CONTINUATIVE MEANING AND COMMUNICATIVE DISTANCE. NOTES ON THE USE OF THE SPANISH GERUNDIAL PERIPHRASES WITH SEGUIR / CONTINUAR / PROSEGUIR

This paper examines the differences in distribution between three Spanish verbal periphrases with continuative meaning: gerundial constructions with seguir , continuar and proseguir . Based on quantitative data extracted from two Spanish language corpora, our diachronic and synchronic analysis shows that over time the three auxiliaries have experienced changes in frequency which can be related to text genre and to the concept of communicative immediacy and distance. We argue that in Modern Spanish the whole discourse continuum of communicative immediacy and distance is covered by continuar and seguir . Contrastively, proseguir mainly survives in the prefab proseguir diciendo , which also seems to have played a significant role in the original expansion of the periphrasis. Thus, our data suggest that prefabs are not only involved in increasing the use of a grammaticalizing construction, but that they arguably also play an important role in the preservation of a construction whose use has decreased significantly over time.


INTRODUCTION 1
The Spanish language is known for its frequent use of verbal periphrases, pluriverbal expressions composed of a finite verb form that serves as an auxiliary, and an auxiliated non-finite verb form, i.e. an infinitive, gerund or past participle (see, for example, Bravo & García Fernández 2016;Fernández de Castro 1999;García Fernández 2006;Gómez Torrego 1999;Olbertz 1998Olbertz , 2023;;RAE-ASALE 2010: 2105-2222;Squartini 1998;Yllera 1999).Examples (1) to (3) are instances of periphrases with an infinitive, a gerund and a past participle respectively. (1) Isabel, contestando a Andrea, que acaba de quejarse de la sosería del matrimonio, le rebate, escandalizada, tal rebeldía.(Usos amorosos del dieciocho en España, Carmen Martín Gaite, España, 1972, CORDE) 'Responding to Andrea, who has just complained about the dullness of the marriage, Isabel rebukes such rebellion, scandalized.'(2) Parece que la policía la estuvo interrogando, fue por lo de los estudiantes.(Últimas tardes con Teresa, Juan Marsé, España, 1966, CORDE) 'It seems that the police were questioning her, it was because of the students.'(3) Ahora anda metido en un estudio sobre la justicia o la injusticia del latifundio.(La mortaja, Miguel Delibes, España, 1948-1963, CORDE) 'He is now involved in a study on the justice or injustice of landownership.' In these syntactic constructions the finite and the non-finite verb constitute a single unit with a unitary meaning that conveys the aspectual and modality features of the event or state denoted by the non-finite verb.They are the outcome of the grammaticalization of a formerly lexical verb in conjunction with a nonfinite verb form.
There seems to be a fair amount of consensus among authors with regard to the semantic and syntactic status of seguir + GER and continuar + GER.Few studies, however, mention proseguir, although the reason for this lack of attention is unclear.It might be that they do not consider proseguir + GER a periphrasis or that they just overlook the construction because of its relatively low frequency.In this sense, García Fernández (2006: 237) constitutes an exception, since in this study the periphrastic character of proseguir is explicitly denied on the basis of a number of syntactical tests involving its use with meteorological verbs, in passive constructions and in contexts of focalization . Fernández Martín (2019: 68) Spaulding (1925Spaulding ( -1928: 262-265) : 262-265) and Yllera (1980Yllera ( : 83, 1999: 3421-3422): 3421-3422), on the other hand, do classify proseguir + GER as periphrastic, although they recognize that its use is very limited and literary in nature.
From a historical linguistic point of view, it seems logical to incorporate proseguir + GER in the same group as seguir and continuar, because, as stated by Keniston, proseguir is the only auxiliary that is used in conjunction with a gerund in the sixteenth century to express the continuation of the action (Keniston 1937: 469).Thus, Keniston's observation suggests that in earlier times proseguir used to have a similar meaning and function to modern-day seguir and continuar.
The observed differences in frequency and stylistic properties between the three periphrases in Modern Spanish raise the following questions: − Does the current distribution reflect an earlier pattern or has it changed over time?− How can the difference(s) in frequency be accounted for?We think that these research questions are pertinent since, as far as we are aware, there are no studies that treat the diachronic development of the three auxiliaries together and focus on their similarities and differences.With the present paper, therefore, we hope to contribute to a better understanding of the expression of continuative meaning in Modern Spanish.
In the following sections we will address these questions starting with a brief discussion of the origins of the three auxiliaries (section 2).Next, we will outline the main characteristics of the corpora that were used to collect the data on which our analysis is based (section 3).Subsequently, in section 4 we will discuss the existence of a discourse continuum and the concept of communicative immediacy and distance as elaborated by Koch and Oesterreicher (2007: 20-42).In addition, we will propose a classification of different text genres based on this concept of immediacy and distance.In section 5 a diachronic overview is given of the distribution of the three periphrases, while in section 6 the distribution of the constructions is examined across different text genres with the aim of finding significant patterns of use that might be explained on the basis of communicative immediacy and distance.In section 7 we will focus on another parameter relevant to the concept of immediacy and distance, the distribution of the periphrases in both physical and online written versus oral communication.Section 8 deals with the status of proseguir + GER in Modern Spanish.The paper closes in section 9 with a discussion of the results and some concluding remarks.

THE ORIGIN OF THE THREE PERIPHRASES
The auxiliaries seguir, continuar and proseguir find their origin in Latin, corresponding respectively to SEQUI ('to follow', 'to come after'), CONTINUARE ('to join', 'to make continuous') and PROSEQUI ('to follow', 'to accompany').Note that the last verb is composed of the preposition PRO ('before') and the aforementioned Latin verb SEQUI.As an independent verb seguir is found in 'Ignacio goes up to his room, the dog follows him wanting to play.'

CORPUS
Our study of the diachronic distribution of the three gerund constructions is based on quantitative data taken from the Spanish language corpus CORDE.The corpus was created by the Royal Academy of the Spanish Language (Real Academia Española).It contains some 250 million words and covers the period from around 800 to 1975 (see RAE banco de datos, CORDE).Although it includes texts from different Spanish-speaking countries, for the present purpose we focused only on texts written by Spanish peninsular authors, because it is known from other studies that verbal periphrases are susceptible to diatopic variation (see, for example, Nieuwenhuijsen 2019, Sedano 2000).Furthermore, the history of writing in the Latin-American countries only started in the sixteenth century, and given that we want to examine the diachronic evolution of the constructions, Peninsular Spanish texts provide far greater diachronic scope.
The CORDE corpus also allows us to distinguish between different text genres.The corpus is divided into fiction texts that account for 44 % of the total amount of texts and non-fiction texts that make up the remaining 56 % of the database.The fiction texts distinguish between prose and poetry and the subcategories lyric, narrative and drama.The non-fiction category includes didactic, religious, historical, society, scientific and legal prose (for details see RAE, the manual of the corpus).
We collected a dataset of n = 9 233 examples, 5 736 cases of seguir + GER, 2 702 cases of continuar + GER and 795 cases of proseguir + GER.Thus, seguir represents 62 % of the total number of cases, continuar 29 % and proseguir 9 %. 3ince the CORDE corpus is not lemmatized, the data were extracted by introducing different forms of each verb followed by a gerund using wildcards to collect all the relevant occurrences.Afterwards, every case was analysed manually in order to exclude tokens that did not correspond to one of the three periphrases.
In addition, we collected a dataset of cases of continuar/proseguir/seguir + GER extracted from CORPES XXI, a corpus also created by the Real Academia Española.The corpus is still growing but at the date of access it comprised some 350 million words and contained language utterances produced in the period 2001-2021 (version 0.94, see RAE banco de datos, CORPES XXI).The lemmatization of this corpus offers the possibility of conducting more specific searches than in the CORDE corpus.We searched for strings consisting of the three verb headwords followed by a gerund.Moreover, one of the assets of this present-day Spanish language corpus is that it allows us to distinguish between written and oral communication, and thus provides an additional parameter with which to analyze the distribution of the three auxiliaries in Modern Spanish.Nonpeninsular data were filtered out to keep the dataset comparable with the diachronic dataset.These selection criteria rendered a total of n = 55.536examples, 48.992 cases of seguir + GER, 6 432 cases of continuar + GER and 112 cases of proseguir + GER.This dataset contains a larger representation of seguir + GER (88.2 %) compared to the CORDE corpus, and lower percentages for both continuar + GER (11.6 %) and proseguir + GER (0.2 %).

COMMUNICATIVE IMMEDIACY AND DISTANCE
The communicative conditions and verbalization strategies that are involved in the production of a text were visualized by Koch and Oesterreicher (2007: 20-42; also see Koch 1999;Oesterreicher 1996Oesterreicher , 1997Oesterreicher , 2004) in a model of communicative immediacy and distance (see Figure 1). 4IGURE 1: Model of communicative immediacy and distance (Koch 1999: 400).
The model takes into account two dimensions, the medium (graphic or phonic) and the conception of linguistic discourse, i.e., the communicative strategies that are employed.These can be divided in two general modes (written and spoken).Whereas the medium entails a dichotomy whose members exclude each other mutually (a graphic code cannot be phonic at the same time), the written and spoken conceptions give rise to a continuum that ranges from the purely spoken to the purely written conception of linguistic discourse.In this way, the two extreme poles of the continuum correspond to forms of communication or linguistic utterances that imply a maximum degree of communicative distance (written mode) and of communicative immediacy (spoken mode) respectively.Interestingly, in spite of the previously mentioned dichotomy in the medium of realization, transpositions from one medium to the other are possible, e.g., an interview that is written down or a letter read out aloud.
The communicative discourse scale principally refers to linguistic utterances that respond to particular social and professional needs, such as telephone conversations, private letters, sermons, academic conferences or legal contracts.(Oesterreicher 1996(Oesterreicher , 1997;;Koch & Oesterreicher 2007: 34-35, 2012: 444).The CORDE corpus, on which the diachronic analysis for this paper is based, focusses on the larger category of text genres.We might generalize and argue that, since academic conferences and legal contracts are assigned to the far end of the discourse continuum, where communicative distance is maximally present, the corresponding text genres of scientific prose and legal prose are also expected to reflect a high degree of communicative distance and can be characterized as formal.However, since both scientific prose and legal prose incorporate different subtypes of texts, such a generalization probably lacks justification.
In order to concretize the concept of communicative immediacy, Koch (1999: 400) identifies a number of parameters, related to the participants involved, the temporal and spatial setting in which the discourse takes place and the topic.According to these parameters, communicative immediacy is grounded in the deictic immediacy of time ("now") and place ("here") and in the presence of the participants that directly participate in the discourse, which takes place in a familiar, informal setting ("you" familiar or intimate, and "I").Hence, if we take the presence of concrete words that refer to these parameters of temporal, spatial and participant deixis as a measure of the degree of immediacy of the text genres in which they appear, we will be able to make more specific statements about the relative position of these genres on the discourse scale of communicative immediacy and distance.
Accordingly, we calculated the presence of the Spanish words that correspond to the concepts of temporal (ahora, 'now') and spatial (aquí, 'here') immediacy, as well as the Spanish forms of address that refer to the firs-person singular (yo, 'I') and to the second person singular familiar (tú, 'you' familiar) for each text genre of the CORDE corpus.However, since Spanish is a pro-drop language, in which the overt expression of subject pronouns is constrained by specific contextual factors (RAE-ASALE 2010: 2547-2559), which are not necessarily the same for all text genres, we selected the corresponding object pronouns me ('me') and te ('you' familiar), since their use is not affected by the type of contextual constraints imposed on subject pronouns.We focused on the period of the eighteenth to the twentieth century, because, as we will see in the next section, there are hardly any instances of the three periphrases before the eighteenth century and we are particularly interested in the relationship between the usage of the three constructions and text genre.In addition, we limited the search to peninsular texts, for the reasons we explained in section 3. Since the number of texts within each text genre and the number of words per text varies, Figure 2 expresses the relative frequency of the mentioned deictic adverbs and pronouns under scrutiny per ten thousand words. 5For convenience of presentation, we have ordered the genres by their overall scores in the twentieth century, from the highest score on the left side to the lowest score on the right side of the figure.Supposing a high score indicates a high degree of communicative immediacy, whereas a low score indicates a low degree of communicative immediacy, i.e., a high degree of distance, the horizontal axis of Figure 2 reflects the discourse continuum of immediacy and distance, where the left end corresponds to maximum immediacy and the right end to maximum distance.
FIGURE 2: Relative frequency of words of immediacy (aquí-ahora-me-te) per ten thousand words per text genre and per century in peninsular documents (data extracted from CORDE). 6f we assume that the references to deictic immediacy and familiar participants directly involved in the discourse are indicators of the degree of communicative immediacy of the corresponding text genre, the results of Figure 2 leave no doubt as to the clear differences that exist between the analysed genres.At the left end of the scale, we find dramatic prose and verse, which have very high frequencies across all three centuries, thus reflecting their communicative immediacy.On the other end of the scale scientific and legal prose stand out with the lowest scores in the nineteenth and twentieth centuries, endorsing the idea that these are genres in which communicative distance is inherent.Moreover, within the fictional genres, the highest scores after dramatic prose and verse are found in narrative prose, while in the group of non-fictional genres historical and religious prose present the highest scores.
In the following sections, where we discuss the distribution of the three auxiliaries that are involved in the expression of continuity, the ordering of the text genres according to the presence of words of immediacy will serve to interpret possible differences in distribution across the text genres.

A DIACHRONIC OVERVIEW
In this section we will briefly discuss the distribution of the three auxiliaries in conjunction with the gerund across the centuries.We calculated the relative frequencies for all three constructions for each century in which we found at least one example of either seguir, continuar or proseguir + GER. Figure 3 displays the relevant data.The results reveal that the earliest cases of continuar/proseguir/seguir + GER date from the fifteenth century. 7Our data also make clear that, whereas proseguir is the most frequent verb in this type of construction before the nineteenth century, continuar begins to emerge as a competitor of proseguir in the eighteenth century.Another radical change is also observed from the nineteenth century onwards; the use of seguir rises dramatically and replaces continuar and proseguir as the dominant auxiliary.Although the results suggest that continuar also starts to expand its use in the eighteenth century, it is seguir that takes the lead in the twentieth century, occupying almost three quarters of the functional niche of the continuative meaning.

TEXT GENRES
As far as we are aware, there are no in-depth studies yet that treat the three periphrases together, and authors that do mention the constructions only make general statements about the stylistic properties of the auxiliaries (see section 1).Therefore, we will take a closer look at the text genres in which the periphrases appear in order to clarify whether the distribution of the constructions and their stylistic properties can be linked to their use in particular text genres.Figure 4 offers a panchronic overview of the three periphrases as they appear in the different text genres distinguished in CORDE.
Seguir is the most frequent verb in all text genres, except for legal prose, which favours continuar (10 occurrences per million words), and narrative verse, in which proseguir and seguir show the same frequency.In narrative verse, however, the overall presence of the three constructions is limited, with a total of 31 occurrences.
On the one hand, it is not surprising that seguir displays the highest frequency in almost all text genres, given its high overall frequency in the nineteenth and twentieth centuries.On the other hand, the fact that continuar is 7 Prior to the fifteenth century our corpus does testify the existence of combinations of seguir and a gerund, but in these cases the construction has not yet acquired its continuative meaning.Instead, seguir still maintains its lexical meaning of 'to follow' and frequently has its own complement, while the gerund expresses the way in which the action is carried out, as the following example illustrates: (i) & lo pusieron a cauallo & lo siguian aquexando el cauallo con correas que andas mas ayna.
(Juan Fernández de Heredia, Gran crónica de España, I, 1385, CORDE) 'And they placed him on a horse and they followed him spurring the horse with belts so that it would run faster.'This type of usage is also found in Modern Spanish: (ii) Yo las seguí arrastrando el echarpe por el suelo.(Elena Quiroga, Escribo tu nombre, 1965, CORDE) 'I followed them dragging the scarf across the floor.'Obviously, since they are not instances of the periphrasis, they were filtered out when the sample set was checked manually.ANU.FILOL.ESTUD.LINGÜÍST., 13/2023, pp.55-77.ISSN: 2014-1408.DOI: 10.1344/AFEL2023.13.3 much more frequent than seguir in legal prose might explain why previous studies claim that continuar is more formal and seguir is favoured in colloquial speech.According to the results in Figure 2, legal prose falls at the end of the continuum where communicative distance is greatest, and it is characterized as a very formal text genre with a highly elaborated architecture (Aguirrezabala & Fanduzzi 2012;López Izquierdo 2008: 428;Oesterreicher 1996Oesterreicher , 2004)).Even so, in view of our results, it seems more accurate to say that seguir is the unmarked auxiliary, which, in principle, makes it suitable for every discourse, including colloquial speech.In line with this, one might expect that the more formal continuar would also be preferred in scientific prose, which is also formal in nature.As we will see below, this used to be the case until the twentieth century, when seguir consolidated its status as the most frequent and unmarked auxiliary.
Taking into account that proseguir outnumbered both continuar and seguir until the nineteenth century when the use of seguir started to grow substantially at the cost of proseguir, and continuar initially kept pace with the expansion of seguir, the data in Figure 4 may be too rough and obscure differences between the use of the three verbs in the various text genres per century.Therefore, we will examine the parameter of text genre separately in each century, starting with the eighteenth century, because before that time the absolute numbers of seguir, and partially also those of continuar, are too low to make a feasible and reliable subdivision of text genres.We already saw in Figure 3 that proseguir is overall the most common verb to express continuity of the action indicated by the gerund in the eighteenth century.The data in Figure 5 demonstrate that this auxiliary has the highest frequency in narrative, historical and didactic prose, three genres allocated at different points of the discourse continuum of immediacy and distance (see Figure 2).The distribution of the three auxiliaries in scientific prose stands out, bearing in mind that it is considered a highly elaborated, learned text genre (Oesterreicher 1996(Oesterreicher , 2004;;Sevilla Muñoz & Sevilla Muñoz 2003) and that in the eighteenth century it already involves relatively low communicative immediacy (see Figure 2).Interestingly, the fact that it is continuar that predominates in scientific prose in this century suggests that proseguir might have been less formal in character in the eighteenth century than in Modern Spanish, or at least less formal than continuar.Moreover, while seguir still has a very low overall frequency in the eighteenth century (see Figure 3), it is rather frequent in historical prose, although the differences in frequency between the three verbs are less marked in this genre than in the other genres.
We may infer from the data visualized in Figure 6 that proseguir definitely loses ground in the nineteenth century, because it is no longer dominant in any of the text genres.On the other hand, continuar clearly gains in importance in this century.It is not only the most frequent auxiliary in scientific prose, as in the previous century, but also in, religious, historical, journalistic and legal prose.As already mentioned, legal prose belongs to the right end of the discourse continuum where communicative distance is high, while the other genres, according to Figure 2, are less distant.In turn, seguir also becomes quite competitive in this century.It is predominant in dramatic prose and verse and in narrative, didactic and society prose.Noteworthy, some of these genres received high scores on the parameter of immediacy in Figure 2.  The results of Figure 7 leave no doubt as to the prevalence of seguir in the twentieth century.The verb is dominant in all text genres, except for legal prose.As we already saw in Figure 3, the use of proseguir is extremely limited in the twentieth century.It appears in narrative, lyric, religious, historical, didactic, society and scientific prose but generally with fewer than 10 instances in each genre type, except for narrative prose with a total of 41 occurrences and scientific prose with 14 occurrences overall.Note that in lyric prose proseguir reaches a frequency of 29 occurrences per million words, but this is based on only 4 actual cases.Continuar, in turn, has lost its relevance in almost all text genres in which it prevailed in the previous century and only predominates in legal prose (43 occurrences per million words).
From the data collected from CORDE we may tentatively conclude that in the eighteenth century proseguir is the most common verb to express the meaning of continuity in conjunction with a gerund and dominates in a number of text genres (narrative, historical and didactic prose) situated at different points of the discourse scale.In this century, seguir starts to appear in historical prose, while continuar is introduced in the highly distant genre of scientific prose.The nineteenth century is clearly a period of competition, in which continuar and seguir extend their use through a number of different text genres that involve different degrees of communicative immediacy and distance.Remarkably, two of the genres (legal and scientific prose), in which continuar is the preferred verb, are clearly marked for communicative distance.As a consequence, the use of proseguir definitely drops off in the nineteenth century.On the other hand, twentieth century Spanish attests the overall success of seguir.This occurs, not only at the cost of proseguir, a development that had already started in the nineteenth century, but also at the cost of continuar, which is now only dominant in the communicatively distant genre of legal prose.These results lead us to conclude that in the twentieth century seguir is the unmarked auxiliary preferred in all text genres, independently of the degree of communicative immediacy or distance they imply, except for legal prose, which is the domain of continuar.
In the next section we will further explore the relationship between the distribution of the three auxiliaries and the concept of communicative immediacy and distance.

LANGUAGE
Until now we have concentrated on linguistic utterances transmitted graphically and pertaining to the written mode.Yet, if we want to further test the idea that in contemporary language continuar is used in discourse that maintains a high degree of communicative distance, whereas seguir, as the unmarked form, in principle covers the whole discourse scale, it would be interesting to examine the possible influence of the channel by which a message is communicated on the choice of the auxiliary in the gerund construction with continuative meaning.Therefore, we will use the parameter of mode, i.e., the distinction between spoken and written language, given that spoken language is strongly related to communicative immediacy, while in written communication a certain degree of communicative distance is inherent (Oesterreicher 1997, Koch & Oesterreicher 2012).For this purpose, we compiled a second dataset extracted from the corpus CORPES XXI, since CORDE only contains written texts.Apart from this, an additional advantage of CORPES XXI is that it covers the twenty-first century, whereas CORDE only contains texts prior to 1976.The spoken section of CORPES XXI includes a variety of recordings and transcriptions of debates, speeches, interviews and press conferences, among others.Within the broad category of written communication, it is possible to make a subdivision based on whether the documents are traditional written texts on paper or web documents.Generally speaking, web documents are different from other written documents insofar as they are much more unpredictable and individualized than paper documents (Santini 2006: 67).It should be noted, however, that they constitute a relatively heterogeneous category, depending on whether they are based on paper models or have been created exclusively for the digital medium.As Kanaris and Stamatatos state: Webpages form new kinds of genre, taking advantage of a new communication medium and different type of interaction with the receiver.Some of the webpage genres are variants of traditional genres (e.g., online newspapers) while others have no antecedents in paper documents (e.g., personal home pages, blogs).(Kanaris & Stamatatos 2009: 499) CORPES XXI distinguishes between different kinds of web documents, and blogs are the most common type of web documents in the corpus.Therefore, for the present discussion we will compare this particular kind of web documents with paper documents and spoken utterances.Paper documents and blogs together with the spoken mode can be seen as different manifestations of the discourse continuum, where paper documents are assumed to possess the lowest degree of immediacy, spoken utterances have the highest degree of immediacy, and blogs hold a position in between.
Given this characterization, we anticipate finding differences in the distribution of the three periphrases.We expect continuar, on the one hand, to have a higher frequency in paper documents than in spoken language and, on the other hand, a lower frequency in blogs than in paper documents, but its frequency in blogs will be higher than in spoken language.Seguir, in turn, is likely to exhibit the opposite pattern, with increasing frequencies from paper documents to blogs and spoken language, although we expect it to show high frequencies in all three types of utterances because of its supposedly unmarked status.As far as proseguir is concerned, in view of its limited distribution in the twentieth century CORDE data, we do not expect to find clear differences in frequency between the three categories.The results are given in Figure 8.
The results in Figure 8 confirm our expectation that continuar + GER is used far more frequently in paper documents (47 times per million words) than in spoken language (28).In addition, its frequency in blogs is higher (38) than in spoken utterances but lower in comparison to its frequency in paper documents.Seguir is very common in all three types of utterances, but as expected, its frequency increases as communicative immediacy increases (345 occurrences per million words in paper documents, 399 in blogs and 572 in spoken language).Not surprisingly, proseguir is almost non-existent in paper documents, blogs and spoken language (0.8, 0.2 and 0.6 times per million words respectively).
The results in Figure 8, thus, support the claim that continuar is likely to be used in communicative situations of distance rather than of immediacy, whereas seguir confirms its status of unmarked verb, although it has a particularly high frequency in spoken language.In a broader sense, the results also confirm the idea that communicative immediacy and distance play a role in the choice of one of the three verbs in periphrases with a continuative meaning.

PROSEGUIR + GER IN MODERN SPANISH
As Figure 3 made clear, proseguir was the most common of the three auxiliaries until the nineteenth century.Yet in principle, our description of the discourse continuum leaves no space for the auxiliary proseguir, since continuar is associated specifically with communicative distance, while seguir is the unmarked auxiliary that covers the whole discourse spectrum, but is the non-preferred option in contexts of high communicative distance.In Modern Spanish, rather than being associated with a particular text genre, proseguir has become mainly restricted to the context in which it has always been particularly common, i.e., in conjunction with the verb decir, rendering proseguir diciendo.In the CORDE corpus, proseguir diciendo covers 38 % (302/795) of all the cases with proseguir, whereas for continuar diciendo the overall percentage is 3.5 % (94/2,702) and for seguir diciendo 2 % (115/5,736).In Figure 9 we have combined the data from CORDE (fifteenth to twentieth century) on the three periphrases with the gerund diciendo with the corresponding data from CORPES XXI (twenty-first century).The percentages express the relative frequency of diciendo compared to other auxiliated verbs    A comparison of the three figures shows that, whereas proseguir diciendo constitutes a substantial part of all the instances of proseguir + GER in almost every century, for continuar this is only the case in the fifteenth and sixteenth century.The results in these two centuries, however, are based on two and three occurrences of diciendo respectively, against four and fourteen instances of continuar + GER overall.As far as seguir is concerned, this auxiliary hardly selects the gerund of decir compared to other verbs, not even in contemporary Spanish, when seguir is the preferred auxiliary to express the continuation of the action.Looking at the twenty-first century, there is still a strong connection between diciendo and the auxiliary proseguir, since this gerund covers 20 % of all the cases of proseguir + GER in this century.By contrast, continuar diciendo only accounts for 2 % of all the cases of the corresponding periphrasis, while seguir diciendo is only found in 0.8 % of the cases of seguir + GER in the twenty-first century.
Therefore, we might infer that proseguir diciendo has become a conventionalized multi-word combination, a prefab, which is defined by Erman and Warren as a combination of at least two words favoured by native speakers in preference to an alternative combination which could have been equivalent had there been no conventionalization.(Erman & Warren 2000: 31) Bybee and Torres Cacoullos ( 2009) stipulate an important role for these conventionalized multi-word strings in the process of grammaticalization.According to the authors, prefabs are the driving force behind the increased use of the general construction they relate to and serve as the locus of extension of the construction.9It goes beyond the scope of this paper to offer a thorough analysis of the grammaticalization of proseguir + GER but, considering the data, it is likely that proseguir diciendo played a similar role in the expansion of proseguir + GER from the fifteenth to the eighteenth century.10

DISCUSSION AND CONCLUDING REMARKS
Our first research question addressed the distribution of the three auxiliaries used in the gerundial periphrasis with continuative meaning, and specifically whether these constructions have experienced changes in frequency over time.Our quantitative analysis revealed that from the fifteenth century onwards, when the first instances of the periphrasis appeared in texts, it was mainly proseguir that served as an auxiliary for the construction.Moreover, the present-day pattern of use, according to which seguir + GER is far more frequent (96 occurrences per million words) than the two other verbs (continuar 31 occurrences per million words, proseguir 2), dates back to the nineteenth century when both continuar and seguir started to extend their use at the cost of proseguir.
Subsequently, we raised the question of how the differences in frequency could be accounted for.We argued that the current distribution of the three auxiliaries is linked to the concept of communicative immediacy and distance.When continuar + GER appeared as a periphrasis in the eighteenth century, it was prevalent in scientific prose, a text genre that is characterized by a high degree of communicative distance.Although there is evidence that the auxiliary moved away from the far end of the discourse continuum in the nineteenth century and extended its use into genres which involve less communicative distance, in Modern Spanish continuar is only dominant in the highly distant genre of legal prose, while seguir has managed to encroach into all other genres.This outcome endorses the characterization of continuar as a rather formal form, which is also confirmed by the fact that it is relatively more frequent in written documents than in spoken utterances and more frequent in texts on paper than in weblogs.Our analysis supports the claim, made by previous authors, that seguir is more colloquial in nature than continuar, since we found a higher frequency of seguir in spoken language than in weblogs and in written paper documents.However, the present analysis also provides evidence that seguir is not just a colloquial variant.It expanded its usage along the discourse continuum from a certain degree of communicative distance to communicative immediacy and ended up becoming the unmarked form to convey continuative meaning.
Proseguir started in the fifteenth century as the most common auxiliary in gerundial constructions with continuative meaning, which suggests that the verb must have been less formal or literary than it is nowadays.Over the centuries, as it came to be perceived increasingly as a formal and literary verb, it became less eligible for text genres with a relative high degree of immediacy, thus drifting away from more speech-like written utterances, which came to be the domain of seguir.In present-day Spanish, the overall use of the periphrasis proseguir + GER is very low.The prefab proseguir diciendo, which from the sixteenth century onwards has always been the most frequent combination compared to other gerunds that occur with this auxiliary, presumably played a crucial role in the expansion of the periphrasis until the eighteenth century.Interestingly, this conventionalized multi-word string seems to be playing an equally important role in Modern Spanish, since nowadays it is largely responsible for the preservation of proseguir + GER.Thus, our findings arguably indicate that across the lifespan of a construction a prefab might not only be pivotal at the start, when .ESTUD.LINGÜÍST., 13/2023, pp.55-77.ISSN: 2014-1408.DOI: 10.1344/AFEL2023.13.3

FIGURE 3 :
FIGURE 3: Relative frequency of continuar/proseguir/seguir + GER per million words per century in peninsular documents (data extracted from CORDE).

FIGURE 4 :
FIGURE 4: Relative frequency of continuar/proseguir/seguir + GER per million words per text genre in peninsular documents (data extracted from CORDE).

FIGURE 5 :
FIGURE 5: Relative frequency of continuar/proseguir/seguir + GER in the eighteenth century per million words per text genre in peninsular documents (data extracted from CORDE).

FIGURE 6 :
FIGURE 6: Relative frequency of continuar/proseguir/seguir + GER in the nineteenth century per million words per text genre in peninsular documents (data extracted from CORDE).

FIGURE 7 :
FIGURE 7: Relative frequency of continuar/proseguir/seguir + GER in the twentieth century per million words per text genre in peninsular documents (data extracted from CORDE).

FIGURE 8 :
FIGURE 8: Relative frequency of continuar/proseguir/seguir + GER per million words per mode (written/spoken) and type of written document in peninsular Spanish (data extracted from CORPES XXI).

FIGURE 9a :
FIGURE 9a: Relative frequency of continuar diciendo compared to other instances of continuar + GER in percentage per century in peninsular documents (data extracted from CORDE and CORPES XXI).

FIGURE 9b :
FIGURE 9b: Relative frequency of proseguir diciendo compared to other instances of proseguir + GER in percentage per century in peninsular documents (data extracted from CORDE CORPES XXI).

FIGURE 9c :
FIGURE 9c: Relative frequency of seguir diciendo compared to other instances of seguir + GER in percentage per century in peninsular documents (data extracted from CORDE and CORPES XXI).