Conceptual Integration in Natural Oral Narratives – María Dolores Porto & Manuela Romano

This paper belongs to a wider project1 that proposes to develop an integrative and cross-disciplinary line of research of natural narratives (oral and written) within the socio-cognitive models of language, models which promote the integration of the cultural component in the explanation of discursive dynamics. One of the research lines within this project has been to analyse the structure and grammar of emotionally charged oral narratives in Spanish and English. Until the moment, thus, the following conclusions have been drawn regarding the texts under study: (i) This kind of narrative, far from having a linear and causal structure, is constantly broken by comments, justifications, explanations, background information, side stories, etc. in order to create empathic bonds with the listener (Romano, 2008). (ii) Narrators use specific internal parameters or linguistic strategies to disclose highly intimate and painful information and attract the hearer’s attention to the most salient segments of the narrations -long pauses, deep respiration and clicks; the profusion of details; repetitions, redoings and repairings of information and the insertion of dialogue, among other (Romano, 2008). And (iii), narrators use specific pragmatic markers to guide the listeners’ attention through the string of stories, transitions and emotions, characteristic of this text type. These markers are, on the one hand, polysemous or polyfunctional, and, on the other, show the ideolectal or individual preferences of speakers (Romano and Porto, 2009; Romano and Porto, forthcoming). In this paper new research questions are addressed. In the first place, we wish to unravel how hearers cope with the chaotic information present in oral narratives of charged events. And, in the second, how the micro or linguistic level and the macro-level (the integration or blending of stories into a meaningful narrative) interact online in order to construct the final emergent story.

In order to answer these new research questions, the paper focuses on the following issues. First, in section 2, we introduce the corpus –Spanish radio oral narratives of highly emotional or charged events, as well as a detailed description of the narrative SpN3 ‘Hijo no quiere salir’ (Spanish Narrative 3: Son won’t go out).  Secondly, section 3 discusses how the theoretical concepts coming from Mental Spaces and Conceptual Integration Theory (Fauconnier 1994, 2008; Turner, 2008, Semino & Culpeper, 2002; Dancygier, 2007) can help to understand emotional oral narratives. In this section, thus, the notions of base space, input narrative space, space builder and narrative anchor are applied to SpN3 ‘Hijo no quiere salir’, in order to understand the main processes which enable hearers to make sense of the apparently chaotic information presented in this particular type of discourse by integrating their fragmented, non-linear structure into a coherent and emergent whole. In the third place, section 4 explains the advantages of incorporating central notions of attentional phenomena coming from Cognitive Grammar, such as windowing of attention (Talmy, 2000, 2007) to the narrative under scrutiny. And fourth and last, the study concludes that oral narratives of charged events can best be understood as complex and dynamic systems in which a series of incomplete and fragmented stories or spaces compete for the hearer’s attention. The emergent story arises through a gradually increasing network of emergent cross-mappings and blends and their interaction with our collective cultural knowledge.

The Narratives Under Study

The Corpus

The narrative under study, SpN3 ‘Hijo no quiere salir’, belongs to a corpus of Spanish oral narratives from an agony column radio programme in which people call to speak about their most intimate worries or problems. The data were recorded and then transcribed from the Spanish programme Hablar por hablar (Cadena Ser), a late night programme broadcast in Spain for the last ten years. 10 narratives, 5 recounted by men and 5 by women, were selected randomly for the study. No details about the narrators except gender and that they are all native speakers of Standard Spanish, were accessible due to the characteristics of the media and programme used for the sources. In addition, the average length of the narratives under study ranges from 300 to 600 words, including a total number of 4048 words for this analysis. This typology of programme was chosen because it contains natural and non-planned discourse in which speakers feel free to talk about their concerns in an anonymous setting and are therefore very close to natural language.

Let us summarize the most significant features of these texts and the context in which they are delivered, which will help to understand why the tools of MSCI can be so useful for their comprehension. (i) In the first place, compared to other similar agony column radio programmes, Hablar por Hablar does not show a very collaborative structure. People call either to share their problems with the listeners overtly or in response to previous callers seeking advice. In this sense, anonymous listeners can only provide indirect feedback by calling back later during the same programme or some days or even weeks later. The presence of the radio presenter is barely noticeable and passive, using continuers only to help narrators proceed with their stories and decide whether they have come to an end or not. The narratives are, thus, delivered with almost no interruptions. (ii) Secondly, because of the communicative setting, narrators and listeners, as complete strangers, share no common individual background knowledge, a fact that is also going to influence the linguistic features of this text type. We must bear in mind that these narratives are told only once by completely unknown narrators and, thus, there is no chance of retrieving lost information for clarification. These narratives, as shown previously (Romano and Porto forthcoming), contain structural and modal markers which guide the listeners’ attention throughout the different stories or narrative spaces while modulating the interpretation of the utterances. But, in addition to the different discourse and pragmatic markers, listeners have other socio-cognitive tools which are going to facilitate their ‘journey’ through the process: a shared and collective knowledge. As will be shown in section 3, Shank and Abelson’s 1977 concept of script, schemata for generalized episodes or events, comes in very handy to explain how listeners complete the missing information by using their previous knowledge and experiences. It would be impossible to interpret any kind of narrative without them, since there are always gaps in the stories, information that narrators take for granted. In Herman’s words, “what makes a story a story [is] the relation between the explicit cues included in a text or a discourse and the scripts on which readers or listeners rely in processing those cues” (2003, p. 10). (iii) In the third place, the oral narratives under study have been defined along a ‘continuum of narratives’ (Ochs and Capps, 2001; Romano, 2008), somewhere in between monologue and dialogue. Among the most salient ‘monologued’ characteristics of these texts we can point out that the recipient of the narration is not a prototypical listener since s/he is not present and therefore, cannot be as verbally active as the participant in a conversation and that there is just one single active speaker telling the story, with few verbal interruptions, commentaries or conversational-enhancers from the interlocutor. As for the more ‘dialogical, multiple, co-constructed’ features we can mention that they are still interactive texts because listeners are ‘present’ and the speaker feels that their presence matters. And that they show a highly fragmented, non-linear temporal and causal organization, which is constantly broken by different parenthetical material such as commentaries, digressions, flashbacks, flashforwards, etc., -contained in the many side-stories of the narratives and related to their highly emotional contents. Let us now have a closer look at a prototypical narrative: SpN3: ‘Hijo no quiere salir’.

SpN3: ‘Hijo no quiere salir’

SpN3 is about a mother who calls the radio programme for help because her son does not want to go out after his girlfriend died in a car accident two years ago. As can be seen below, the narrative consists of five stories or narrative spaces. The main storyline is, thus, broken or interrupted 9 times (lines 2, 7, 14, 18, 20, 27, 36, 38 and 52), as the minor stories are also cut several times throughout the narrative process. This means, that the main story is presented and recovered in 8 independent sections or chunks. See the structure and complete narrative below2 (see the translation into English in the Appendix):

narrative:

1 Tengo un hijo que… (click)

2 bueno, hace ya… como dos años que tenía una… una chica.

3 Se iban a casar y… ya tenían todo

4 y resulta de que mmm… la víspera de Reyes (respiración)

5 pues iba en el coche ella

6 y… tuvo un accidente y murió.

7 quería saber a ver que…

8 porque es que no hay forma de sacarle de casa… desde entonces.

9 Está totalmente mmm… metido en casa,

10 no quiere salir (respiración),

11 está agobiadísimo

12 y… y siempre igual

13 y no hay forma (respiración).

14 Entonces yo quería que mmm…

15 a ver si puede salir… y mmm…

16 Ya él yo ya creo que ya si…

17 si encontrara una buen chica que le… le gustara,

18 Porque yo hablo mucho con él… y él y…mmm… me cuenta (click)

19 pero (…..) no hay forma de hacerle salir a ningún sitio ni… (respiración).

20 Y es una pena

21 porque es mmm… un hijo tan bueno,

22 es un cielo (respiración),

23 muy trabajador… (titubea),

24 pero no… no… no… no quiere salir de casa,

25 está agobiado, siempre lo mismo, pensando… (suspiro)

26 y no hay forma.

27 Yo sí hablo muchísimo con él (respiración)

28 Y le y le digo  “Pero bueno, ya es hora de que…” mmm… pues tenga   una chica.

29 Tiene 38 años (respiración)

30 Y dice que… que  

31 “Sí, si yo… pero si encuentro una buena chica” (respiración)

32 “Pero…, es que, pero si no sales de casa cómo la vas a encontrar si la chi..

33 “Aquí, aquí no va a venir a buscarte” (respiración)

34 “Ya pero… es que no…”

35 No hay forma, no hay forma de que salga y mmm…

36 me gustaría saber a ver qué..

37 qué puedo hacer para que salga.

38 Si tenemos una parcela muy grande.

39 A su padre y a mí no nos deja que hagamos nada.

40 Él lo hace todo… (respiración):

41 limpia la piscina…, siega el césped…

42 Hace todo.

43 No (titubea), no piensa ná más que en eso,

44 o sea, pero de salir y de eso nada.

(P) ¿Cómo era antes del accidente él?

45 Pues él es alegre… y… y eso y mmm….

46 Tenía amigos y….

47 pero ahora mmm no,

48 es que es no es el mismo.

49 No, no, no hay forma.

50 Se mete en su habitación y así (respiración).

51 Cuando ya termina de mmm… estar en la…

52 No nos deja ni sacar la basura a la calle (muy rápido).

53 Él la saca, él… riega… Siempre está liao..

54 y… no piensa en otra cosa.

structure:

S1: My son just won’t go out (main story)

S2: His girlfriend died in a car accident two years ago (background information)

S3: I don’t know what to do, I need help (purpose for calling)

S4: We have a very close relationship, we talk a lot about this problem (expanding information)

S5: He is a very good and hardworking son, helps us at our country house, takes the garbage out, waters the plants, etc. (expandinginformation)

Let us now see how the analytic tools of Mental Spaces and Conceptual Blending Theory, as well as Attention Phenomena in language, can help to understand this structure and the process which enable listeners to build the final coherent story.

Blending stories in oral narratives

In order to explain how a narrative can be made up of so many fragmented pieces, we have carried out an analysis of the narratives from a Mental Spaces approach, also known as Blending (Fauconnier, 1994, 2008; Fauconnier and Turner, 2002; Turner, 2008; Oakley and Hougaard, 2008).

Blending, or rather Mental Spaces and Conceptual Integration Theory, is a model that aims to explain how people construct meaning by keeping track of currents of information. In analysing discourse, the theory claims that, as the discourse unfolds, a rich array of mental spaces is set up with mutual connections from one space to another, interacting and blending with each other, as well as activating new ones, in order to satisfy any number of expressive purposes. These mental spaces are only partial conceptual structures constructed as we think for purposes of local understanding and they do not constitute entire domains of knowledge. Instead, they make use of any previous general knowledge, personal direct experience, context, etc. as required by discourse in order to make sense of the information received. Therefore, they are framed, as they pick up elements from different frames and cognitive models we already know about (Fauconnier, 2007). When applied to fictional narrative (Dancygier, 2007; Semino, 2006; Turner, 2003; among others), the theory explains how a complete story emerges from the blending of several fragments and secondary stories provided by the text. It is our purpose in this paper to show that the same processes apply for non-ficitonal narratives.The tools of  MSCIT, we think, are especially useful to explain the online global construction of meaning in broken or chaotic oral discourse –as in the oral narratives of emotional events under study, since these texts will only be listened to once, and there is no possibility of retrieving/re-reading any missed information.

Mental Spaces and Conceptual Integration Theory is an overarching, quite complex theory and therefore it is beyond the scope of this paper to fully explain it. However, there are some basic concepts that will be especially useful for the present analysis. In the first place, input spacesare those spaces constructed from the information provided by linguistic expressions and which then blend into a new one, the blended space, which provides the final constructed meaning. The generic space, which contains some general knowledge that is projected onto the input spaces, allows the mappings that will lead to the blending. Since all spaces in a blend are interconnected, the connections made between them are the cross-mappings. The linguistic expressions that indicate that a new mental space is needed are called space builders. Finally, the base space is one of the input spaces, the one that serves as starting point for the developing of the network of mental spaces and which is also constructed by the linguistic expressions in the narrative (space-builders)3.

For our purposes, the different side stories that compose the oral narratives under study –flashbacks, comments, self-justifications, explanations, etc.–, are the input spaces or narrativespaces (Dancygier, 2008). Figure 1 is a representation of the narrative SpN3 “Hijo no quiere salir” from a Mental Spaces and Conceptual Integration Theory approach4:

In Figure 1, the several stories that compose the narrative (see section 2.2) are presented as input spaces. One of them, the starting point, is the base space, the here-and-now of the narrative, which is constructed by the very first sentence in the narrative (I’ve got a son who…) and then developed by other fragments (lines 8-13, 19, 24-26, 35, 44, 49-51, 54). Other input spaces are derived from this one (the background space, the consequences of that situation, the reason for calling, etc.). On top of them all, we have represented a genericspace, which explains the connections between the input spaces and makes the blended space possible. The generic space contains general knowledge on how things usually are, which is not usually expressed in discourse but belongs to our sharedcollective knowledge; that is, the knowledge by which individual minds and cognitive processes are shaped by their being together and interacting with other embodied minds (Frank 2008). It is in this sense that we can associate this notion of generic space to that of scripts or ‘experiential repertoires’ (Shank and Abelson, 1977; Herman, 2003; Bernárdez, 2008)5. Therefore, in the narrative under study, the content of that generic space is the common belief that young people must go out and socialise and if they do not do so, then something is wrong. Through the explicit cues included in SpN3, namely the repetition, up to six times (lines 8-13, 19, 24-26, 35, 44 and 49-51), of the idea that the narrator’s son has a problem because he doesn’t want to go out, we activate the main script or shared collective knowledge behind this narrative: healthy young people need to socialize. This information is implicit throughout the whole process, from the very opening of the narrative to its closing. Other  implicit scripts which help hearers construct the global final story are: healthy/good young people help their parents (lines 20-23, 38-43, 50-53), it is important to have a good relationship/talk a lot with your children (lines 18, 27-34) and you need a girlfriend/boyfriend to be happy (lines 28-31). All these will also act as generic spaces for other blendings that could be considered in a more fine-grained analysis.

As for the cross-mappings between spaces, they are mostly triggered by identity or analogy, that is, by the existence of the same elements in all of them. In our narrative, it is quite straightforward that the presence in all spaces of the narrator and/or her son is the most obvious link between spaces. However, we will see that there are also other narrative anchors that will help the hearer to link the input spaces. Finally, a blended space is represented in Figure 1 as an emergent structure that gathers elements from the input spaces but which is, at the same time, a new, different space that constitutes the global meaning of the narrative.

It must be noted, though, that all the input narrative spaces are not clearly set up at the beginning of the narrative. In fact, as already stated in section 2.2, some of these spaces are embedded into one another, so they are not actually at the same level as could be presumed from the way they have been depicted in Figure 1. As a matter of fact, Figure 2 would be a more accurate representation of the narrative if we want to take into account the embeddings.

Therefore, the spaces are not provided and complete at the beginning of the narrative. They are constructed by the hearer by connecting the several fragments present in the story as it develops and so are the embeddings, which do no always depend on textual clues alone (Dancygier, 2008, p. 64). In fact, we will see that the embeddings are mostly constructed through our shared, collective knowledge on the world, in the form of scripts, on which a part of the interpretation relies.

In order to help the listener construct the spaces and guide him/her through the fragmented spaces, the narrator will use several strategies. One of them is the use of different attentional markers, which act as space builders by signalling the starting or the end of a space. These markers include pragmatic and linguistic devices, such as discourse markers6, tense shifts, repetitions, sighs, clicks, breathing, repairs, etc. (signalled in bold in section 2.2). Another device to guide the listener is the use of narrative anchors (Dancygier, 2000) which help the hearer to link the different fragments of a space. Those narrative anchors can be linguistic expressions, or concepts, or ideas that are repeated or re-elaborated at different points in the narrative.  In the next sections, we will see how both strategies are used in the narrative under study; Hijo no quiere salir.

Pragmatic markers as space builders

Right after a first attempt to tell the main story, that is, to enter the base space (S1) in line 1, we find the first interruption. The main story or base space, that the narrator’s son does not want to go out, is suddenly cut in line 2 with a click and a chain of discourse markers. These markers introduce the painful background information that the listener needs to know and which constitute the second input space (S2) –her son’s girlfriend died in a car accident two years ago: bueno, hace ya… como… (well,.. about…, like..,). Then, in line 7, the first information on the narrator’s purpose for calling (S3) is inserted: she needs help. In this case, it is the conditional form of the verb in formal requests that is acting as a space builder, since the precedent information was in the past:quería saber… (I would like to know…). However, before she even finishes the sentence, S1 is recovered in lines 8 to 13 by means of a series of connected or compound markers: porque es que…, (because there’s just), which is closed with a deep breath in line 13. After that, lines 14 to 17 resume input space S3 –the narrator’s purpose for calling the radio programme–, again through the use of the Spanish conditional for requestsyo quería (I would like..) as a space builder, as well as through the markers entonces… mmm, a versi.., y mmm.. (then… mmm, lets see.., and mmm..). In line 18, the narrator introduces a completely new segment, which expands the information given to that point. The hearer constructs a new space (S4) cued by porque (because), mmm… and a final loud click that closes the space. It is worth noticing that porque, the prototypical causal connector in Spanish, does not introduce here any causal relation at the linguistic-discursive level of the text. Porque is used rather to open a new mental space in which the narrator is self-justifying her behaviour or relationship with her son, she thinks she is doing everything possible to help him. Line 19, signals a new transition and brings us back very shortly to S1 through the discourse marker pero (but), only to repeat the problem: there is just no way, no way to make him go out. Again, S1 is immediately interrupted by another side story, S5 –he is a very good son– in lines 20 to 23, whose function is to expand the background information about the son: he is the perfect son. This time, a mere y (and) builds the space. Lines 24 to 26 bring the listener back into S1, again by means of repeating the return marker: pero, no, no, no, no quiere salir de casa.. (but, no, no, no, he just won’t go out..). In lines 27 to 34, the listener is reintroduced into input space 4 (S4) –we have a very close relationship-, again in a rather sudden way, by introducing the first person singular personal pronoun yo, plus a series of deep breaths and direct quotations which intend to reproduce real conversations between mother and son and which help to justify the narrator’s behaviour within the whole situation: she is always talking to her son and trying to convince him to go out. Another return to base space S1 takes place in line 35, by repeating again the main problem underlying the whole narrative: no hay forma, no hay forma de que salga (There is no way, no way he’ll go out ).

Next, the conditional form for requests brings back space 3 (S3) in lines 36 and 37 and then listeners are guided back to S5 in lines 38 to 43. This time the transition marker is si (if), the Spanish prototypical connector, but as in line 18 with porque, this is not its main meaning or function. Si is used by the narrator to reopen the input space where the mother justifies all her son’s behaviours, except him not going out. This space is then abruptly interrupted to take the listener back to S1 and the space builder this time is o sea, pero de salir y eso,nada (I mean, but about going out and so, nothing). Base space S1 is resumed in lines 49 to 54, after the narrator’s answer to a direct question from the radio presenter in lines 45 to 48, again by repeating the main storyline or anchor: No, no, no hay forma..(No, no, there is just no way…).  S1 is interrupted for the last time with input space S5 –he always takes the garbage out and water the plants- a sub-story or second specification of S5, which is introduced with no pragmatic or linguistic markers. Finally the narrative is closed with a last return to S1: y… no piensa en otra cosa (and he won’t think about anything else).

In short, these pragmatic markers act as space builders and have both, structuring and modal functions. They are helping the listener to segment and assimilate the broken information while creating an empathic bond with the listener, as well as helping the narrator bring out the painful experience by means of stallers, repairs, redoings, etc., It is worth emphasising that most of these discourse markers have the main function of opening or building new narrative spaces and that these functions or meanings do not always coincide with their more traditional or prototypical meanings, see porque (line 18) and si (line 38) as good examples of this polyfunctionality7. Finally it must also be noted that a more fine-grained analysis is always possible, that is, more breakings of the stories, and therefore other mental spaces, can be considered. For instance when direct speech is introduced in S4 in lines 27 to 34, or in S2 after the marker y resulta de que mmm (and  it turned out that mmm…) in line 4, which opens a new space –a definite date in the time when the son and his girlfriend were preparing their wedding.

Narrative anchors

In addition to discourse markers, the narrator also provides the hearer with some narrative anchors. As stated above, these can be repetitions of expressions or a re-elaboration of some ideas which will guide the cross-mappings between spaces in order to eventually make a coherent blend. Thus, for instance, the narrator repeats at several points of her story that her son “does not want to go out” (no quiere salir) (lines 10, 24), that he is always at home (está siempre metidoen casa) (line 9), “there is no way to make him go out” (es que no hay forma de sacarle de casa) (line 8, 35), “no way to make him go somewhere else” (no hay forma de hacerle salir a ningún sitio) (line 19), “about going out, just not at all” (de salir y eso, nada) (line 44), etc. Particularly, the phrase “there is no way” (no hay forma) can be found seven times (lines 8, 13, 19, 26, 35, 49). All these expressions with the same meaning serve for the purpose of connecting, not only the several fragments of one space, as in this case of the fragments of S1, but also to connect the different input spaces, that is, to guide the cross-mappings that make the final blend possible. Thus, in S3 we find “if he could go out” (a ver sipuede salir) (line 15) and “what can I do to make him go out” (qué puedo hacer para que salga), and in S4 “but… if you don’t go out…” (peroes que si no sales de casa…) (line 32).

As we have just seen, it is the hearer who connects the different input narrative spaces as the process unfolds, following the clues provided by the text, such as scripts, pragmatic markers and narrative anchors, until the story reaches a satisfying degree of coherence by means of emergent cross-mappings or projections. However, in order to do so, the hearer is not only meant to build and interconnect the different mental spaces that constitute the narrative, but also to keep track of the way in which the speaker goes from one space to another, and to focus his attention on a different space at a time as the narrative evolves. In the next section, we will describe some of the strategies to manage attention through the mental spaces that compose the narrative.

Attention management in oral narratives

Attention is one of our main cognitive abilities and it plays a crucial role in the structuring and interpretation of the world. It allows us to focus on small parts of our environment in order to understand it all. This way of making sense of the world is reflected in language (Langacker 1987, 2001; Talmy 2000, 2007, 2008). According to Talmy, the attentionalsystem of languagefunctions in a similar way to attention in other cognitive, perceptual systems, such as vision or hearing. So, linguistic forms can direct our attention over a referent scene in a process that he calls windowing of attention, by which “portions of a referent scene will be placed in the foreground of attention while the remainder of the scene is backgrounded” (Talmy, 2000, p. 258).

In the first place, in order to better explain how attention works in discourse processing, some relevant concepts related with our ability to focus attention will be presented as they are studied in psychology. When processing a visual scene, for example, attention is commonly explained as a spotlight which can be directed and so highlight different parts of a given scene at a time, so that each part can be subjected to analysis before they all are put together by the viewer in his/her mind (Posner, 1980; Sperling and Weichselgartner, 1995). Alternatively, in hearing, the cocktail party effect has been studied as the capacity of focusing attention on one conversation among many others happening at the same time in the same place. In a similar way, when processing discourse, we also direct our attention to small parts of it, what Langacker calls attentional frames (Langacker, 2001). Thus, the active frame at a given point in discourse, the Current Discourse Space, can be conceived as a mental space constantly being modified and updated by successive frames.

Attention is guided by a combination of top-down and bottom-up factors (Desimone and Duncan, 1995; Navalpakkam and Itti, 2005). Bottom-up factors are those features of the stimulus that make it perceptually salient, for example, in the case of visual stimuli, a bright colour or a  big size would be bottom-up factors that would make a particular stimulus more salient. In those cases, attention is stimulus-driven. But attention can also be strongly modulated by top-down, user-driven factors, which include features such as volition, expectations, beliefs, task demands, etc… For instance, big size or bright colours can be overridden if we have been specifically asked to search for something small or black. When processing discourse, such top-down factors play an important role on the side of the listener, whose personal aims, expectations or previous experiences will influence his interpretation by focusing more or less attention on a part of the whole discourse. Finally, working memory is another component that influences the management of attention. It refers to our capacity of storing information temporarily for a particular task. When focusing attention, bottom-up, sitmulus-driven factors, are memory-free, whereas top-down factors are memory dependent (Henderson and Hollingworth 1999, 2003). In discourse, and especially in oral discourse, the temporary quality of this capacity accounts for the need of repeating those parts of the discourse that are considered more significant and have to be kept in focus.

A very important feature of attention in general is that it is gradable. As pointed out by Talmy, “the particular level of attention on a linguistic entity is set in terms of foregrounding or backgrounding relative to a baseline for the entity, rather than absolutely on a zero-based scale”(Talmy, 2007, p. 266). Therefore, language has an extensive system that assigns different degrees of salience to the parts of an expression and those phenomena that exert increased attention on a linguistic entity are regularly accompanied by additional cognitive effects, such asdistinctiveness, clarity and significance and, what is more, several factorsregularly combine and interact in order to produce attentional effects (Talmy, 2008).

Consequently, all the narrative spaces that compose the narrative in our study cannot be considered as equally salient. Figure 3 is a representation of the different degrees of salience, corresponding to differences in size in the picture, of the input spaces in SpN3.

As explained above, both bottom-up and top-down factors will combine in order to provide different salience to each space. Among the stimulus-driven factors, size can be considered in the first place. Both in terms of time and total number of words (116 words), S1, the base space, is the biggest one. Besides, S1 is made up of eight fragments spread all through the whole narrative and therefore is repeatedly activated. The repetition of the very same expression “there is no way” (no hay forma) keeps the space activated in our working memory, and therefore in focus, for a longer time. On the contrary, S2, the background space that informs of the reason why the narrator’s son does not want to go out, is far shorter (41 words) and it is only activated once at the beginning of the story, which causes it to fade away from working memory as new spaces are being opened or kept activated, and by the end of the narrative it has definitely been backgrounded, i.e. it is not active any longer. This is the reason why we have placed it in the last position in the graphics.

Among the top-down, user-driven factors, we must consider the relevance of S3, the purpose space. This space establishes the reason why the woman called the radio and is telling her story: she wants to know what else she could do to help her son and make him go out. Consequently, S3 attracts a high level of attention, even if it only amounts to a total of 46 words, because it provides the final aim of the narrative: asking for advice. For this reason, we have positioned S3 in a second place, only after S1. Moreover, as discussed in the previous section, the narrator makes use of markers, often by juxtaposing more than one, in order to attract the hearer’s attention to those parts that she considers more significant. So, for instance, S4 is introduced in line 27 by a marked change of subject. It shifts from the third person –he, the son– to the first one, I (yo), and S5 is first introduced in line 20 with the simplest marker, and (y). We can conclude then that S4 and S5 are not highly salient and not much attention is driven towards them.

Finally, it is important to point out that, since attention is gradable, as stated above, the multiple spaces built for the narrative are not just in or out of focus, but rather foregroundedor backgrounded to a certain degree at any given point of discourse, which implies that more than one space can be active at the same time. In our narrative, S1 is foregrounded all the time, whereas, as explained above, S2 is only activated once and then gradually backgrounded until it fades away completely before the end of the narrative.

In short, S1 – the fact that her son does not want to go out –  is the most important part of the narrative for the woman who calls the radio program and a great deal of attention is driven towards this space through different strategies: frequency (eight fragments), repetitions, markers, etc. The purpose for calling is asking for advice, S3, a narrative space which is also highly relevant for the global meaning of the narrative. Other narrative spaces contribute to the whole by expanding the information on the relationship between the son and his parents (S5) and how the woman tries to convince her son to go out (S4), but they are not particularly significant. Finally, the reason why the son does not want to go out, i.e. the fact that his girlfriend died in an accident, is only additional information for the listeners. The narrator does not regard this information is worth much attention and for this reason she does not devote more than a few words to it and does not reopen the mental space again in the rest of her discourse. Clearly, in the construction of the blend, not all spaces will have the same “weight” in the final emergent story.

Conclusions

From the analysis presented in this work, we must conclude that Mental Spaces and Conceptual Integration Theory is a very useful approach to understand online meaning construction phenomena in narrative discourse, especially in oral narratives of charged events. From this approach, oral narratives can be understood as complexanddynamic systems in which a series of incomplete and fragmented stories compete for attention. We have seen that not all narrative input spaces are equally salient, but can attract different degrees of attention. In that sense, we have analyzed how space builders are also attentional markers, which attract the hearer’s attention to a specific space at any point in discourse and help the hearer to select the figure (foregrounded information) against the landmark (mass of background information presented by the narrator). The global meaning of the narrative is an emergent story that arises through a gradually increasing network of cross-mappingsand blends from multiple narrative mental spaces that have a different weight in the final product, that are often embedded in other spaces and that are interconnected in different ways. In conclusion, the online construction of meaning in an oral narrative of charged events implies the same cognitive processes for meaning construction used elsewhere in language: maintaining the activation of a space already set up, elaborating many spaces at the same time, blending the spaces into a higher level structure, adding cultural knowledge, etc.