation between prosody and speaker meaning has been carried out by phonologists, and progress has been made in the formal analysis of prosodic structure and its relation to meaning (WHARTON, 2012).More recently, this interaction has been approached by pragmatists, especially in the wake of Paul Grice's influential views.Grice (1989) proposes that an essential feature of human communication is the expression and recognition of intentions: the speaker's intended interpretation is inferred by a process of hypothesis formation and evaluation, guided by general principles of communication that communicator and audience are supposed to follow.The question then arises as to how different prosodic features, and intonation in particular, contribute to this process.
This paper explores the function of different intonation patterns in the framework of Relevance Theory, a cognitive pragmatic theory based on Grice's central claims on the role of decoding, inference and relevance in communication.In particular, it applies the idea of procedural encoding to the analysis of intonation in political discourse.
Political discourse is a form of action meant to exert symbolic pressure on the public to legitimize political power.The power of discourse emerges through the control of the belief systems of individuals and groups.It is argumentative, essentially polarized and polemic (VERÓN, 1987).Van Dijk (1997, p. 28) regards political speech as evaluative, and postulates that all ideological and political discourse follows the strategic principle of the Ideological Square, which consists of emphasizing OUR good actions and THEIR bad actions, and of de-emphasizing OUR bad actions and THEIR good actions.Each argumentative move in political discourse follows the overall principle of positive selfpresentation and negative other-presentation.We analyse how intonation contributes to manipulating audiences in order to control their belief-systems, and indirectly their actions, on the basis of those beliefs ( VAN DIJK, 2006).

Página423
The paper is organized in the following way.In section 2, we introduce the pragmatic and prosodic framework.In section 3, we present the corpus and methodology.In section 4, we analyse the data from a pragmatic point of view.In section 5, we evaluate our analysis in the light of other cognitive pragmatic work on political discourse.Finally, in section 6, we summarize the contribution of intonation to information processing.

PRAGMATIC FRAMEWORK
The theoretical framework for this paper is Relevance Theory (RT), a cognitive inferential pragmatic theory (SPERBER; WILSON, 1995;WILSON;SPERBER, 2004).It is basically a generalization about the human cognitive system which forms the basis of a communicative theory of utterance interpretation.Any external stimulus, internal representation or utterance (information) may be relevant when it combines with a context of background assumptions the individual already possesses and yields positive cognitive effects, that is, changes in an individual's representation of the world.These effects are mainly of three types: contextual implications, or conclusions that cannot be derived from new information or from the context independently, but only from a productive combination of both, revisions leading to the strengthening of background information, and revisions leading to the weakening and eventual abandonment of background information.The more cognitive effects these stimuli or representations trigger, the greater the relevance.But cognitive processing involves effort of perception, memory and inference.The greater the effort, the less relevant the information will appear to be; the lesser the effort, the more relevant it will appear to be.Greater effort will be acceptable only when it leads to greater cognitive gain.Relevance results from the contextualization of a stimulus, mental representation or utterance in a context which is not given in advance, but which is selected ad hoc on the basis of its accessibility to maximize the relevance of the utterance.
Any pragmatic theory makes assumptions as to the information the participants in a conversation must share for communication to be possible.RT proposes the concept of mutual cognitive environment.A cognitive environment is the set of facts an individual can entertain in their mind and accept as true or probably true.In a mutual cognitive environment, for every fact that an individual entertains, they also entertain the fact that their interlocutor entertains it (SPERBER; WILSON, 1995).

Página424
According to Relevance Theory, inferential communication involves two layers of intention: the informative intention (the intention to inform an audience of something) and the communicative intention (the intention to inform the audience of one's informative intention).Understanding is achieved when the communicative intention is fulfilled, that is, when the audience recognizes the informative intention.The informative intention must be recognized as such, but it may or may not be fulfilled, because its fulfilment depends on how much the audience trusts the communicator.In other words, there is difference between understanding and believing (WILSON; SPERBER, 2004, p. 611).

PROSODIC FRAMEWORK
The prosodic framework of analysis adopted here is the Autosegmental-Metrical model (AM, PIERREHUMBERT, 1980;LADD, 2008), initially applied to English and later used in the analysis of Spanish and other languages.Intonation contours result from the concatenation of H (high) and L (low) tones, which form two different types of unit: pitch accents (T*), which are associated with the tonic syllables (those that carry stress) and edge tones (T%), associated with the margins of prosodic constituents at different levels of the prosodic hierarchy.Edge tones can also end at a mid level (M%)1 .In this paper we concentrate on tonal movements at the end of the intonation phrase, the socalled nuclear pitch accent.The nuclear pitch accent and the boundary tone form the nuclear tone configuration.The meaning of the nuclear configuration applies to the whole intonation phrase of which it is part.
The AM model is applied to the analysis of intonation in Spanish through the Sp_ToBI transcription, a notational system for the different varieties of Spanish (PRIETO; ROSEANO, 2010;HUALDE;PRIETO, 2015).Argentinian Spanish (River Plate variety) intonation has been analysed in experimental conditions in Colantoni and Gurlekian (2004), and Gabriel et al. (2010, 2011, 2013), among others.Spontaneous production in this dialect has been approached in Labastía (2011, 2016, 2018), and Dabrowski and Labastía (2013), among others.
According to Gabriel et al. (2010Gabriel et al. ( , 2013)), there are three falling nuclear configurations in Argentinian Spanish: H+L* L% (↘high falling), L* L% (↘low falling) and L+H*+L L% (↗↘rising falling).These authors propose that the first two configurations appear in broad focus statements, while they associate the third with narrow focus, contradiction, emphatic and exclamative statements.In the analysis of these in spontaneous discourse, the three configurations indicate that the content in those intonation phrases is relevant in itself and orient the hearer to the derivation of the different cognitive effects proposed in RT.H+L* L% is associated with the derivation of contextual implications, while L* L% conveys a high degree of certainty on the part of the speaker and qualifies as a categorical assertion, and L+H*+L L% indicates that the content is highly relevant and may be taken as a reinforcement or a contradiction of background information (LABASTÍA, 2016).

Página425
As regards the level nuclear configuration, Gabriel et al.'s proposal (2010) does not assign it a specific meaning.In spontaneous discourse data, this configuration often appears in incomplete statements and is used to signal to the hearer that he should hold off investing processing effort until the speaker instructs him to do so through a falling pitch configuration (LABASTÍA, 2011), because, with the intonation phrases with a level configuration, he is in the process of building a larger discourse structure.With this configuration, the speaker is providing the context in which the point of the utterance will achieve relevance.The information in those intonation phrases is to be incorporated into the mutual cognitive environment, but it is to be considered as background to the relevant information, which usually follows it and is intoned with a falling configuration (LABASTÍA, 2018).
Besides the falling and the level nuclear configurations, there are also other configurations which end at a mid or high level, which we have labelled as 'rising'.In Gabriel et al. (2010) L*M%, or the low rising configuration, appears in uncertainly statements, while L+H* HM% appears in echo wh-questions, imperative and confirmation yes-no questions and in certain types of vocatives.In Labastía (2018), four rising configurations are identified, L* M% (↗low rising), L+H* H% (↗high rising), L+H* LM% (↗↘↗rising-falling-rising)2 and L+H* HM%3 (↗→↘rising-rising-falling), which also appear in incomplete statements.They combine nuclear tones with a high or mid boundary tone.These nuclear configurations can also appear in incomplete utterances, and the processing instruction seems to be, as in the case of the level configuration, to postpone deriving cognitive effects because the speaker is building up a larger discourse structure.But, unlike the level configuration, the rising configurations seem to mark the content of the intonation phrases in which they appear as already present in the context.Consequently, they may operate as a reminder of background information which is already present in the cognitive environment, and which is made accessible to reduce processing costs.According to Labastía (2018), the different rising configurations (those ending in H% or M%) encode different degrees of accessibility of background information, from more to less accessible: L* M%, L+H* H%, L+H* LM%, L+H* HM%.The nuclear configurations in River Plate Argentinian Spanish and their procedural meaning are summarized in Table 1.
In the procedural view of intonation, the use of these tonal configurations does not necessarily coincide with the objective conditions of discourse.Each configuration indicates a piece of computational information.That is, the speaker does not necessarily mirror the context as it is, but rather presents the information as she wants it to be perceived by the hearer (ESCANDELL-VIDAL, 1998, p. 200).
The different intonation phrases are grouped together in a larger unit, which we call 'sequence' ('processing unit' in House, 1990).The first phrases in the sequence tend to have mid-level or rising configurations, and indicate the information is to be considered part of the background, while the final phrases usually have falling configurations, indicating they are part of the foreground, and contain the point of the utterance.The sequence usually starts on a relatively high pitch and progressively descends to a low pitch level at the end.Example (1) shows part of a sequence in an extract from the corpus Página426 (Macri's interview, 2021).Slant bars separate intonation phrases and syllables in capital letters mark prominences.The underlined syllable indicates the beginning of the nuclear pitch configuration.In this example, intonation phrase 1, with a low rising configuration, provides the context in which phrase 2, with a low falling configuration, achieves its relevance.Figure 1 shows the acoustic analysis of these intonation phrases, made with the help of PRAAT, a free-access speech analysis programme (BOERSMA; WEENINCK, 2020).Figure 2 shows the schematic representation of a sequence.This proposal is in keeping with Sperber and Wilson's (1995, p. 202-217) observations about prosody and relevance.They consider that it is natural for information that the speaker presents as given or uncontroversial to come before new information, in the same way as "it is natural to raise a question before answering it, or to communicate a complex piece of information step by step" (SPERBER; WILSON, 1995, p. 211).Background information is information that contributes only indirectly to relevance, by reducing processing effort; it need be neither given nor presupposed.Foreground information is information that is relevant in its own right by having contextual effects; it need not be new (SPERBER; WILSON, 1995, p. 217).Indicating the relative relevance of different pieces of information helps reduce the hearer's cognitive effort.

CORPUS AND METHODOLOGY
In this paper, we would like to analyse the information these politicians present as background with the help of intonation and the way they (intend to) guide listeners to certain assumptions in which to process the main point of the utterance.We would like to show that backgrounding and foregrounding with the help of intonation plays an important role in the interpretation of discourse, and in particular political discourse.Backgrounding may enable politicians to insinuate information as if it had already been consented to as part of the mutual cognitive environment shared with their audience.This could be a mechanism to attempt to manipulate audiences into inadvertently adopting their viewpoint.
The video interviews were downloaded from Youtube.We watched the video recordings of the four events and selected some fragments which we found useful to put our hypothesis to the test.We transformed the video into WAV audio files and extracted these fragments as 44.100Hz, 16 bit mono files and analysed them with Praat (BOERSMA; WEENINCK, 2020).We used the Sp_ToBI labelling system applied to

Página429
Argentinian Spanish (GABRIEL et al., 2010(GABRIEL et al., , 2013) ) with some modifications resulting from the analysis of spontaneous speech (LABASTÍA, 2018).We submitted some key examples to the scrutiny of three colleaguesexpert phoneticianswho agreed on 90% of the cases with respect to the non-falling/falling contrast.
In the analysis of the examples, we are guided by intonation to identify the foregrounded and backgrounded information.On the basis of phonological organisation, we identify the central point of the utterance, the response to the journalist's question, and focus on the information in phrases with level or rising configurations preceding the central point.We present examples as pairs in which the information backgrounded by one politician is in conflict with the one backgrounded by the other.
In the transcription of intonation, we follow the same conventions explained in 2.2.The translation of each phrase can be found below the Sp_ToBI analysis.The backgrounded intonation phrases are marked in black letter type.

ANALYSIS OF THE DATA
In example [2], at the beginning of the interview, journalist Carlos Cué asks Cristina Kirchner about the situation Argentina is in.She considers that the country is not doing well, and explains why this is so.

Página430
Kirchner organizes her answer in two sequences: the first one groups intonation phrases 1-3, and the second one, phrases 4-14.The focus of Kirchner's answer lies in intonation phrases 3 and 13-14, with a falling configuration, where she declares that the indebtedness level in Argentina is 'staggering', and where she claims that the level of deleveraging during her administration had not occurred since 1976.Instead, in phrases 1 and 2, with a level and rising configuration, she presents as background the fact that under the present (Macri's) administration, the country is getting heavily into debt.Then in phrases 4-12 (with the exception of 11), once again with level and rising configurations, she presents as background the fact that her administration saved the country from default and led Argentina to deleveraging.The information Cristina Kirchner presents as background (with non-falling configurations) constitutes a defence of her achievements as an administration.Besides, intonation phrase 4 is introduced with a conditional clause at the beginning of the sequence, in which she suggests that this fact is undeniable and has to be admitted as such.This also favours its interpretation as background.Cristina Kirchner's backgrounded information helps her to mark a clear contrast with the current administration: in this way, she is able to distance herself from the alleged current trend of financial indebtedness.The (foregrounded) negative evaluation of her political opponent's management is based on the (backgrounded) positive evaluation of her own administration concerning indebtedness.This move is meant for Kirchner to present herself as someone who has not incurred the same errors as the current administration, and so is likely to be an alternative to it in future.
In example [3], the then president Macri uses background intonation to introduce an aspect of Kirchner's administration which clearly contrasts with her positive selfpresentation in [2].The interviewer has just asked him why he waited for two years to announce his administration's policies.Macri answers in the following way.In this extract, Macri organizes his answer in two sequences, the first one including phrases 1-6, and second one phrases 7-11.The point of Macri's utterance in this example is expressed in intonation phrases 6 and 11 (with falling intonation): we were living at a level of disorder which could not be measured even then, and what we were living at the time of the previous (Cristina Kirchner's) administration was absolutely false.These phrases answer the interviewer's question as to why Macri took so long to explain his plan.The rest of the phrases in this utterance have level or rising configurations.What is being backgrounded here is the claim that Macri had inherited this covert crisis (phrases 3 and 4) that was leading the country to the edge of absolute disaster (phrase 9), and that his delay in announcing his policies was due to the fact that a transition was necessary for the world to be able to become fully aware of the depth of the plight the country was in (phrase 10).The fact that this covert crisis was inherited from the previous (Kirchner's) administration in phrases 3 and 4 are the subject/topic of the sentence, and, as such, also favour their interpretation as background.Macri's backgrounded information positions him in the role of saviour of a country on the brink of tragedy, and justifies his not having announced his policies beforehand.Macri is trying to garner support for his policies from those who might still be undecided (VERÓN, 1987), and at the same time to discredit his political opponent on the basis of the alleged chaos resulting from her administration.
Examples 4 and 5 introduce contrasting backgrounded claims concerning an aspect of Cristina Kirchner's administration: the management of the country's hydrocarbon resources.In example 4, the interviewer asks Cristina Kirchner why it is so difficult to leave positions of power, hinting at the fact that, after finishing her term as president, she has now run for the position of senator.Kirchner denies any attachment to power, and claims that it is a stressful job to be part of a government which makes decisions against the circles of power.This extract is organised as a single sequence.The relevant part of this utterance, marked with falling intonation (phrases 6-9), concerns the issue of the reaction her administration might have caused in the powers that be.In this section, Kirchner is trying to persuade her interviewer that she is not interested in retaining power because of the strain it demands from someone who defies those powers.In order to do so, she presents some contextual assumptions to provide evidence for the point she is trying to make by claiming two measures her late husband's and her own administration took: the recovery of pension funds and of the country's natural resources from private hands.These assumptions are expressed in phrases 1-5, pronounced with a level configuration, and are intended to be interpreted as the reason for Kirchner's administration not being accepted by the circles of power.One of these backgrounded claims is that her administration recovered the management of hydrocarbons for Argentina.The intented conclusion would be that leading a revolutionary administration with those achievements can't have been easy, and therefore she is not interested in continuing to wield that power, which has been a continual source of stress for her.Yet, she presents herself as a political activist whose role in politics has been and is to serve her people and country and defend it from greedy economic interests.
Example 5 is part of Mauricio Macri's presidential address in 2018, when, in the middle of a financial crisis, the then president explains the international events that contributed to the crisis.Among them, he mentions his predecessor's hydrocarbon policy:

Página433
In this extract, the first sequence groups intonation phrases 1 and 2, and the second, phrases 3 and 4. The issue which is relevant in this fragment, expressed in phrases 2 and 4 (both with falling intonation), is that at that time Argentina imported oil, and the unfortunate fact that the price of oil went up.This part of the assertion is contextualised in the points made in phrases 1 and 3: that this was a bad policy of the previous administration (i.e.Cristina Kirchner's), and that this event happened throughout the world, not just in Argentina.Phrase 1 introduces the concept of a bad policy as the reason for importing oil through the use of por (due to).With these two pieces of information, backgrounded with a level and a rising configuration respectively, Macri is disclaiming responsibility for this unfortunate situation, which is meant to partly account for the current crisis, and at the same time he is shifting the blame to the previous administration.In both cases, he attributes this knowledge to his audience ('and you also know that…'), as if this was already part of the mutual cognitive environment shared with them.This view is in clear contrast with Cristina Kirchner's contextual assumptions in example 4 (phrases 3 to 5), where she claims to have achieved a recovery of these natural resources for the country.Macri's purpose in this fragment is to show that he is not entirely to blame for what is happening to the country at the time of the speech, which results from a combination of her predecessor's bad policy and the current international situation.
In example 6, in discussing the difficulties that a politician has to face, Macri refers to an accusation made against him: that he had been involved in an eavesdropping operation on the opposition.

Página434
This extract is organized in three sequences: the first one includes phrases 1-4; the second groups phrases 5 and 6, and the third, phrases 7-11.The point of this utterance is Macri's experience of being unfairly accused and how he felt at the time.This is foregrounded (with falling pitch configurations) in phrases 4, 6 and 11.The context Macri provides in which to process this relevant information is expressed in 1-3, 5 and 7-10, all pronounced with level or rising configurations.The issue which he backgrounds is the alleged fact that this criminal case was invented by the opposition, and that he was unjustly accused of something which was false, and that the judges who heard the case played along.The conditional sentence introduced by means of cuando (when) at the beginning of the sequence may also contribute to this backgrounding effect in the second sequence.The overall effect he creates is one of presenting a background in which he appears as an innocent victim of the opposition's maneuvers to discredit him as a future presidential candidate.The contextual implication is that he was a powerless victim of an evil plot by Cristina Kirchner's administration, and at the same time deserves great respect for having endured this plight within the rule of the law, and shows his moral strength in the face of adversity.This also contributes to showing that he is in control of the country.
It is interesting to contrast Macri's claim in example 6 with Cristina Kirchner's claim in example 7.In this case, Carlos Cué asks Cristina if she thinks she could end up in jail, since she has been charged with corruption.Kirchner refers to cases of people who were allegedly imprisoned unfairly, and answers in the following way:

Página435
This example is organized as a single sequence.The relevance of this utterance lies in the answer to the question posed by the interviewer, as to whether Cristina Kirchner may or may not end up in jail.The affirmative answer to the question can be found in intonation phrase 14.The rest of the utterance, a long subordinate clause headed by "when," which could be considered a temporal clause with a conditional value, describes the context in which her affirmative answer will achieve relevance.Without specifically referring to Macri's administration, she is introducing as background a context in which the rule of law is practically no longer in force, and in which justice has become a weapon wielded by the state to fight the opposition.Her veiled accusation is made patent by the use of the term task force, an armed force organized for a special operation, which equates Macri's government with a military dictatorship.As Macri does in example 7, here Mrs. Kirchner presents herself as the helpless victim of a violent government, thus showing that she is just on a par with the many who are suffering unfairly for not supporting the current administration, which is using the judicial system to its own advantage.
An interesting point to consider is that the only cases of level or rising configurations appear towards the end of this section, just before the answer, in phrases 10, 11 and 13, whereas the previous phrases are uttered with a low configuration (L* L%), whose meaning we described above as used to present foregrounded, relevant information as a categorical assertion made with a high degree of certainty.We think that these phrases could also have been uttered on a level or a rising configuration.The fact that this configuration appears in what we consider to be part of the background seems to contradict our characterization of this configuration as used for foregrounding.A similar case can be found in example 8, an extract from a recent interview of Macri (no longer president) by a board of journalists who question him about his responsibility for Cristina Kirchner returning to power as vice-president, and for him and Cristina Kirchner being responsible for the lack of national unity.Macri answers that the widening gap is a question of values, and accuses Kirchner of sequestering Peronism and of trying to destroy the republican system of government.The central point Macri makes in this extract, organized as a single sequence, can be found in phrases 8 and 9: the Kirchnerite branch of Peronism and Cristina Kirchner are there to destroy the republican system.The previous phrases can be considered to constitute the context in which to process these final phrases: phrases 1 to 7 combine the low configuration in phrases 1-3 and 7 with the level configuration in 4 and 5, and a falling configuration in 6.The context of these phrases introduces the reason why the Kirchnerite branch is in a position to break the system: they have captured Peronism, and the Peronist party is now in power.Consequently, it is Cristina Kirchner (at the time vicepresident) who is actually in power, not the president Alberto Fernández.As in example 7, we believe that these early phrases might as well have been uttered with a level or rising configuration.It is possible to relate the choice of L* L% (low) nuclear configuration to the fact that in both Cristina Kirchner's interview with Carlos Cué and Mauricio Macri's more recent interview with a board of journalists, neither Kirchner nor Macri were in power any longer, and the journalists adopted a somewhat aggressive, confrontational style which is absent from Macri's interview as president, in which the interviewer acted with special deference towards him as the highest authority in the country.In both cases, the interviewees reacted with categorical replies to defend their views and attack their political opponents from a position devoid of any privileges.In these phrases with L* L%, the status of the information as background might be given by its early position in the sequence rather by the nuclear pitch configuration and, in the case of example 7, also by its being part of a conditional clause early in the sequence.

DISCUSSION
Through the analysis of some fragments of political discourse, we have tried to show how the use of non-falling nuclear configurations help process information as contextual assumptions, in which other claims gain relevance by having positive cognitive effects.Backgrounding serves the purpose of the Ideological Square of positive self-presentation and negative other-presentation.With the intonationally backgrounded information, the speaker makes accessible or more accessible the premises for the hearer's intended conclusion and the fulfillment of the task at hand, which is usually set by the interviewer's questions.As we hope to have shown, the information Kirchner backgrounds is in sharp contrast with the one Macri backgrounds.This reveals the ideological bias each politician is trying to marginally convey.
In the analysis of examples 7 and 8 we have also noticed the appearance of L* L%, a falling configuration, where we would have expected level or rising configurations.We have attributed it to the interviewee's need to assert their views categorically, even in introducing the background assumptions, in the face of the interviewers' rather aggressive questioning.We have also found that the interpretation of part of the utterance as background does not depend only on the intonation used, but also on other factors, such as its initial position in the utterance, i.e. in initial intonation phrases in the sequence (cf.SPERBER; WILSON, 1995, p. 211), the appearance of certain nominalized expressions in subject/topic position, or the use of logical connectors like cuando (when), introducing conditions in subordinate clauses early in the utterance.
According to Wilson (2011, p. 20-21), a speaker who produces an utterance has two goals: to get the hearer to understand her meaning, and to persuade him to believe it.The hearer has two tasks: to understand the speaker's meaning, and to decide whether to believe it.The first task concerns the pragmatic ability to infer the speaker's intended meaning from linguistic and contextual cues.The second involves a capacity for 'epistemic vigilance', that is, the ability to avoid being accidentally or intentionally misinformed.This ability is related both to the assessment of the reliability of the source of information (its benevolence and competence), and reliability of its content in connection to background beliefs.
Within a massive modularist view of the human mind, Mercier and Sperber (2009) postulate the existence of two kinds of inferences: intuitive inferences, which result from the direct output of the inferential modules in comprehension and take place without attention to reasons for accepting them, and reflective inferences, which result from the operation of a particular metarepresentational module, the argumentation module, which takes the input of the comprehension module and subjects it to close scrutiny, and yields an argument for or against a given conclusion.Intuitive inferences are fast, automatic and mostly unconscious, relying on a fast and frugal heuristics of comprehension: following a path of least effort in constructing an interpretation of the utterance, the addressees will stop when their expectations of relevance are satisfied (WILSON; SPERBER, 2004, p. 613).As such, these mechanisms may lead to erroneous conclusions.We trust our mental processes to yield true conclusions, but in order to increase their performance, we take cognitive shortcuts.Sperber et al. (1995) affirm that we are "nearly incorrigible cognitive optimists."We take for granted that our spontaneous cognitive processes are highly reliable, but we are likely to sacrifice precision for the sake of efficiency.Instead, reflective inferences are slow, consciously controlled and effort-demanding, but make it possible to overcome the shortcomings of intuitive inferences.According to Mercier and Sperber (2009), humans rely massively on communicated information, and this reliance is a source of vulnerability to misinformation and deception.Epistemic vigilance mechanisms have arisen to filter communicated information in order to assess the trustworthiness of communicated information.
Speakers have linguistic resources to get past the hearer's epistemic vigilance mechanisms and convince him.Wilson (2011, p. 23-4) considers that, to achieve this, the speaker can produce an argument showing that her point follows logically from, or is strongly supported by other background information.These resources consist of logical and discourse connectives, as well as indicators of epistemic modality and evidentiality.In this perspective, called the argumentative theory of reasoning, the function of these resources would not be to guide the comprehension process, but to display the communicator's competence and benevolence, and to get the hearer to trust her.Backgrounding through intonation could be included as part of these resources.
De Saussure (2012) has introduced the idea of discursive presupposition, that is, assumptions that get accommodated as part of the assumed background knowledge without the hearer noticing it.They may be used for persuasion and manipulation.Discursive presuppositions tend to be processed at a low degree of awareness and may bypass critical evaluation.They are necessary conditions for relevance, but they are not grasped as the speaker's point in the utterance.They are economically incorporated because they are not evaluated by the relevance-checking procedures or by the epistemic vigilance controls.This is the basis for their persuasive power.They get spontaneously accommodated even when they are inconsistent with otherwise accessible information.Although de Saussure is referring here to a different notion, his description fits well with what we have discussed in relation to background information in intonation phrases with level or rising configurations.
Maillat and Oswald (2009Oswald ( , 2011) ) discuss the cognitive aspects of manipulation in a similar way.Manipulation exploits the cognitive dynamics of the search for relevance: finding the optimal ratio between cognitive effects and processing efforts.Due to economy constraints on efficiency, the addressee may not take into account all relevant information and take shortcuts, sometimes at the expense of consistency and logical validity.The manipulator may impose external restrictions on the interpretative process, so that the addressee is led to shallow-process certain contextual assumptions in order to optimize the use of cognitive resources, and consequently he interprets the utterance in a limited context.Manipulation also ensures that any alternative set of contextual assumptions is not accessed.A manipulative speaker will increase the accessibility of some contextual assumptions so that the addressee does not process the utterance in a larger context, where he might find inconsistencies or contradictions with other background knowledge or previously held beliefs he might entertain.Manipulation is an attempt at controlling the context selection process of an utterance by making some contextual assumptions so salient that they cannot but be processed, and thus prevent the argumentative module from applying critical information and bypass epistemic vigilance.According to Oswald (2016), this is a key feature for argumentative effectiveness.
It is our hypothesis that background intonation operates in this way: By making information accessible, it favours its interpretation as background assumptions, and may thus contribute to their shallow processing in such a way that they get incorporated as if they were already part of the mutual cognitive environment shared by communicator and audience.As such, it is likely to favour accommodation and acceptance without close scrutiny, and bypass the filters of epistemic vigilance.They may lead to the acceptance of certain propositions which may be at odds with the audience's background beliefs without their noticing.At the same time, it may prevent the audience from entertaining other assumptions which might lead them to question the validity of those premises.But, as we have seen, intonation does not, by itself, determine the interpretation of information as background.It operates in conjunction with other devices, such as logical operators (cf.conditional sentences in examples 6 and 7), early appearance in the sequence of intonation phrases, or other linguistic devices such as topicalisation (example 8).All the properties of the ostensive stimulus together help to set the inferential process in the track intended by the speaker.

CONCLUSION
In this paper, we have discussed the contribution of intonation to the processing of spoken political discourse, but we would like to suggest that the conclusions that we have drawn from it can apply to other types of spoken discourse as well.Our data indicate that backgrounding through non-falling configurations may be used to signal, together with other linguistic devices, that part of the utterance contains the premises in which relevant information may be processed and positive cognitive effects achieved.This resource may be used to marginally communicate information, and also to manipulate audiences into believing certain facts, as in the case of political discourse.The audience may be led to infer certain content without realizing it is problematic, and "its acceptance may help satisfy the communicator's covert perlocutionary goals" (OSWALD, 2014, p. 101).
The results of this work are also significant in the light of Sperber and Wilson's (1995) comments.As the communicator makes her informative intention mutually manifest, the audience may accept it or reject it.In face-to-face communication, the audience will be able to convey their acceptance or refusal to accept it.However, Where communication is non-reciprocal, there are various possible situations to be taken into account.The communicator may be in a position of such authority over her audience that the success of her informative intention is mutually manifest in advance.Journalists, professors, religious or political leaders assume, alas often on good grounds, that what they communicate automatically becomes mutually manifest.(SPERBER; WILSON, 1995, p. 63).
Of course, this does not mean that the propositions that get communicated will automatically become part of the mutual cognitive environment shared with their audience.Whether manipulation is actually effective or not depends on other factors related to epistemic vigilance: whether the communicator is considered trustworthy and benevolent in the first place, especially in relation to his/her past history as a source of information, and whether his/her pronouncements are actually subjected to deep scrutiny for consistency and accuracy, and pass epistemic vigilance controls through reflective inference, as Mercier and Sperber (2009) have pointed out.Besides, most, if not all of us, are prone to information bias, depending on our previously held beliefs, and will tend to accept or reject these pronouncements accordingly.
at the stressed syllable which rises to a mid-level until the end of the intonation phraseL+H* LM% ↗↘↗A rise to the stressed syllable followed by a fall and a rise to a mid-level towards the end of the intonation phrase L+H* HM% ↗→↘ A rise to the stressed syllable which continues into the following syllable and ends at a mid-level at the end of the intonation phrase is the last populist administration in our history /2 y que VENgan VEINte Años de creci↘MIENto / L+H* L+H* L+H* L* L% and there come twenty years of growth.

Figure 1 :Figure 2 :
Figure 1: Acoustic analysis of the end of a sequence with two intonation phrases in example 1.The panel above shows the first phrase, with a low rising nuclear configuration (inside the circle), and the one below the second phrase, with a low falling configuration (inside the circle).