Complex Words, Causatives, Verbal Periphrases and the Gerund: Romance Languages versus Czech (A Parallel Corpus-Based Study)

COMPLEX WORDS, CAUSATIVES, VERBAL PERIPHRASES AND THE GERUND The monograph focuses on the typological diff erences between the four most widely spoken Romance languages (Spanish, Portuguese, French and Italian) and Czech. Utilizing data from InterCorp, the parallel corpus project of the Czech National Corpus, the book analyses various categories (expression of potential non-volitional participation, iterativity, causation, beginning of an action and adverbial subordination) to discover diff erences and similarities between Czech and the Romance languages. Due to the massive amount of data mined, as well as the high number of languages examined, the monograph presents general and individual typological features of the four Romance languages and Czech that oſt en exceed what has previously been accepted in the fi eld of comparative linguistics. KAROLINUM COMPLEX WORDS, CAUSATIVES, VERBAL PERIPHRASES AND THE GERUND ROMANCE LANGUAGES VERSUS CZECH (A PARALLEL CORPUS-BASED STUDY) EDITED BY PETR ČERMÁK DANA KRATOCHVÍLOVÁ OLGA NÁDVORNÍKOVÁ PAVEL ŠTICHAUER complex words_mont.indd 1 23/04/2020 09:09 Complex Words, Causatives, Verbal Periphrases and the Gerund Romance Languages versus Czech (A Parallel Corpus-Based Study) Edited by: Petr Čermák Dana Kratochvílová Olga Nádvorníková Pavel Štichauer Reviewed by: Prof. Bohumil Zavadil Dr. Jana Pešková KAROLINUM PRESS Karolinum Press is a publishing department of Charles University Ovocný trh 560/5, 116 36, Prague 1, Czech Republic www.karolinum.cz This work was supported by the European Regional Development Fund-Project “Creativity and Adaptability as Conditions of the Success of Europe in an Interrelated World” (No. CZ.02.1.01/0.0/0.0/16_019 /0000734) and by the Charles University project Progres Q10, Language in the shiftings of time, space, and culture. © Karolinum Press, 2020 © Edited by Petr Čermák, Dana Kratochvílová, Olga Nádvorníková and Pavel Štichauer, 2020 © Text by Leontýna Bratánková, Petr Čermák, Štěpánka Černikovská, Jan Hricsina, Jiří Jančík, Jaroslava Jindrová, Dana Kratochvílová, Zuzana Krinková, Petra Laufková, Olga Nádvorníková, Daniel Petrík, Pavel Štichauer and Eliška Třísková, 2020 Designed by Jan Šerých Typeset by Karolinum Press First edition ISBN 978-80-246-4554-4 ISBN: 978-80-246-4616-9 (pdf) Univerzita Karlova Nakladatelství Karolinum 2020 www.karolinum.cz [email protected] Contents 1. expressions of potential participation, iterativity, causation, ingressivity and adverbial subordination in the light of parallel corpora Petr Čermák, Dana Kratochvílová, Olga Nádvorníková, Pavel Štichauer –––– 9 1.1 Investigation project and its history –––– 10 1.2 Objectives and scope of the present monograph –––– 11 1.3 Organisation of the monograph –––– 12 1.4 Terminological remarks –––– 13 1.4.1 Romance languages under scrutiny and use of the term Romance –––– 13 1.4.2 Use of the terms counterpart and respondent –––– 14 2. corpus design & corpus-based contrastive research methodology Olga Nádvorníková –––– 15 2.0 Introduction –––– 16 2.1 Corpus-based contrastive research methodology –––– 16 2.2 Corpora used in this study –––– 21 3. morphologically complex words in romance and their czech respondents Pavel Štichauer, Jan Hricsina, Jiří Jančík, Jaroslava Jindrová, Zuzana Krinková, Daniel Petrík –––– 25 3.0 Introduction –––– 26 3.1 Word-formation: complex vs simple words –––– 27 3.2 Romance and Czech: common and different word-formation patterns –––– 27 3.3 The typology of Czech respondents –––– 29 3.3.1 Typology of Czech respondents of the adjectives with the suffix -bile/-ble/-vel –––– 29 3.3.2 Typology of Czech respondents for verbs with the prefix re-/ri- –––– 31 3.4 The modal suffix -ble/-bile/-vel –––– 32 3.4.1 Data elaboration and analysis –––– 33 3.4.2 Quantitative distribution of the types –––– 35 3.4.3 Discussion of various examples –––– 36 3.5 The iterative prefix re-/ri- –––– 39 3.5.1 Data elaboration and analysis –––– 39 3.5.2 Quantitative distribution of the types –––– 40 3.5.3 Discussion of various examples –––– 41 3.6 Concluding remarks –––– 43 4. causative constructions in romance and their czech respondents Petr Čermák, Dana Kratochvílová, Petra Laufková, Pavel Štichauer –––– 45 4.0 Introduction –––– 46 4.1 Definition of causativity and its forms of expression –––– 46 4.2 Causativity in Romance languages –––– 48 4.2.1 Analytic type –––– 48 4.2.2 Synthetic type –––– 49 4.2.3 Characteristics of the Romance construction hacer/fare/faire/fazer + infinitive –––– 49 4.3 Causativity in Czech –––– 50 4.3.1 Word-formatting causativity –––– 51 4.3.1.1 Verbs derived from another verb –––– 52 4.3.1.2 Verbs derived from an adjective –––– 53 4.3.1.3 No change in the lexical basis, expressing causativity through a prefix roz- –––– 53 4.3.2 Semantic causativity –––– 54 4.3.2.1 Suppletive types –––– 54 4.3.2.2 Causative interpretation resulting from syntax –––– 55 4.3.3 Analytic causativity –––– 55 4.3.3.1 Causative verbs followed by a subordinate clause –––– 55 4.3.3.2 Causative verbs followed by a nominal syntagma –––– 55 4.3.3.3 (Semi-)causative verbs followed by an infinitive –––– 56 4.4 Our typology of Czech respondents –––– 57 4.5 Methodology –––– 59 4.6 Causative constructions in Romance – formal comparison –––– 60 4.7 Analysis of Czech respondents –––– 62 4.7.1 Primary Czech respondents –––– 64 4.7.1.1 Type 3 – shodit type (hacer caer / far cadere / faire tomber / fazer cair) –––– 64 4.7.1.2 Type 8 – what makes you think that > proč myslíte? (‘why do you think that?’) –––– 66 4.7.1.3 Type 4 – dát vypít type –––– 68 4.7.2 Secondary Czech respondents –––– 69 4.7.2.1 Type 5 – dohnat kslzám type –––– 69 4.7.2.2 Type 9 – other translation –––– 71 4.7.2.3 Type 7 – způsobit, že tál type –––– 71 4.7.2.4 Type 1 – rozplakat type –––– 73 4.7.2.5 Type 2 – posadit type and type 6 – způsobit tání type –––– 73 4.7.2.6 Type 10 – no translation –––– 73 4.8 Conclusions –––– 73 5. ingressive periphrases in romance and their czech respondents Dana Kratochvílová, Jaroslava Jindrová, Pavel Štichauer, Eliška Třísková –––– 79 5.0 Introduction –––– 80 5.1 Verbal periphrases in Romance –––– 80 5.1.1 Approaches to verbal periphrases and the goal of our study –––– 81 5.2 Aspect and Aktionsart –––– 82 5.2.1 Aspect and Aktionsart in Romance languages –––– 82 5.2.2 Aspect and Aktionsart in Czech –––– 83 5.2.3 Verbal periphrases and the relationship to aspect and Aktionsart –––– 84 5.2.4 Ingressive MoA –––– 85 5.2.4.1 Initial ingressivity in Romance languages –––– 85 5.2.4.1.1 Derivative ingressive MoA in Romance –––– 85 5.2.4.1.2 Analytical ingressive MoA in Romance –––– 85 5.2.4.2 Initial ingressivity in Czech –––– 88 5.2.4.2.1 Derivative ingressive MoA in Czech –––– 88 5.2.4.2.2 Analytical ingressive MoA in Czech –––– 91 5.3 Corpus analysis –––– 93 5.3.1 Methodology –––– 93 5.3.2 Results of the corpus analyses –––– 95 5.3.2.1 Ingressive constructions expressing the mere beginning of aprocess –––– 95 5.3.2.2 Ingressive constructions expressing the beginning of aprocess and the notion of effort by part of the subject –––– 96 5.3.2.3 Ingressive constructions expressing the beginning of aprocess and the notions of suddenness and unexpectedness –––– 99 5.3.2.4 Ingressive constructions expressing the beginning of aprocess and the notions suddenness, abruptness and previous retention –––– 101 5.3.2.5 Ingressive constructions expressing the beginning of aprocess and the notions of suddenness, abruptness and vehemence –––– 102 5.4 Concluding remarks –––– 103 6. the romance gerund and its czech respondents Olga Nádvorníková, Leontýna Bratánková, Štěpánka Černikovská, Jan Hricsina –––– 107 6.0 Introduction –––– 108 6.1 Morphology of the Romance gerund –––– 108 6.2 Romance gerund as aconverb –––– 109 6.2.1 Syntactic functions of the Romance gerund –––– 110 6.2.2 Semantic interpretation of the (adverbial) Romance gerund –––– 112 6.3 Typology of Czech respondents of the Romance gerund –––– 114 6.4 Data elaboration and quantitative analysis of the Romance gerund –––– 115 6.4.1 Factors influencing the frequency of the Romance gerund –––– 117 6.4.2 Syntactic functions of the Romance non-periphrastic gerund –––– 118 6.5 The adverbial Romance gerund and its Czech respondents –––– 121 6.5.1 Semantic types of the Romance adverbial gerund and the Czech transgressive –––– 121 6.5.2 Czech respondents of the Romance gerund –––– 129 6.5.2.1 Finite verbs as respondents of the Romance gerund –––– 131 6.5.2.1.1 Coordinate finite clause as arespondent of the Romance gerund –––– 131 6.5.2.1.2 The subordinate finite clause as arespondent of the Romance gerund –––– 132 6.5.2.2 Nominalisations as respondents of the Romance gerund –––– 134 6.5.2.3 Non-finite verb forms as respondents of the Romance gerund –––– 136 6.6 Conclusion –––– 144 7. formal expressions vs abstract linguistic categories: coming to terms with potential (non-volitional) participation, iterativity, causation, ingressivity and adverbial subordination Petr Čermák, Dana Kratochvílová, Olga Nádvorníková, Pavel Štichauer –––– 147 7.0 Introduction –––– 148 7.1 Correspondences of the analysed phenomena across Romance languages –––– 149 7.2 Czech respondents of the analysed phenomena vs systemic counterparts –––– 151 7.3 Exploiting the parallel corpus in search of language universals and abstract categories –––– 153 Bibliography –––– 154 1. expressions of potential participation, iterativity, causation, ingressivity and adverbial subordination in the light of parallel corpora petr čermák dana kratochvílová olga nádvorníková pavel štichauer 16 2.0 introduction Multilingual corpora strongly changed the research paradigm in contrastive studies, making it possible to base the contrastive statements not only on intuition but on large corpus data. As pointed out by Altenberg – Granger (2002, 7), bilingual and multilingual corpora have brought about a revival of interest in contrastive linguistics, since they opened up new possibilities of research, based on empirical data. According to these authors, “the information gained from corpora is both richer and more reliable than that derived from introspection” (ibid.). Specific methods and approaches subsequently developed, e.g. bi-directional analysis (‘Johansson’s procedure’, see Johansson 2007) or the use of ‘translation counterparts as markers of meaning’ (Malá 2013 and 2014). With the analysis of the overall pattern of translation correspondence, we can ‘see through multilingual corpora’ (Johansson 2007) and shed new light on the differences and similarities between the languages compared. These developments would not be possible without the constitution of a rigorous methodology of the exploitation of multilingual corpora, taking into account, on the one hand, the limitations of the representativeness of these corpora in terms of size and composition, and, on the other hand, the potential specific features of the language of translation (see Nádvorníková 2017a and 2017b). This chapter first provides a brief summary of the basic methodological principles of corpus-based contrastive research (Section 2.1) to subsequently explain the strengths and the limitations of the corpora used in the research introduced in this book (Section 2.2). 2.1 corpus-based contrastive research methodology Most corpora used in contrastive corpus-based research is comprised of original, non-translated texts and the corresponding translations. These corpora are mostly called ‘parallel’ (see Xiao – Yue 2009, 241–242; Aijmer 2008, 276; Granger 2003, 21), with a potential distinction between unidirectional parallel corpora (i.e. containing 2. corpus design & corpus-based contrastive research methodology 17 translations only in one translation direction, e.g. from English into Norwegian and not from Norwegian into English) and bi-directional ones (i.e. comprising source and target texts in both directions of translation).2 If in a bidirectional parallel corpus, the non-translated components have the same characteristics in terms of size and composition (and, eventually, sampling techniques), the parallel corpus may be called ‘comparable’.3 Nevertheless, the use of the terms ‘parallel corpus’ and ‘comparable corpus’ in contrastive corpus-based research is not consistent. First, a comparable corpus cannot contain translations or cannot be multilingual. In the former, the corpora in the two (or more) languages are of the same size and composed of the same text types, but are not translations of each other.4 In the latter, the comparable components are written in the same language but differ in specific properties: e.g. the corpus Jerome, comprising translated and non-translated texts in the same language – Czech (see Chlumská 2013 and 2017 and an example of its exploitation in 2.2). Asimilar terminological confusion can be observed in the term ‘parallel’: Granger (1996, 38) used the term ‘parallel corpus’ for corpora comparable in terms of size and composition. In this research, we will follow the most consensual use of the aforementioned terminology, reserving the term ‘parallel’ for bilingual or multilingual corpora containing translationally equivalent texts (see e.g. Peters – Picchi – Biagini 2000, 74) and the term ‘comparable’ for corpora with the same size and composition (see also Xiao and Yue 2009, 240–241 or Aijmer 2008, 276). However, more important issues discussed in the literature related to the use of parallel corpora in contrastive research concern methodological principles and restrictions that have to be taken in consideration while making contrastive statements on the basis of the comparison of original texts and the corresponding translations. The first question that arises in this context is the delimitation of the units compared: what is the source item and what is its ‘equivalent’ in translation? The identification of the source unit and its potential counterparts requires a deep insight into their valeur, i.e. their position in the system of all the languages under scrutiny. In the research introduced in this book, based on the comparison of four different Romance languages and Czech, this question becomes even more pressing since the language units entering the comparison may have a different valeur in the source Romance languages. The gerund, for example, has a different frequency, different functions and a different position in the system of non-finite verb forms in Italian, French, Spanish and Portuguese. For this reason, a tertium comparationis of the cross-linguistic term 2 Xiao and Yue (2009, 241) also mention multidirectional corpora where the same source text can be compared with its translations into several languages. 3 The most prominent example of comparable parallel corpus English-Norwegian Parallel Corpus (ENPC, see Johansson 2007). 4 See the definition of a comparable corpus in Aijmer (2008, 276): “A comparable corpus on the other hand does not contain translations but consists of texts from different languages which are similar or comparable with regard to a number of parameters such as text type, formality, subject-matter, time span, etc.” 18 converb was suggested for its comparison with Czech (see Nádvorníková et al. this volume).5 The identification of the ‘equivalent’ of the source unit in translation has to address numerous issues. First, from the point of view of translation studies, the analysis of ‘translation equivalence’ at the level of only words or sentences is inaccurate since translators do not translate words or sentences but texts. In addition, the term ‘equivalence’ is itself questionable, as it can be understood both in a descriptive and prescriptive meaning (what corresponds to the source item or what should correspond, see e.g. Guidère 2011, 83). Thus, in our book, we distinguished the two meanings by using the term ‘respondent’ for concrete translation solutions in Czech, and by reserving the term ‘counterpart’ for potential systemic equivalents, see Čermák – Kratochvílová – Nádvorníková – Štichauer (this volume, Section 1.4.2).6 However, in the actual analyses of bilingual parallel concordance, a researcher has to encounter a large range of respondents, i.e. also multiple candidates to the systemic counterparts of the search unit. The crucial issue, in this case, is the distinction between the particular translation solutions and the prevailing types of respondents (recurrent translation patterns, see Krzeszowski 1990, 27), which potentially reveal the systemic equivalences. In fact, solid contrastive statements can only be formulated on the latter, whereas the former can be used in a study in the domain of translation studies focussed on special translation techniques (e.g. modulation or transposition, see Vinay – Darbelnet 1995) or translation quality assessment (e.g. omissions or additions).7 The last issue defining the usability of parallel (translation) corpora in contrastive research is related to potential specific features of the language of translated texts, different from the non-translated ones. These differences may be due to the influence of the source language (interference, shining through), but also due to the translation process itself (so-called translation universals, see Baker 1996, 176–177 for the definitions given below). The specific features of translation that are the most discussed in literature are simplification (“The idea that translators subconsciously simplify the language or message or both”; for research see e.g. Vanderauwera 1985; Laviosa 2002, or Cvrček – Chlumská 2015), explicitation (“The tendency to spell things out in translation, including, in its simplest form, the practice of adding background information”, see e.g. Blum-Kulka 1986; Olohan – Baker 2000; Pápai 2004, or Nádvorníková 2017c) and normalisation (“The tendency to conform to patterns and practices that are 5 The necessity of tertium comparationis in contrastive linguistics is mentioned e.g. in Goddard – Wierzbicka (2008); see also Altenberg – Granger (2002, 15–18). Barlow (2008) points out that without a common basis for the comparison of the analysed phenomena, the contrastive analysis will always compare pears and apples; in the best of the cases, however, contrastive analysis compares different kinds of apples (Barlow 2008, 101). 6 See a similar distinction in Johansson (2007, 5; translation correspondence vs systemic equivalence). 7 Missing equivalents in translation may be due not only to the (voluntary or involuntary) omissions performed by the translator, but also to technical issues (misaligned segments). Moreover, the missing counterpart may be compensated outside the given parallel segment. 2. corpus design & corpus-based contrastive research methodology 19 typical of the target language, even to the point of exaggerating them”, see e.g. May 1997 or Kenny 2001).8 Contrastive research based on parallel (translation) corpora implemented several methodological principles designed to identify and/or avoid the influence of the specific features of translation. The basic principle is the systematic identification of the direction of translation: indeed, in a corpus of mixed directions of translation, potential sources of interference are multiplied. This principle is often combined with the bi-directional analysis, which also compares the translation respondents of a given item in the opposite direction of translation. A bi-directional analysis is especially in use in comparable corpora, where the components in all the directions of translation are comparable in size and composition (see above). We did not apply the bidirectional analysis systematically to all the topics in our analysis because the subcorpora of translations from Czech into Romance are much smaller than those in the opposite direction of translation and thus not comparable. Therefore, the bidirectional analysis was tested only in the case of the gerund, in order to establish to what extent the Czech transgressive corresponds to the Romance gerund (see Nádvorníková et al. this volume). The specificities of parallel corpora (both in the translated and non-translated parts) can also be identified by the comparison with the corresponding monolingual reference corpora. In fact, parallel (translation) corpora, by definition, cannot be representative of the entirety of the language use, since they are limited to texts and the types of text being translated (some types of text, e.g. letters or e-mail messages, are rarely translated) or because there are more translations in one direction of translation (e.g. from English into Czech) than in another (e.g. from Czech into English), cf. Granger – Lerot – Petch-Tyson (2003, 20). For this reason, it is recommended to compare the results obtained from parallel corpora to those extracted from monolingual corpora, referential for the given languages (see e.g. Altenberg – Granger 2002, 9). However, we did not apply this procedure to our study, since a systematic comparison of the results in the four topics to five reference corpora (in Czech and in the four Romance languages) would be to go beyond the scope of this book. Nevertheless, we decided to verify the potential specificity of the language of translation in our research at least in the first topic addressed in this book: causative constructions (see Čermák – Kratochvílová et al. this volume). As explained in that chapter, the Romance causative construction (hacer/fare/faire/fazer + infinitive) has a wide range of types of respondents in Czech (synthetic as well as analytic, see Section 4.3). If the Czech translations were influenced by the source language, we could expect there to be a higher frequency of the analytic respondent nechat + infinitive (the closest by its form to the Romance causative constructions), in comparison with the non-translated texts. In order to test this assumption, we used the corpus Jerome (comparable translation 8 The specific language of translation is sometimes called ‘translationese’ (see e.g. Baker 1993 or Mauranen 1999). However, as pointed out by Chlumská (2017, 23), ‘specific features of translation’ and ‘translationese’ are not synonymous, since the latter conveys a negative evaluation. 20 corpus of Czech, see Chlumská 2013 and http://wiki.korpus.cz/doku.php/en:cnk:jerome). The corpus comprises translated and non-translated texts in equal amounts (mostly fiction, but also a subcorpus of non-fiction). The whole corpus contains 85 million tokens but also includes a smaller subcorpus (5 million tokens) balanced according to the source languages (14 languages, including the four Romance languages under study in this book).9 The proportions of source languages in the unbalanced corpus correspond to their proportion in the Czech publishing market; consequently, English as a source language prevails. For our experiment, we used both variants of the Jerome corpus – balanced as well as unbalanced. The results of the corpus search are shown in Table 2.1: Tab. 2.1. Comparison of the frequency of Czech causative construction nechat + infinitive in the Jerome corpus Jerome corpus (nechat + infinitive)10 Unbalanced corpus Balanced corpus Non-translated texts Translated texts Non-translated texts Translated texts Size of the corpus (in tokens) 42,401,470 42,563,842 2,547,367 2,540,043 Abs.fq. 5,107 6,389 401 297 Rel.fq. (ipm) 120 150 157 117 Dice coefficient 0.22 –0.30 Table 2.1 shows that in absolute as well as in relative frequencies (ipm), the frequency of the construction nechat + infinitive in translated and non-translated texts are different. In the unbalanced corpus, the frequency is higher in the translated texts, whereas in the balanced corpus, the result is the opposite. According to the chi-squared test, both differences are statistically significant (at p<.001). However, as shown in Cvrček – Kodýtek (2013), the statistical significance does not necessarily mean the statistical relevance (the so-called effect size), i.e. whether it is possible to identify a relevant factor behind it. In order to test the effect size, we used the Dice coefficient, based on the comparison of the relative frequencies: Dice = 2x (ipm1 – ipm2) / (ipm1 + ipm2) The Dice coefficient results vary between −2 and 2, which are both extreme values signalling high relevance of the difference in frequency. However, in our analysis of 9 Since the design of the corpus is synchronic, it only includes translations published after 1992. In addition, it avoids the potential influence of the authors’ idiolects by limiting the number of texts written by one author to three books only. More books translated by one translator are accepted although the authors of the originals must be different. 10 In order to reduce the amount of extraction noise and increase the comparability of the results in the two subcorpora, we used a simplified regular expression [lemma="nechat"] [tag="Vf.*"], without potential elements between the verb nechat and the infinitive. Despite this limitation, we consider the results reliable. 2. corpus design & corpus-based contrastive research methodology 21 the frequency of the Czech causative construction nechat + infinitive in translated and non-translated texts, the Dice coefficient is 0.22 in the unbalanced corpus and −0.30 in the balanced corpus. These results show that the statistical relevance of the differences observed in Table 2.1 is minimal, which means that the influence of the translated text on the frequency of nechat + infinitive is not found. Although the detailed analysis in the Jerome corpus was conducted on only one topic addressed in this book (causative constructions), we dare say that the other domains under examination in this study are not substantially influenced by the specificities of the translated language either. However, other limitations of the corpus, especially those related to the composition of the corpus and the size of the different language subcorpora, may come into play. For this reason, we introduce below (Section 2.2) a detailed description of the corpus used in this study and take into consideration its possible limitations throughout this book. 2.2 corpora used in this study Data for the research in the four topics introduced in this book was drawn from a large multilingual (parallel) corpus named InterCorp (http://ucnk.korpus.cz/intercorp/?lang=en, Čermák and Rosen 2012 or Nádvorníková 2016 in French). The InterCorp parallel corpus project was started in 2005 by the Institute of the Czech National Corpus (http://ucnk.ff.cuni.cz) with the first version of the corpus published on the internet in 2008. Since then, a new version of the corpus has been launched every year, which has improved the corpus interface functions and added new texts and sometimes new languages to the corpus (see the versions listed at http://wiki.korpus.cz/doku.php/en:cnk:intercorp). The present study was carried out on data extracted from version 6 of the corpus (see http://wiki.korpus.cz/doku.php/ en:cnk:intercorp:verze6 and the detailed description below).11 The search in the corpus is free for non-commercial uses after registration (https:// www.korpus.cz/signup)12 and extensive research has already been conducted on it (see the database of publications at https://www.korpus.cz/biblio). Nevertheless, the corpus is not used only in (contrastive) linguistic research but also in everyday practice by translators, students and language teachers (see e.g. https://korpus.cz/proskoly). In addition, in 2015, an online dictionary based on the InterCorp data was made available on the internet (http://treq.korpus.cz/, see Škrabal – Vavřín 2017). 11 As mentioned in Čermák – Kratochvílová – Nádvorníková – Štichauer (this volume, Introduction), the first (Czech) version of this book, resulting from the first stage of our research, was published in 2015, and the data was extracted in 2013, from the latest version of the corpus available at that time (version 6). The InterCorp parallel corpus is nowadays (in 2020) at version 12, which is obviously larger than the previous versions. However, for the present monograph, it was not possible to extract and examine completely new data. 12 After signing a non-profit licence agreement, the Institute of the Czech National Corpus can also provide texts from the InterCorp parallel corpus as bilingual files including shuffled pairs of sentences. 22 InterCorp parallel corpus contains the originals and the corresponding translations in 40 languages, including the four Romance languages under study in this book (Spanish, Italian, French, and Portuguese).13 For more than half of the languages included in the corpus, the texts were lemmatised and POS-tagged (with the exception of e.g. Arabic, Hindi, Hebrew, Chinese, Vietnamese etc.). For all the four Romance languages under examination in this study, both lemmatization and POS-tagging are available in the corpus. Nevertheless, in order to exclude potential mistagged tokens, we preferred regular expressions to the POS-tags in the specific corpus searches, wherever it was possible (for example, the gerund was searched via the suffixes in the four Romance languages, see Nádvorníková et al. this volume).14 The originals and the translations in the corpus are aligned at the sentence level using the sentence aligner hunalign (http://mokk.bme.hu/resources/hunalign/, see Varga et al. 2005).15 Since Czech is the pivot language of the project, all texts are aligned with this version, and through this version to other languages, which makes it possible to perform a multilingual search in the corpus interface (KonText). Among the fiction texts available in most language versions, are obviously translations from English,16 but, surprisingly, the text available in the InterCorp parallel corpus in most of the translations was not written in English, but in French: it is Le Petit prince by Antoine de Saint-Exupéry (available in 29 translations):17 [FR] « Moi, se dit le petit prince, si j’avais cinquante-trois minutes à dépenser, je marcherais tout doucement vers une fontaine… » (Antoine de Saint-Exupéry, Le Petit prince) [CS] „Kdybych já měl padesát tři minuty nazbyt,“ řekl si malý princ, „šel bych docela pomaloučku ke studánce…“ (transl. Zdeňka Stavinohová) [DE] „Wenn ich dreiundfünfzig Minuten übrig hätte“, sagte der kleine Prinz, „würde ich ganz gemächlich zu einem Brunnen laufen…“ (transl. Grete Leitgeb; Josef Leitgeb) 13 Arabic (ar), Belarussian (be), Bulgarian (bg), Catalan (ca), Czech (cs – pivot language), Danish (da), German (de), Greek (el), English (en), Spanish (es), Estonian (et), Finnish (fi), French (fr), Hebrew (he), Hindi (hi), Croatian (hr), Hungarian (hu), Icelandic (is), Italian (it), Japanese (ja), Lithuanian (lt), Latvian (lv), Macedonian (mk), Malay (ms), Maltese (mt), Dutch (nl), Norwegian (no), Polish (pl), Portuguese (pt), Romany (rn), Romanian (ro), Russian (ru), Slovak (sk), Slovene (sl), Albanian (sq), Serbian (sr), Swedish (sv), Turkish (tr), Ukrainian (uk), Vietnamese (vi). 14 The amount of mistagged tokens is minimal in the corpus, with the exception of the past transgressive in Czech, where 50% of the occurrences were noises (see Nádvorníková et al. this volume). 15 For details about the design of the InterCorp parallel corpus and technical aspects of its constitution, see Vavřín – Rosen (2008) or Čermák (2010). 16 E.g. Harry Potter’s stories by J.K. Rowling or books by Lewis Carroll, Georges Orwell, Douglas Adams or J.R.R. Tolkien. 17 Le Petit prince is available in Czech,Belarussian,Bulgarian,Catalan,Danish,German,Lower Sorbian,Greek, English,Spanish,Finnish,Hindi,Croatian,Upper Sorbian,Hungarian,Italian,Latin, Latvian, Macedonian, Dutch,Polish, Portuguese,Romanian,Russian,Slovak,Slovenian,Swedish,Serbian Cyrilic,Ukrainian. On the top list of texts available in the most translations in the core of InterCorp, are other texts also written in Romance languages: Il nome della rosa by Umberto Eco (in 22 translations, the 9th position) and Paulo Coelho’s O Alquimista (also 22 translations). 2. corpus design & corpus-based contrastive research methodology 23 [EN] “As for me,” said the little prince to himself, “if I had fifty-three minutes to spend as I liked, I should walk at my leisure toward a spring of fresh water.” (transl. Katherine Woods) [ES] —Si yo dispusiera de cincuenta y tres minutos —pensó el principito— caminaría suavemente hacia una fuente... (transl. Bonifacio del Carril) [HIN] “  ”  , ‚  -  …  ” (transl. , ) [IT] “Io”, disse il piccolo principe, “se avessi cinquantatré minuti da spendere, camminerei adagio adagio verso una fontana…” (transl. Nini Bompiani Bregoli) [PT] “Eu, pensou o principezinho, se tivesse cinqüenta e três minutos para gastar, iria caminhando passo a passo, mãos em o bolso, em a direção de uma fonte…” (transl. Frei Betto) The corpus is divided into a core part and so-called collections. The core consists mostly of fiction and partly of non-fiction. The collections are comprised of various types of texts: movie Subtitles, Acquis communautaire, transcripts of debates in the European Parliament and journalistic texts (collections SYNDICATE and Presseurop).18 The core of the corpus and the collections differ not only in the text types included but, more importantly for our research, in the quality of the data: unlike the collections, the texts in the core of the corpus are all proofread and the quality of their alignment is semi-manually checked using the InterText editor for aligned parallel texts (see Vondřička 2014). Moreover, texts in the core of the corpus are mostly translated by professional translators and revised in publishing houses, unlike e.g. the collection of movie subtitles, and the direction of translation is identified with certainty in this part of the corpus. As can be seen in Section 2.1, the distinction between the source and the target languages is a crucial factor in corpus-based contrastive research; consequently, we limited the data for the research introduced in this book only to original texts in the core part of the corpus, in the four Romance languages under examination.19 Thus, the research in the four topics introduced in this book was conducted on the same subcorpora, as defined in Table 2.2. Table 2.2 shows that the core of the corpus represents only a minor part of the InterCorp parallel corpus (cf. columns 2 and 3 in the table) and none of the core subcorpora can be considered by its size as representative for the given language. For this reason, in the analyses introduced in this book, the frequency counts were limited to the minimum, and if introduced, their main purpose is to identify the size of the dataset examined. In addition, the four subcorpora are not comparable one to each other: the largest, for Spanish, is six times larger than the others. For this reason, in the analyses, datasets in Spanish may prove to be more quantitatively valuable and reliable than in the other three languages (see e.g. Štichauer et al. this volume, Section 3.4.1). 18 In 2017, a new collection was added to the corpus:18translations of the Bible. 19 This means that we excluded from the core of the corpus the texts translated from a third language (e.g. all translations from English), since different source languages may give rise to other interferences. 24 Tab. 2.2. Subcorpora of the InterCorp parallel corpus used in this research Subcorpus Total No of tokens (collections & core) Subcorpus used in this study (the core of the corpus limited to translations from the four Romance languages into Czech) No of tokens No of texts No of (different) authors No of (different) translators ES>cs 73,002,746 9,326,150 105 (36 Europ. / 69 Am.) 43 (21Europ. / 22 Am.) 44 IT>cs 54,564,618 1,631,204 17 11 9 PT>cs 58,531,766 1,485,541 15 11 (8 Port. / 3Braz.) 9 FR>cs 62,288,229 1,533,451 31 23 24 As mentioned above, most of the texts in the core part of the InterCorp parallel corpus are classed as fiction; in the four Romance languages under scrutiny in this study, non-fiction is mostly represented by history: Giuliano Procacci Storia degli italiani in Italian or Dames du XIIe siècle by Georges Duby or books about history for young people (Marco Polo, Alexandre le Grand) in French. In Spanish, we find in non-fiction e.g. La rebelión de las masas by José Ortega y Gasset. In fiction, the most represented authors (in number of tokens) are, for example, Arturo Pérez-Reverte, Isabel Allende or Gabriel García Márquez in Spanish; Umberto Eco, Elsa Morante or Alessandro Baricco in Italian; Louis Ferdinand Céline, Michel Houellebecq or Bernard Werber in French; and José Maria Eça de Queirós, Jorge Amado or João Giumarães Rosa in Portuguese. Since the InterCorp parallel corpus is synchronic, most texts were published after 1950, with the exception of major works in the given literature, e.g. Marcel Proust in French or the above-mentioned José Maria Eça de Queirós in Portuguese. In addition, both European and American authors are represented in Portuguese and Spanish. In the analyses presented in this book, we attempt to identify and minimise the potential impact of the authors’ or translators’ idiolects by systematically mentioning the name of the author and the translator in the examples.20 Despite the aforementioned limitations in size and in composition of the subcorpora of the InterCorp parallel corpus used in this research, we believe that the results obtained in the four topics introduced in this book (complex words with the suffix -ble/-bile/-vel and the prefix re-/ri-, causative construction hacer/fare/faire/fazer + infinitive, ingressive verbal periphrases and the gerund) are reliable and provide a good example of the use of parallel corpora in contrastive research. Future research may verify and refine our findings on larger corpora (not only a new version of the InterCorp parallel corpus but on monolingual reference corpora for the four Romance languages as well) and develop further contrastive analyses comparing the four Romance languages. 20 Olohan (2004, 28) points out that not mentioning the name of the translator in a corpus-based contrastive study may be a signal of the underestimation of the translation task and process. 3. morphologically complex words in romance and their czech respondents pavel štichauer jan hricsina jiří jančík jaroslava jindrová zuzana krinková daniel petrík 32 pressed by larger contextual information. This is what we particularly find with polysemous verbs where other semantic features are also connected with the prefix re-/ri-. We now turn to a detailed presentation of the quantitative distribution of the single types discussing some of the peculiar cases. We begin with the suffix -ble/-bile/-vel and, subsequently, we move on to the prefix re-/ri-. 3.4 the modal suffix -ble/-bile/-vel The Latin suffix -abilis/-ibilis gave rise, in Romance, to various outcomes, such as both -abile and -evole, in Italian, where eventually the only Latinate form -abile prevails and can be said to be the only productive suffix. In Es., Fr. and Pt. we find a similar evolution with -ble and -vel being the forms of interest. In fact, we restrict ourselves, in what follows, to a synchronically active pattern where the prototypical group of formations are adjectives derived from transitive verbs with a clear modal meaning glossed as ‘what can/cannot be done’. Nevertheless, we are forced to take into account all kinds of deviations from this core semantics as the range of possible meanings (as well as the type of verbs required for the derivation) is larger. We first briefly characterise the suffix (we follow here, e.g., Grossmann – Rainer 2004, 422–426; Bisetto 2009; Val Álvaro 1981; Grevisse – Goosse 2007, 169–173; Mateus 2003, 945; Pires de Oliveira – Ngoy 2007). As previously mentioned, the suffix requires transitive verbs with agentive subjects; we thus find the following restrictions on the base verbs: 1) Psychological verbs (whose subject can be broadly defined as experiencer) are ruled out, as witnessed by the impossibility to have, for instance, It. *preoccupabile (from preoccupare). This constraint is not absolute as, for example, in French where we do find examples of adjectives also derived from these verbs, such as affligeable, agaçable, aguichable, attristable (cf., e.g., Leeman – Meleuc 1990, 33). 2) Stative verbs, especially those where the subject assumes the role of a possessor, as can be seen in It. possedere → *possedibile, Fr. posséder → *possédable. As for the semantic interpretation, we can also find specific nuances, which go beyond the simple passive potentiality, as in Fr. souhaitable ‘desirable’, It. pagabile ‘payable’ while their meaning is rather deontic, paraphrasable as ‘what deserves to be desired’, ‘what must be paid’. We will discuss further issues relative to formal and semantic aspects below when addressing the corpus data. 3. morphologically complex words in romance and their czech respondents 33 3.4.1 data elaboration and analysis In order to obtain a frequency list of all adjectives with the suffix, we ran a simple tagbased query which had an identical form for all four subcorpora, depending on the exact form of the suffix. We thus have three basic queries: Es. + Fr. [lemma=".*ble" & tag="ADJ.*"] It. [lemma=".*bile" & tag="ADJ.*"] Pt. [lemma=".*vel" & tag="ADJ.*"] Of course, such a query yields a raw frequency list where all adjectives ending in the sequence in question appear. In a subsequent step, we have therefore proceeded to manually elaborate the frequency lists to weed out some of the particular lemmas, which we will now briefly discuss. It is possible to delimit four groups of these adjectives, depending on the formal and semantic transparency. Therefore, we define four classes, inspired by Tekavčić (1972, 75–76), where the transparency (referred below as T) is complete, partial or there is none: 1) Group T(1). In this class, we can find exactly what we are interested in: morphologically and semantically transparent adjectives, i.e. formations where the verbal base and the suffix -ble/-bile/-vel can be clearly discerned. Morphologically, this group is required to display no allomorphy between the verbal stem and the stem of the derived adjective. We thus have pairs like It. leggere – leggibile ‘to read – readable’. Semantically, the semantics of the adjective must satisfy, in a reproducible way, the potential reading (regardless of possible contextual nuances) glossed as ‘what can/cannot be done’. 2) Group partial-T(2). This group comprises adjectives, which are semantically compositional in the same way as those in group T(1), but formally they display a kind of allomorphy, as is found in pairs such as Es. ver – visible ‘to see – visible’. Of importance here is not only the straightforward semantic relationship between the base verb and the derived adjective but also the systematic formal relationship in that the allomorphy in question is not confined to just one pair but is found across a wide range of patterns (this requirement comes close to what Corbin 1987, 342 called allomorphic projection). 3) Group partial-T(3). In this group, there is also semantic transparency in that this class of adjectives show the basic modal meaning defined above although what is crucially lacking is the morphological motivation. We find what we could call baseless formations, i.e. adjectives that lack a verbal base altogether. This is the case, for example, of the adjective Es. vulnerable (along with the other Romance respondents), which may be linked to the verb herir (It. ferire, Fr. blesser). These adjectives are direct loanwords from Latin. 34 4) Group non-T(4). This negatively defined group comprises adjectives which, synchronically, do not display any semantic and morphological transparency. They lack a verbal base and do not satisfy the general meaning instruction. Examples, also current in English, which typically illustrate this group are probable, possible, etc. It is clear that, as mentioned above, class T(1) comprises those adjectives that we are interested in. However, it would be too hasty to immediately rule out the other three classes away as the semantic transparency in class T(3) and the morphological transparency in class T(2) might be a relevant factor. Therefore, we have weeded out the adjectives in class T(4) while selectively maintaining some of those pertaining to classes T(2) and T(3). Unsurprisingly, these formations are some of the most frequent adjectives. Their elimination, which in type frequency is not so relevant, affects the overall token frequency in that they represent between 40–50% of all the tokens. We now briefly describe these adjectives for all four languages under investigation. First, it needs to be said that in order to identify and discuss the typology presented above, we set up, as in the other case studies present in this book, a frequency limit of 10 (or 11, depending on the size of the four subcorpora) tokens. Such a limit is comprehensible as we are interested in recurrent translational patterns, and not in nonce-formations (although these are, of course, extremely important for other aspects such as productivity). As a result, there are four datasets for Es., It., Fr. and Pt. For Spanish, which is quantitatively far more represented than the other languages, we obtained a raw frequency list of 197 adjectives with the overall token frequency equal to 16,362. The elimination of 70 adjectives pertaining to classes T(4) and partly to T(2) and T(3) produces a reduced list of 122 adjectives with the token frequency equal to 7,959. Examples of the eliminated adjectives – almost identical for all languages under scrutiny – are posible, estable, irresponsable, temible etc. (for the complete list, see Čermák – Nádvorníková et al. 2015, 110–111). For Italian, we had a significantly smaller dataset with 50 adjectives each having more than 11 tokens with the overall token frequency equal to 1791. Eliminating 23 adjectives of the classes T(4) and, partly, T(2), T(3), we arrived at the list of 27 adjectives with the overall frequency equal to 622 (again, for details, see Čermák – Nádvorníková et al. 2015, 98–99). In the French dataset, the situation is much the same with a raw list including 76 adjectives (each again with the frequency of 11 and more tokens), with the overall frequency of 2,978. Elimination of 52 formations (such as invraisemblable, impassible, implacable, pitoyable etc., see Čermák – Nádvorníková et al. 2015, 140–141) yields a list of 24 adjectives whose overall token frequency equals 587. Finally, Portuguese represents the smallest dataset with 44 adjectives each having more than 10 tokens with the overall token frequency of 1,478. Again, the elimination of the undesired formations yields a list comprising 17 adjectives with the overall frequency equal to only 342 tokens (see Čermák – Nádvorníková et al. 2015, 124). The summary of all four datasets is given in Table 3.1. 3. morphologically complex words in romance and their czech respondents 35 Tab. 3.1. Frequency data for adjectives with the suffix -ble/-bile/-vel Language Type frequency Overall token frequency Spanish 122 7,959 Italian 27 622 French 24 587 Portuguese 17 342 As is clear from this direct comparison, the only quantitatively valuable dataset is that of Spanish, which is due, as for the other phenomena investigated in this book, to the larger size of the Spanish-Czech parallel corpus. Taking into account this serious limitation, which will be overcome in the future enlargement of the Italian, French and Portuguese subcorpora, we can nevertheless present the quantitative distribution of the types defined above. 3.4.2 quantitative distribution of the types Beginning with the absolute figures, which are not particularly telling since they do not allow for direct comparison, we summarise the results in Table 3.2: Tab. 3.2. Distribution of types for adjectives with the suffix -ble/-bile/-vel in the four subcorpora – absolute frequencies Language Frequency A B C1 C2 C3 D1 D2 Spanish 7,959 4,116 2,575 120 114 249 108 677 Italian 622 370 86 11 29 39 30 56 French 587 392 76 4 3 16 27 69 Portuguese 342 141 127 16 29 8 4 17 While acknowledging that the figures, especially for French and Portuguese, are so low, mainly because we have restricted ourselves to only those adjectives that reach the minimum of 10 or 11 tokens, we still have to acknowledge the basic difficulty when directly comparing these results. A better overview is obviously provided by relative frequencies. In Table 3.3 is the percentual distribution of the types according to the above-defined typology. Overall, the relative frequencies, though they must be taken with extreme caution, clearly show as a predominant solution type A, where the direct correspondence between the adjective with -ble/-bile/-vel and the Czech respondent with the suffix -telný is maintained. Indeed, more than half of the identified respondents belong to type A. The second more frequent type is B, where, as defined above, we still find an adjective, 36 but with a different suffix and with a slightly different semantic instruction. What is quite striking is that the syntactic solution of type C, further differentiated in three subtypes, is by far the least exploited respondent. On the contrary, the residual type D, with especially high figures for D2, represents an interesting, albeit a hardly generalisable, group of individual solutions, to which we will turn below. But before doing so, a mention is required of the important correlation between the typology and the four classes T(1), partial-T(2)/T(3), and non-T(4). Indeed, there appears to be a strong correlation between the type and the adjectival class defined in terms of total, partial or null transparency. It is perhaps not a surprising fact that adjectives of class T(1), i.e. those entirely transparent and compositional, cover almost 90% of the type-A respondents while those belonging to class partial-T(2), cover only 25% of the type-A respondents. Conversely, adjectives of the class partial-T(3), i.e. those where the semantic instruction is maintained but the verb base is entirely lacking, tend to be translated by the respondents defined as type-B. This correlation can safely be demonstrated for the Spanish dataset, not just because we have a large amount of data, but in the case of the other languages, this dependency turns out to be less significant. Therefore, we do not dwell on the details. Instead, we move on to the discussion of concrete examples, especially those deserving particular attention. 3.4.3 discussion of various examples As we have already seen, in the majority of cases (between 42% and 67% of all the tokens, depending on the dataset), there is the hypothesised correspondence between the suffix -ble/-bile/-vel and the Czech suffix -telný. Thus, we find a wide range of examples where adjectives, such as Es. invisible ‘invisible’ (along with the other Romance respondents), increíble ‘incredible’ or insoportable ‘unbearable’ match the Czech adjectives neviditelný ‘invisible’, neuvěřitelný ‘incredible’ or nesnesitelný ‘unbearable’. This is probably an uninteresting situation and it is worth exploring the marginal respondents of the other types. We mostly leave aside type B, since in this case, we have a synonymous adjective where not only the modality is coded in a less explicit way, but the choice of the adjective appears to be entirely unpredictable, as shown in example (1). Tab. 3.3. Distribution of types for adjectives with the suffix -ble/-bile/-vel in the four subcorpora – relative frequencies (%) Language % A B C1 C2 C3 D1 D2 Spanish 100 52 32 2 1 3 1 9 Italian 100 60 14 1 5 6 5 9 French 100 67 13 1 1 3 5 12 Portuguese 100 42 37 5 8 2 1 5 3. morphologically complex words in romance and their czech respondents 37 (1) Pt. Tenho uma pessoa respeitável, com bom paladar, muito escrupulosa em contas. → Mám spořádanou osobu, chutně vaří a je pečlivá v účtech. Literally: I have an orderly person. Eça de Queiroz, Bratranec Basilio (O Primo Basílio), transl. Zdeněk Hampl, Prague: Odeon, 1989. We pay some attention to the type-C respondents, simply because these require a kind of lexical transformation in that the adjective is translated by way of a syntactic construction. For instance, adjectives such as It. riconoscibile ‘recognisable’ are rendered here by way of byl k poznání (lit. ‘he was to recognising’), dal se poznat, which is a reflexive form of causative construction ‘he made himself recognise’. A similar, C1-type solution can be seen in example (2). (2) Fr. Deux, que vous restiez absolument invisible. → Zadruhé: aby vás nebylo vůbec nikde vidět. Literally: so that it won’t be possible to see you. Michel Tournier, Tetřev hlušec (Le Coq de bruyère), transl. Václav Jamek, Prague: Odeon, 1984. Within type D, which is a heterogeneous group with hardly classifiable respondents (see Čermák – Nádvorníková et al. 2015, 118 for some examples from Spanish), we have identified at least two recurrent situations worth discussing. The first, D1, corresponds to a solution where some nominal, often deverbal, construction appears, sometimes bordering on idiomatic phrases, as in example (3): (3) Es. Después Rosario me secó, se secó, arregló el cuarto en un santiamén (es increíble lo hacendosa y práctica que es esta mujer) y se puso a dormir pues al día siguiente tenía que trabajar. → Pak mě Rosario utřela, sebe taky utřela, v cukuletu uklidila pokoj (je k nevíře, jak je přičinlivá a praktická) a šla spát, protože další den musí pracovat. Literally: it is to not-belief. Roberto Bolaño, Divocí detektivové (Los detectives salvajes), transl. Anežka Charvátová, Prague: Argo, 2009. The second, type D2, comprises cases where the meaning inherent in the suffixed adjective is coded by way of a completely different construction. A case in point is example (4): 38 (4) It. Poi incominciarono a muoversi in modo visibile anche Mecenate e Virgilio; (...). → Pak jsem uviděl, že se začali hýbat i Maecenas a Vergilius; (...). Literally: I saw that they started moving. Sebastiano Vassalli, Nespočet (Infinito numero), transl. Kateřina Vinšová, Prague: Paseka, 2003. This example, which is not infrequent within the D-type respondents, involves an interesting overhaul of the original text. In Italian, we have an adjective-based sequence they started to move in a visible way, and the “visibility” meaning is taken over by the verb-based construction with see. A similar example is in (5), where again, the “visibility” meaning is expressed by the negative form of the verb escape: (5) Pt. A mudança de tom, visível na forma como dissera a última frase, não escapou a Luís Bernardo. → Luísi Bernardovi neunikla změna tónu v poslední větě. Literally: to Luís Bernard did not escape the change of tone. Miguel Sousa Tavares, Rovník (Equador), transl. Lada Weissová, Prague: Garamond, 2006. Finally, as hinted at above, we also classify within the D2 types those respondents where the adjective with the suffix -ble/-bile/-vel is translated into Czech by way of a directly derived adverb, in which the modal meaning is clearly maintained, as can be seen in examples (6) and (7). (6) It. Ma il cuore di colui, frattanto, s’involava irresistibile verso Bella, (...). → Chlapcovo srdce se však nezadržitelně rozběhlo k Belle (...). Literally: irresistibly. Elsa Morante, Příběh vhistorii (La storia), transl. Zdeněk Frýbort, Prague: Odeon, 1990. (7) Es. Ante la necesidad de consolarme Paulina del Valle cambió de manera imperceptible para todos, menos para Frederick Williams. → Paulina del Valle se tím, jak mě musela utěšovat, změnila – i když tak nepostižitelně, že si toho všiml jen Frederik Williams. Literally: imperceptibly. Isabel Allende, Sépiový portrét (Retrato en sepia), transl. Monika Baďurová, Prague: BB Art, 2003. These examples are of interest because they show the limits of the proposed typology. If we insist on the part-of-speech correspondence, we are bound to rule out these 3. morphologically complex words in romance and their czech respondents 39 respondents from the A-type group. However, the formal and semantic correspondence is so strong that it might be more reasonable to split the A-type into an A1, the adjectival type, and an A2-type, the adverbial type. Such a modification of the typology would lead to a slightly different quantitative distribution, leaving in the D-type group only those respondents that resist straightforward classification. 3.5 the iterative prefix re-/riIn the four Romance languages under investigation, we find the prefix re-/riwhose major semantic instruction, as mentioned above, is a broadly defined iterativity, i.e. repetition of the action, activity expressed by the base verb. It is widely held that this general meaning of repetition also assumes different semantic realisations, some of them contextually induced rather than inherently present in the prefix itself. There appears to be an agreement in the literature that these semantic specifications can be as follows (cf. for Es. Martín García 1998, 45–51, Varela – Martín García 1999, 5012– 5013; for It. Grossmann – Rainer 2004, 155; for Fr. Grevisse – Goosse 2007, 186; Jalenques 2002, 84–87; Mok 1964, 106–109; for Pt. Vilela 1994, 117; Cunha – Cintra 1999, 88; Said Ali 2001, 188; Bechara 2009, 367): 1) Simple iterativity (i.e. a single, repeated instance of an action), e.g. Es. abrir – reabrir ‘to open – to reopen’. 2) Reversibility (i.e. a return to a preceding state of affairs), e.g. Pt. construir → reconstruir ‘to construct – to reconstruct’. 3) Movement in the opposite direction, e.g. Pt. fluir → refluir ‘to flow – to reflow’. 4) Reciprocity, e.g. It. abbracciare → riabbracciare ‘to embrace – to reembrace’. 5) Intensification (i.e. those cases where there appears to be no repetition of the action, just a reinforcing connotation), e.g. Fr. doubler – redoubler ‘to double – to redouble’. While we maintain that some of these semantic nuances are rather context-induced (for example, to reembrace can, in fact, be a repeated action by one and the same person and not just the reciprocal action of the other person), we view the semantic feature of intensification as something that often coexists with the basic meaning of iteration. These polysemous verbs tend to be quite frequent so it is important to keep the two meanings apart (as far as possible). 3.5.1 data elaboration and analysis As in the case of the adjectives with the suffix -ble/-bile/-vel, we ran a series of combined regular and tag-based queries, which had a similar form for all four subcorpora, 40 depending on the exact form of the suffix and on the tagset. In its basic form, the query had the following shape [lemma="re.*" & tag="V.*"] with obvious modifications according to the language in question. Of course, as above, we obtained raw frequency lists from which much extraction noise was to be eliminated. In particular, we weeded out the following two types of formations: 1) Various forms containing the sequence re-/rithat cannot be said to be a verbal prefix. 2) Verbs which are not synchronically segmentable in such a way that a verbal base and a prefix can be clearly discerned. This is obviously the case of verbs such as Es. recordar (cf. *cordar) ‘to remind’, It. ricevere (cf. *cevere) ‘to receive’ etc. Conversely, all verbs that are also polysemous and display an intensifying meaning have been maintained as they can all assume, in a particular context, an iterative interpretation. However, it is clear that these verbs reach high frequencies where the simple iterative meaning represents a marginal case. This is the reason why, as we shall see, type C is by far the most frequent respondent. Once the frequency lists have been elaborated in this way, we obtained the following datasets summarised in Table 3.4. Tab. 3.4. Frequency data for verbs with the prefix re-/riLanguage Type frequency Overall token frequency Spanish 69 7,287 Italian 56 3,526 French 84 7,099 Portuguese 34 1,037 It must be noted that, as in the preceding case, these figures include only those verbs whose token frequency is equal or higher than 10 tokens. Moreover, we should also note an apparently unexpected high type and token frequency of the French prefixed verbs despite the smaller size of the corpus. Although we would need to be reassured about this fact independently (on the basis of comparable frequency lists), we can say that French typically displays a large amount of such prefixed verbs, which thus seem to be particularly productive. 3.5.2 quantitative distribution of the types Table 3.5 shows the absolute figures which, as above, are misleading in that they do not take into account the different sizes of the four subcorpora. 3. morphologically complex words in romance and their czech respondents 41 Tab. 3.5. The distribution of types for verbs with the prefix re-/riin the four subcorpora – absolute frequencies Language Frequency A B C Spanish 7,287 881 2,983 3,423 Italian 3,526 475 640 2,411 French 7,099 1,178 1,431 4,490 Portuguese 1,037 202 98 737 Moving on to the relative frequencies, summarised in Table 3.6, we can see that although the clearly dominant type is the C-respondent, where no overt iterativity is explicitly expressed, a less evident situation is in the interplay between the A and B types. Tab. 3.6. The distribution of types for verbs with the prefix re-/riin the four subcorpora – relative frequencies (%) Language Frequency A B C Spanish 100% 12 41 47 Italian 100% 14 18 68 French 100% 17 20 63 Portuguese 100% 19 10 71 In fact, Spanish – which is, due to the corpus size, the most reliable dataset – displays a distribution where the analytic expression of repetition by way of an extra adverb turns out to be the least exploited. The B-type, where iterativity is inherently coded in the selected verb (or otherwise within the clause), reaches the same frequency as the previously mentioned C-type respondent. This situation, where direct comparison appears difficult, also arises because some verbs that would be clearly iterative in one Romance language are less so in the other languages. A case in point is It. riuscire, which has the dominant, lexicalised meaning ‘to succeed’ but is susceptible to be also used in the iterative meaning ‘to go out again’. The same case can be made for riguardare ‘to concern, to regard’ or ‘to look up again’. It is now obvious that the inclusion or exclusion of such verbs affects the frequencies in a sensible way (see the lists of verbs included for each language in Čermák – Nádvorníková 2015 et al., 94–95, 104– 106, 121, 132–134). We now consider various examples. 3.5.3 discussion of various examples We begin with the type-A respondents. As defined above, we view this type as an overt analytic solution that combines the base verb with an adverb carrying the iterative meaning, such as znovu, opět, zase etc. Examples (8) and (9) illustrate this. 48 1999, 2247; Skytte – Salvi 2001, Salvi – Vanelli 2004). As has also been proven by our corpus analysis, the borderline between activity and state is sometimes unclear (the state being the result of previous activity, i.e., zabít ‘to kill’ means both ‘to cause that somebody died’ and ‘to cause that somebody is dead’).27 4.2 causativity in romance languages Following Comrie’s (1986) typology, systemic expressions of causativity in Romance languages can be divided into two large groups: analytic type (see Section 4.2.1) and synthetic type (see Section 4.2.2). 4.2.1 analytic type The analytic expression of causativity in all the Romance languages studied is primarily represented by a construction that combines a verb meaning ‘to do’, ‘to make’ (hacer/ fare/faire/fazer) and the infinitive. This construction does not display any significant syntactic or stylistic limitations and will be analysed in the following sections. Its formal and semantic properties are comparable to the English construction ‘make sb do sth’: María me ha hecho reír / Maria mi ha fatto ridere / Marie m’a fait rire / A Maria fiz-me rir = ‘Mary made me laugh’. When analysing the hacer/fare/faire/fazer + infinitive construction, we will refer to, for the sake of simplicity, the Romance causative construction, even though we are well aware that this generalisation is somewhat inadequate since we leave aside other Romance languages, especially Romanian, where causative constructions behave differently from the syntactic point of view and are construed with the subjunctive, see Pîrvu (2010) and Ciutescu (2013). Apart from this clearly dominating structure, we can find other analytic expressions of causativity in the Romance languages. The usual form is a personal verbal form + infinitive, the most frequent being verbs corresponding to the English ‘to let’ (dejar/lasciare/laisser/deixar), which can combine with an infinitival completion and collaborate on the expression of causativity. In this construction, the causative meaning combines with other notions (especially the notion of ‘permission’). The grade of causativity of this construction has been explored in numerous studies (see Hernanz 1999, 2258–2265; Grevisse – Goosse 2008, 987; Riegel – Pellat – Rioul 2008, 254 and 443; Gonçalves – Duarte 2001, 660; Enghels – Roegiest 2012 and 2013). Analyses originating in Talmy’s force dynamics (Talmy 1985, 1988 and 2000) appear, along with others, in Soares da Silva (1998, 2001, 2004). As our main concern is the Czech respon27 In this study, we explore cases with the meaning ‘to cause that somebody is doing something’ only because the studied Romance construction does not have a factitive meaning. 4. causative constructions in romance and their czech respondents 49 dents of the Romance construction hacer/fare/faire/fazer + infinitive, constructions corresponding to the English ‘let/have sb do smth’ are not discussed here. The expression of the causee’s state (as a result of causer influence, see Section 4.1) can also be expressed by a semi-copulative verb followed by an adjective: Es. Una máscara me hacía invisible. (‘A mask made me invisible.’) It. Questo mi rende nervoso. (‘This makes me nervous.’) Fr. Il me rend fou. (‘He drives me crazy.’) Pt. O teu sorriso faz-me feliz. (‘Your smile makes me happy.’) 4.2.2 synthetic type Word-formatting processes creating causative verbs are rather limited in all studied languages. Causation resulting in the causee’s state can be expressed by deadjectival verbs such as: Es. triste (‘sad’) > entristecer (‘to make someone sad’, ‘to become sad’) It. pazzo (‘crazy’) > impazzire (‘to become crazy’, ‘to make someone crazy’) Pt. velho (‘old’) > envelhecer (‘to become old’). Nevertheless, in Romance languages (and also in Czech) the most frequent synthetic expression of causativity are unanimously semantic verbs, i.e. verbs that do not express causativity through morphological means (affixes): Es. derribar (‘to push down’) It. uccidere (‘to kill’) Fr. abattre (‘to tear down’) Pt. tombar (‘to knock down’). This numerous group is formally very heterogeneous and its analysis in Romance languages will not be presented in this study. 4.2.3 characteristics of the romance construction hacer/fare/faire/fazer + infinitive Romance causative constructions have been widely studied from different perspectives, the most important ones being: 1) Syntactic features: ● The definition, eventually the comparison with other infinitive constructions (Cano Aguilar 1977; Zubizarreta 1985; Hernanz 1999, 2247; Maldonado 2007; 50 Skytte – Salvi 2001, 499–509; Riegel – Pellat – Rioul 2008, 353; Arrais 1985; Soares da Silva 2005; Lopes – de Menezes 2018, Vesterinen 2012). ● The contrast between causative constructions with the infinitive and with finite verb: Es. Sus palabras me hacen pensar (‘Their words make me think.’) / Sus palabras hacen que piense (literally: ‘Their words make that I am thinking’); (Dowling 1981; Vesterinen 2008a, 2008b). ● For Portuguese, the use of the personal (inflected) infinitive is also a topic (Cunha – Cintra 1987; Bechara 2009, 754; Araújo 2012, 11). ● Syntactic restrictions of causative constructions (Delbecque – Lamiroy 1999, 2012–2013; Riegel – Pellat – Rioul 2008, 231; Labelle 2017; Hu, 2018). ● Use of clitics and reflexive pronouns (Fernández Ordóñez 1999, 1326–1327; Hernanz 1999, 2249; Riegel – Pellat – Rioul 2008, 255; Wilmet 1997, 464; Grevisse – Goosse 2008, 1002; Gonçalves – Duarte 2001, 659; Araújo 2012, 11–14). 2) Semantic features (Campos 1999, 1534; Grevisse – Goosse 2008, 1047; Riegel – Pel lat – Rioul 2008, 255 and 442; Vecchiato 2003). 3) Diachronic analyses (Davies 1995; Simone – Cerbasi 2001; Robustelli 2000). 4) Comparison with other languages (Katelhoen 2011; Gilquin 2015 and 2017; Heidinger 2015; Chen 2015). We shall not discuss formal syntactic characteristics of the analysed constructions; the main concern is the comparison with Czech. Our approach to analysing causativity is the closest to Enghels – Roegiest (2012) and Enghels – Roegiest (2013), who also work with corpus data and compare several languages. However, from the point of view of content, their studies do not overlap with ours since their main concern is the causativity expressed through verbs dejar and laisser. 4.3 causativity in czech In Czech linguistics, causativity has been analysed within the framework of the broader category of semantic relationships that can be found in a sentence (see Daneš – Hlavsa 1981). This category is generally understood as a verbal one (Komárek – Kořenský et al. 1986; Čechová et al. 1996; Čermák 2001; Štícha et al. 2013), and while there have been attempts to approach it as an abstract and wide theoretical notion (see Štícha 1981), the attention has mostly been focused on its concrete formal expressions, i.e. causative verbs. The defining characteristics of Czech causative verbs can be resumed as ‘to cause something to happen or someone to do something’ (Karlík – Nekula – Pleskalová 2002, 413), thus corresponding to the basic characteristics of the hacer/fare/ faire/fazer + infinitive construction. While for the Romance languages it is relatively easy to postulate one dominant systemic expression of causativity, the formal organisation of this category in Czech is more dispersed and comprises all three of Comrie’s types. Following the typology 4. causative constructions in romance and their czech respondents 51 presented in Karlík – Nekula – Pleskalová (2002, 412–413), we can distinguish three large categories of causative expressions that are encoded in Czech: 1) Synthetic causativity – word-formatting causativity 2) Synthetic causativity – semantic causativity 3) Analytic causativity A substantial feature of the majority of synthetic verbal causatives (i.e. types 1 and 2) is the incidence of pairs being made up of a non-reflexive, transitive, causative variant and a reflexive, intransitive and non-causative variant: (1) Pavel rozesmál Marii. Paul make.laugh-pst.3sg.nrefl Mary-acc ‘Paul made Mary laugh.’ (2) Marie se rozesmála. Mary refl-acc start.laughing-pst.3sg ‘Mary started to laugh.’ (3) Pavel otočil klíčem. Paul turn.around-pst.3sg.nrefl key-ins ‘Paul turned the key around.’ (4) Pavel se otočil. Paul refl-acc turn.around-pst.3sg ‘Paul turned around.’ The crucial fact is that the intransitive form is not conceived as causative, i.e. rozesmát se is not conceived as ‘to make oneself laugh’.28 4.3.1 word-formatting causativity This causative type is derived by the affixes of non-causative verbs; the semantic feature of causativity is thus manifested in the morphological structure of the verb (see Comrie’s 1986 morphological type, see also Perissutti 2017). There are several word-formatting processes that can be primarily associated with causativity, the crucial feature for their distinction being whether there is a change in the lexical basis 28 Exceptions are rare, for example, zapálit něco ‘to set sth on fire’, zapálit se ‘to set oneself on fire’. 52 or not. There are causatives that do not have the same lexical basis as their non-causative counterparts; with other verbs, the lexical bases are identical and causativity is expressed only by prefixes (with these verbs, the word-formatting relationship is transparent and the causativity is obvious). 4.3.1.1 verbs derived from another verb Change in the root (primarily -e- > -i-), generally also combined with a prefix such as poor u-. Sedět (‘to be seated’) > posadit (‘to make someone sit somewhere’): (5) Pavel sedí na židli. Paul-nom sit-prs.3sg.ncaus on chair-loc ‘Paul is sitting on a chair.’ (6) Marie posadila medvídka na židli. Mary-nom sit-pst.3sg.caus teddy.bear-acc on chair-acc ‘Mary put the teddy bear on a chair.’ Ležet (‘to be lying’) > položit (‘to put something on a surface’): (7) Pavel leží v posteli. Paul-nom lie-prs.3sg.ncaus in bed-loc ‘Paul is lying in the bed.’ (8) Marie položila papíry na stůl. Mary-nom lie-pst.3sg.caus papers-acc on table-acc ‘Mary put the papers on the table.’ With the prefixed verbs (posadit, položit), causativity is actually expressed through two formal features (prefix, root change). Therefore, these structures are transparent today for Czech speakers. However, with some old verbs without prefixes, the relationship with their original non-causative counterpart is not transparent anymore: vařit vodu (‘to boil water’) – vřít (‘to be boiling’), mořit (‘to bother’) – mřít (‘to be dying’ – archaic), točit (‘to spin’) – téci (‘to flow’), trápit (‘to torment’) – trpět (‘to suffer’) etc. 4. causative constructions in romance and their czech respondents 53 4.3.1.2 verbs derived from an adjective These are frequent causative verbs formed from adjectives. Several Czech adjectives can be used to form a deadjectival verb that expresses a change of state:29 modrý (‘blue’) > modřit (‘to make something blue’, ‘to paint something the colour of blue’) suchý (‘dry’) > sušit (‘to dry something’) (9) Látka je modrá. fabric-nom be-prs.3sg blue-f ‘The fabric is blue.’ (10) Celý den Marie modřila látku. whole-m day-nom Mary-nom make.blue-pst.3sg fabric-acc ‘Mary spent the whole day colouring the fabric blue.’ 4.3.1.3 no change in the lexical basis, expressing causativity through a prefix rozThe prefix rozis considered to be a prototypical means for expressing morphological causativity in Czech. It is also one of the most productive and most polyfunctional prefixes in the Czech language, which complicates its analysis. Some of its meanings (e.g., spatial expansion foukat – ‘to blow’/ rozfoukat – ‘to disperse something through the air’ or destruction mixovat – ‘to mix something’/ rozmixovat – ‘to liquidise something through mixing’, see Štícha et al. 2013, 260–261) are not related to causativity. However, the meaning of ingressiveness (see Kratochvílová – Jindrová et al. this volume) is different: causativity and ingressiveness are closely related since, depending on the context, the same verb can bear both meanings: ● roz + intransitive reflexive verb → meaning ‘sudden beginning of an action’: smát se (‘to laugh-refl’) – rozesmát se (‘to start laughing’) ● roz + transitive non-reflexive verb → causative meaning – rozesmát někoho ‘to make sb laugh’ (cf. Section 4.3). As for the meaning ‘gradual beginning of an action’ (see Štícha 2018, 1060; Pane vová – Karlík 2017), non-reflexive causative meaning is usually impossible (rozhořet se – ‘to start burning’, but *rozhořet něco – *‘to put sth on fire’; exceptions are rare, rozpovídat někoho – ‘to make sb talk’, roztančit někoho – ‘to make sb dance’). 29 According to the above-mentioned Čermák’s concept (Čermák 2001, 253 and 257), these constructions are factitives, not causatives. 54 Examples of the causative use of verbs with roz-: smát se (‘to laugh’) > rozesmát (‘to make sb laugh’) plakat (‘to cry’) > rozplakat (‘to make sb cry’) (11) Marie se smála jako šílená. Mary refl laugh-pst.3sg.ncaus like crazy-f ‘Mary laughed like crazy.’ (12) Ten vtip Marii rozesmál. that joke-nom Mary-acc laugh-pst.3sg.caus ‘The joke made Mary laugh.’ 4.3.2 semantic causativity Causativity might also be present in the very semantics of a verb, here we can distinguish two categories: 4.3.2.1 suppletive types This group is mainly comprised of causative verbs that do not display any formal vicinity to their non-causative counterpart: spadnout (‘to fall down’) > shodit (‘to make something fall down’). Our data shows that this group is extremely heterogeneous, both for the contents and the form. Formally, verbs with prefixes prevail. However, with these verbs it is almost impossible to assign to the prefix a (purely) causative meaning, i.e. causativity is thus expressed by the verb as a whole: přijít (‘to come’) > povolat (‘to make somebody come’). In some cases, the base verb itself is causative, so the prefix can also express other meanings besides the causative (temporal, aspectual and others): kroutit ‘to twist-ipfv’ – zakroutit ‘to twist-pfv’ točit ‘to spin-ipfv’– otočit, pootočit, zatočit ‘to turn-pfv’. 4. causative constructions in romance and their czech respondents 55 4.3.2.2 causative interpretation resulting from syntax This group is formed by verbs that can be interpreted either as causative or non-causative. This change is not reflected in any way in their form and results only from syntax. An example could be the verb zblbnout (‘to become a fool’, ‘to make someone act like a fool’): (13) Pavel zblbnul. Paul-nom become.a.fool-pst.3sg ‘Paul became a fool.’ (14) Pavel zblbnul Marii. Paul-nom become.a.fool-pst.3sg Mary-acc ‘Paul made Mary act like a fool.’ 4.3.3 analytic causativity Analytic causativity, which is formally closest to the Romance type, is also present in the Czech system, its main manifestations being causative verbs followed by a subordinate clause or a nominal syntagma and semi-causative verbs followed by an infinitive. 4.3.3.1 causative verbs followed by a subordinate clause Partially synonymous verbs that could be translated as ‘to cause something’ (způsobit, zapříčinit, vyvolat…) can be combined with a subordinate clause: (15) Neobvykle teplé počasí způsobilo, že tál led. unusually warm-n weather-nom cause-pst.3sg that melt-pst.3sg ice-nom ‘The unusually warm weather caused the ice to melt.’ (literally: ‘The unusually warm weather caused that the ice melted.’)30 4.3.3.2 causative verbs followed by a nominal syntagma Partially synonymous verbs that could be translated as ‘to cause something’ (způsobit, zapříčinit, vyvolat…) can also be combined with a nominal syntagma: 30 For an extended discussion, see Macháčková (1982, 120). 56 (16) Neobvykle teplé počasí způsobilo tání ledu. unusually warm-n weather-nom cause-pst.3sg melting-acc ice-gen ‘The unusually warm weather caused the ice to melt.’ (literally: ‘The unusually warm weather caused the melting of the ice.’)31 4.3.3.3 (semi-)causative verbs followed by an infinitive There are several verbs in Czech that can be considered semi-causative and enable completion by an infinitive (i.e. the type that is formally closest to the Romance construction). The most frequent are nechat (prevailing meaning being ‘to leave’; the causative interpretation partially responds to the English ‘to let’; cf. Section 4.2.1) and dát (prevailing meaning being ‘to give’; the causative interpretation translates as ‘to have something done’). For an extended discussion regarding this type and the semantic properties of the verbs nechat and dát, see Perissutti (2010; 2017) and Toops (1992; 2013). (17) Dal si změřit tlak. dát-pst.3sg refl.dat measure-inf blood.pressure-acc ‘He had his blood pressure measured.’ (18) Nechal si změřit tlak. nechat-pst.3sg refl.dat measure-inf blood.pressure-acc ‘He had his blood pressure measured.’ (literally: ‘He let someone measure his blood pressure.’) (19) Dal mu vypít lahvičku jedu. dát-pst.3sg he-dat drink-inf bottle-acc poison-gen ‘He made him drink a bottle of poison.’ The choice of the verb influences the interpretation of the causative process. The causativity here can be conceived as a wide-scale with two poles: one of them being to leave, not to prevent from doing sth; the other expressing the notions of forcing someone to do something. The notion of forcing is also the main semantic characteristics of another Czech semi-causative verb přimět (‘to move sb to sth’, ‘to force’): 31 For an extended discussion, see Macháčková (1982, 120). 4. causative constructions in romance and their czech respondents 57 (20) Přiměl ho vypít lahvičku jedu. force-pst.3sg he-acc drink-inf bottle-acc poison-gen ‘He forced him to drink a bottle of poison.’ 4.4 our typology of czech respondents In Sections 4.3.1, 4.3.2 and 4.3.3, we presented all the types of Czech causative constructions that are generally considered in Czech grammars. For the data analysis, we decided to use a typology based on these types. Nevertheless, the corpus material has revealed causative expressions that did not fit either of the above-presented categories. Therefore, we adapted and enriched the original typology so as to cover as many expressions of causativity as possible. The typology we used for the data analysis can be resumed as follows: Fusional types Type 1: rozplakat type (cf. Section 4.3.1.3) This type corresponds to the causativity expressed via the prefix roz-. Syntactic roles are maintained: ‘X made Y cry’ = ‘X-nom rozplakat-pst Y-acc’. Type 2: posadit type (cf. Section 4.3.1.1) This type comprises morphologic causativity expressed by the root change -e- > -iand, eventually, also by a prefix. Syntactic roles are maintained: ‘X made Y sit down’ = ‘X-nom posadit-pst Y-acc’. Type 3: shodit type (cf. Section 4.3.2.1) All semantic causatives (in the broadest sense of this term), where no formal relationship between a non-causative verb and its causative counterpart can be found will be analysed as a shodit type. Syntactic roles are maintained: ‘X made Y fall down’ = ‘X-nom shodit-pst Y-acc’. Analytic types Type 4: dát vypít type (cf. Section 4.3.3.3) Analytic causatives that correspond to the pattern of semi-causative verb + infinitive (i.e. dát, nechat, přimět + infinitive) are analysed as a dát vypít type. Syntactic roles are maintained: ‘X made Y drink something’ = ‘X-nom dát vypít-pst Y-dat something-acc’. Type 5: dohnat k slzám type This kind of analytic causative was not mentioned in the general typology of Czech causativity, its inclusion into this set is based on the results of the corpus analyses. It is defined as an idiosyncratic union that is close to idioms. The notion of causativity results from the construction as a whole and the combinatorics is strongly limited with this being the main difference between this type and type 6. This group 64 as has been explained above – we worked with the whole corpus, also including into our analysis the infinitives with one incidence. Given the differences in the size of the different subcorpora we worked with, we will present a more detailed commentary only with regard to Spanish, which offers the largest dataset. When analysing the Czech respondents with respect to the infinitive completion of the Spanish construction, we can postulate a scale ranging from the verbs that are generally translated by one single type to those that have very heterogeneous typological respondents: ● Matar ‘to kill’ (96% of respondents corresponded to type 4); constar ‘to state’ (95% of respondents corresponded to type 3 and the rest to type 10), brillar ‘to shine’ (90% of respondents corresponded to type 8) ● Hablar ‘to speak’ (6 typological respondents, the most frequent ones appeared in only 23% of cases), decir ‘to tell’ (the most frequent respondents appeared in only 29% of cases). There are other verbs between these two poles. The data shows that the most frequent typological respondent appears in 60% with 26 verbs, in 50% with 10 verbs, in 40% with 15 verbs, in 30% with 12 verbs and in 20% with 2 verbs. The most frequent number of typological respondents is 5 (for example, caer ‘to fall’, cambiar ‘to change’), 4 (e.g., dormir ‘to sleep’) and 3 (e.g., abrir ‘to open’). The verb displaying the largest scale of typological respondents (7) in our material is the verb girar ‘to turn’. This short commentary on the Spanish data enables us to draw several partial conclusions: 1) with most verbs, Czech typological respondents are significantly heterogeneous; 2) there are just a few verbs with one unanimous respondent, i.e. the tendency to lexicalisation is not very frequent. 4.7.1 primary czech respondents The three most frequent Czech respondents of the Romance causative construction, which altogether constituted approximately 70% of all respondents, proved to be types 3, 8 and 4. We will analyse each one of them in greater detail in the following sections. 4.7.1.1 type 3 – shodit type (hacer caer / far cadere / faire tomber / fazer cair) In approximately 40% of cases, the Romance construction is translated by an inherently causative verb that does not possess any structurally related non-causative counterpart. The dominance of this type for expressing causativity in Czech is also 4. causative constructions in romance and their czech respondents 65 proven by the fact that it was used for the translation of a large set of infinitives in all four languages (see Table 4.3). It is also worth observing that among the Czech respondents, there were many verbs which are usually not analysed as causatives in Czech grammars. This is a strong reason to consider the Czech causativity in the widest possible sense and to approach it as a highly abstract category, which is connected to the semantics of concrete verbs and to the Aktionsart in the narrowest sense of the meaning, see (26), (27), (28), (29). For an extended discussion regarding Aktionsart in Czech and Romance, see Kratochvílová – Jindrová (this volume). (26) Es. Me pareció distinguir la cabellera plateada del Maestro en el Segundo palco de mi lado, pero en ese instante mismo desapareció como si lo hubieran hecho caer de rodillas. → Měl jsem dojem, že v druhé lóži od sebe vidím Mistrovu stříbrnou hřívu, ale vzápětí zmizela, jako by ho srazili na kolena. Literally: as if they had brought him to his knees. Julio Cortázar, Konec hry (Final del juego), transl. Mariana Housková, Brno: Julius Zirkus, 2002. (27) It. Ancora anni dopo, a Bad Hollen, raccontavano che era stato come se qualcuno, dal campanile, avesse fatto cadere un pianoforte dritto su un deposito di lampadari di cristallo. → Ještě léta potom v Bad Hollenu vyprávěli, že to bylo jako kdyby někdo shodil ze zvonice klavír přímo na skladiště křišťálových lustrů. Literally: as if someone had thrown a piano from the belfry. Alessandro Baricco, Oceán moře (Oceano mare), transl. Alice Flemrová, Prague: Eminent, 2001. (28) Fr. Il s’agissait juste de faire couler un peu d’encre pour rappeler que d’autres avaient fait couler un peu de sang. → Vždyť šlo pouze o to vyplýtvat trochu inkoust, aby se druhým připomnělo, že oni vyplýtvali trochu krve. Literally: that had wasted some blood. Pierre Assouline, Zákaznice (La Cliente), transl. Lubomír Martínek, Prague: Prostor, 2000. (29) Pt. Como se tudo fosse muito simples e a memória o fizesse perder tempo. → Jako by všechno bylo velice prosté a paměť ho jenom zdržovala. Literally: and memory only slowed him down. Hélia Correia, Ďáblova hora (Montedemo), transl. Vlasta Dufková, Prague: Odeon, 1986. 66 The aim of this study is not to analyse in detail the factors influencing the choice of Czech respondents (this being a desideratum for future studies). Nevertheless, it can be indicated that these factors are numerous and two of these appear to be particularly important: the meaning of the fully semantic verb and the presence/absence of a verb complement (and its meaning). Both these factors are closely related as the following example shows. Our data shows that with the Spanish construction hacer caer ‘make fall’, one-seventh of all appearances correspond to the locution hacer caer en la cuenta, where caer en la cuenta means ‘to realise’. In these cases, Czech respondents are the verbs vysvětlit, upozornit, připomenout, oznámit, all of which have the approximate meaning of ‘to make realise’, ‘to announce’, see (30): (30) Es. El sirio Moisés le hizo caer en la cuenta de una novedad: llegaba un circo. → Syřan Moisés mu však oznámil novinku: přijel cirkus. Literally: he told him the news. Gabriel García Márquez, Zlá hodina (La mala hora), transl. Vladimír Medek, Prague: Odeon, 2006. In other cases – where no complement is present – the Czech respondents of hacer caer are, for example, srazit, porazit, spustit, shodit, povalit, upustit, all of which have the approximate meaning ‘to make fall’: (31) Es. —Venga el que ha hecho caer ese fusil —gritó. „Kdo upustil pušku, ať vystoupí,“ zvolal. Literally: who dropped the riffle. Mario Vargas Llosa, Město a psi (La ciudad y los perros), transl. Miloš Veselý, Prague: Mladá fronta, 2004. Therefore, the possible combination of a verb with a complement influences the selection of the Czech respondent. A similar situation can be observed with the vast majority of verbs. The choice of a specific verb is also influenced by other factors (style, language variant etc.). 4.7.1.2 type 8 – what makes you think that > proč myslíte? (‘why do you think that?’) The second most frequent respondent type is characterised by the change of syntactic roles. This type also appeared with the widest range of verbs (see Table 4.3). Its most frequent realisations can be summarised as follows: 4. causative constructions in romance and their czech respondents 67 1) The Czech respondent contains two main clauses with causation generally being expressed by conjunctions such as a tak (‘and so’) or a proto (‘and because of that’): (32) Fr. Elle ne parlait pas, ne semblait pas avoir le sens commun et ne cessait de sourire, ce qui la faisait prendre communément pour une arriérée. → Nemluvila avypadala trochu tak, jako by ani neměla všech pět pohromadě, jen se usmívala, asnad proto ji sousedé považovali za zaostalou. Literally: and maybe because of that, the neighbours considered her retarded. Frédéric Tristan, Hrdinné útrapy Baltazara Kobera (Les Tribulations héroïques de Balthasar Kober), transl. Oldřich Kalfiřt, Prague: DharmaGaia – Dauphin, 2003. 2) The Czech respondent contains a subordinated clause, usually with a temporal or consequential meaning: (33) Es. —¡Basta! —interrumpió el capitán dando un puñetazo sobre la mesa que hizo bailar los platos y las copas. → „Tak dost!“ přerušil je kapitán ranou pěstí do stolu, až se talíře a sklenky roztančily. Literally: so that plates and glasses started to dance. Isabel Allende, Dcera štěstěny (La hija de la fortuna), transl. Monika Baďurová, Prague: BB art, 2004. 3) In the Czech respondent the causer is expressed in different ways, usually by an adverbial (34) or by the modal verb muset (‘have to’) (35): (34) Es. Es el género de calamidades que un día te harán caer los brazos con desaliento o gritar con indignación. → To je přesně typ pohromy, kvůli níž ti malomyslně klesnou ruce a nejradši bys křičel vzteky. Literally: because of which your arms will decline helplessly. Ernesto Sábato, Abbadón zhoubce (Abbadón el exterminador), transl. Anežka Charvátová, Brno: Host, 2002. (35) Pt. Não queria ser visitada assim, tão acabada, fizera a amiga jurar nada dizer a ninguém. → Nepřála si, aby ji někdo navštěvoval tak zničenou, přítelkyně musila přísahat, že nikomu nic neřekne. Literally: the friend had to swear. Jorge Amado, Pastýři noci (Os Pastores da Noite), transl. Pavla Lidmilová, Prague: Odeon, 1983. 68 As can be observed, Czech respondents maintain causativity to different extents and while the causer is also expressed by different means, in some cases causativity is not overtly expressed at all. 4.7.1.3 type 4 – dát vypít type The most frequent analytic expression of causativity was represented by type 4, i.e. a combination of a semi-causative verb with the infinitive. Although this respondent type formally corresponds to the original Romance construction, we must bear in mind that this correspondence is only partial. The Czech verb udělat, which would be the best candidate for the translation of hacer/fare/faire/fazer cannot appear in this kind of construction and all Czech verbs that allow completion with an infinitive, thus forming acausative construction, are only semi-causatives, which also have other non-causative interpretations. The corpus analysis reveals that beside the traditionally mentioned semi-causative verbs such as dát, nechat, přimět (see Section 4.3.3.3) and donutit (‘to force’), there are also others that can gain acausative interpretation when followed by an infinitive form, such as: pomoci (‘to help’), poručit (‘to order’), přikázat (‘to command’), mínit (‘to have in mind’), dovolit (‘to allow’), dokázat (‘to manage’), umožnit (‘to enable’), poslat (‘to send sb somewhere’), naučit (‘to teach’). Given the frequency the Romance construction was translated by asemi-causative Czech respondent, it is possible to state that the contrastive analysis reveals awide range of semantic notions associated with hacer/fare/faire/fazer + infinitive that can be considered only partially causative and that this construction clearly displays secondary meanings that cannot be attributed solely to the category of causativity. (36) It. Il Comandante, che si faceva chiamare re di Roma, era un mangione e beone coatto; e l’alcool serviva come eccitante e narcotico usuale agli occupanti sia nel quartier generale come alla base. → Velitel, který si nechal říkat římský král, byl velký jedlík apijan, alkohol sloužil okupantům jako nejobvyklejší dráždidlo adroga jak vhlavním štábu, tak dole vzákladních složkách armády. Literally: who had others call him Roman king. Elsa Morante, Příběh vhistorii (La storia), transl. Zdeněk Frýbort, Prague: Odeon, 1990. (37) Fr. Quant à Bruno, le jugement se poursuivrait jusqu’ à épuisement du dossier – ce qui signifiait, que le Saint-Office tenterait de faire avouer au malheureux tous les secrets qu’il avait juré sur son âme de ne jamais révéler. → Bruna zamýšleli vyšetřovat tak dlouho, dokud soud nedospěje kúplnému vyčerpání spisu – jinými slovy dokud svatá inkvizice nebožáka nedonutí přiznat veškerá tajemství, včetně těch, onichž by přísahal při své nesmrtelné duši, že je nikdy neodhalí. 4. causative constructions in romance and their czech respondents 69 Literally: force to confess. Frédéric Tristan, Hrdinné útrapy Baltazara Kobera (Les Tribulations héroïques de Balthasar Kober), transl. Oldřich Kalfiřt, Prague: DharmaGaia – Dauphin, 2003. (38) Pt. É novo… Parece que oéter desenvolve, faz aflorar aalma de as frutas… → To je úplná novinka… Éter prý pomáhá odhalit duši ovoce… Literally: helps to reveal the soul of fruits. José Maria Eça de Queirós, Kráčej ačti (ACidade e as Serras), transl. Marie Havlíková, Prague: Academia, 2001. 4.7.2 secondary czech respondents Types 1–2, 5–7 and 9–10 constituted in total less than one-third of all Czech respondents of the Romance causative construction. The most frequent representatives of this group proved to be rather heterogeneous respondents subsumed under type 5 (dohnat kslzám type, i. e. idiomatic expressions of causativity) and respondents marked as type 9, where causativity is expressed by means others than those described in types 1–8. This can be seen as yet further proof that Czech causativity is ahighly complex category, which cannot be identified by one or two dominant forms of expression. 4.7.2.1 type 5 – dohnat kslzám type The above-mentioned inherent complexity of Czech causativity and the heterogeneity of its formal manifestations can be observed even within the very category of the idiomatic expressions of causativity. As described in Section 4.4, these constructions are lexical idiosyncratic units bordering with phraseological units and collocations (combinations of verbs and complements are not free; there are important restrictions). (39) Es. Vaciló, sin saber qué hacer, hasta que las señas insistentes de Borobá le hicieron dudar de que su amiga se encontrara allí. → Zaváhal, nevěda co dělat, až mu Borobiny naléhavé posunky vnukly pochybnost, jestli tam jeho přítelkyně vůbec je. Literally: Borobá’sinsisting gestures inspired doubts in him. Isabel Allende, Království Zlatého draka (El Reino del Dragón de Oro), transl. Monika Baďurová, Prague: BB-art, 2004. (40) It. Se c’è una cosa capace di farti impazzire è l’elastico delle calze troppo stretto. → Pokud existuje něco, co tě může dohnat kšílenství, je to příliš úzká gumička uponožek. 70 Literally: drive you crazy. Alessandro Baricco, City (City), transl. Alice Flemrová, Prague: Volvox Globator, 2000. (41) Fr. Je suis sûr que ça la fait jouir, de mariner un obèse sans défense, nu et imberbe. → Určitě jí to dělá potěšení máčet bezbranného obézního chlapa, nahého abez jediného chlupu na těle. Literally: it certainly gives her pleasure. Amélie Nothomb, Vrahova hygiena (Hygiène de l’assassin), transl. Jarmila Fialová, Prague – Liberec: Paseka, 2001. (42) Pt. Mas oque atorturava, afazia chorar todos os dias era aidade de ele ser um enjeitadinho. → Denně ji však pohnula kpláči trýznivá myšlenka, že dítě bude sebranec. Literally: every day, the tormenting thought that the child would be abastard moved her to tears. José Maria Eça de Queiroz, Zločin pátera Amara (OCrime do Padre Amaro), transl. Zdeněk Hampl, Prague: Státní nakladatelství krásné literatury aumění (SNKLU), 1961. The corpus analyses reveal that respondents forming this category can differ even when analysing the respondents of constructions with the same infinitive completion, see (42) and (43). (43) Es. Cada animalito que se echaba ala boca venciendo la repugnancia, lo hacía sonreír pensando en su maestro, aquien tampoco le gustaban los cangrejos. → Každé to zvířátko, které spřemáhaným odporem pojídal, uněj vzbuzovalo úsměv, když pomyslil na svého učitele, který raky také neměl rád. Literally: arose his smile. Isabel Allende, Dcera štěstěny (Hija de la fortuna), transl. Monika Baďurová, Prague: BB Art, 2004. (44) It. (…) e subito ricostruivate la grande catena dell’ essere, in love and joy, perché tutto quel che nell’ universo si squaderna nella vostra mente si era già riunito in un volume, e Proust vi avrebbe fatto sorridere. → (…) hned jste se pustily do velikánského řetězce bytí podle vzoru love and joy, protože všechno, co ve vesmíru jsou jen pouhé listy, ve vaší mysli je už svázáno do jednoho svazku, aProust by vám pouze vyloudil úsměv na rtech. Literally: wheedled asmile on the lips. Umberto Eco, Foucaultovo kyvadlo (Il pendolo di Focault), transl. Zdeněk Frýbort, Prague: Český klub, 2001. 4. causative constructions in romance and their czech respondents 71 4.7.2.2 type 9 – other translation Analyses of the respondents attributed to this type revealed several different possibilities of expressing notions included in the hacer/fare/faire/fazer + infinitive construction. The most common type can be defined in terms of using adeverbative noun/adjective to substitute the Romance infinitive, the general pattern is: ‘wind makes leaves fall’ > ‘leaves fallen by the wind’, see (44): (45) Pt. (…) uma linda capela branca que as freiras fizeram construir em a frente de ocolégio e que dominava acidade desde omorro (…). → Tato deska byla zasazena vpěkné bílé kapli zbudované jeptiškami naproti ústavní škole ashlížející na město se svého pahorku (…). Literally: built by nuns. Jorge Amado, Země zlatých plodů (São Jorge dos Ilhéus), transl. Jaroslav Rosendorfský, Prague: Československý spisovatel, 1950. Other respondents attributed to this category could be defined by the missing respondent of the verb hacer/fare/faire/fazer: (46) It. Soprattutto però era una città istruita: è sempre il Villani afarci sapere che gli 8–10.000 bambini fiorentini sapevano tutti leggere e scrivere (...). → Především však byla městem vzdělaným: podle Villaniho umělo 8–10 tisíc florentských dětí číst apsát (...). Literally: according to Villani. Giuliano Proccaci, Dějiny Itálie (Storia degli italiani), transl. Bohumír Klípa – Drahoslava Janderová – Kateřina Vinšová, Prague: Nakladatelství Lidové noviny, 2007. 4.7.2.3 type 7 – způsobit, že tál type The most analytic type of expressing causativity (i.e. averb corresponding to ‘to cause something’ followed by asubordinate clause) was the sixth or seventh most frequent type (depending on the concrete Romance language). Given the wide range of its possible completions (see Table 4.3), this type proved to be productive, but also arather marginal means of expressing causativity in Czech. Nevertheless, the scale of Czech verbs that correspond to the Romance verb hacer/fare/faire/fazer is very wide and underlines notions other than the pure causativity that can be attributed to the Romance construction. Apart from the verbs způsobit (‘to cause-pfv’) and způsobovat (‘to cause-ipfv’), this respondent type included verbs expressing: 72 1) The subject’sconsent or permission: svolit (‘to consent’): (47) Es. Lepprince le hizo pasar y me rogó que me quedase. → Lepprince svolil, ať vstoupí, apoprosil mě, abych zůstal. Literally: Lepprince permitted that he entered. Eduardo Mendoza, Pravda opřípadu Savolta (La verdad sobre el caso Savolta), transl. Petr Koutný, Prague: Odeon, 1983. 2) The subject’swill: přikázat (‘to order’), požadovat (‘to demand’), nařídit (‘to command’), chtít (‘to want’), říct (‘to tell’), přimět (‘to force’): (48) It. Il giovane ufficiale con secchi ordini li fece portare via. → Mladý důstojník stroze přikázal, aby je odnesli. Literally: commanded that they were taken away. Italo Calvino, Naši předkové (Inostri antenati), transl. Zdeněk Digrin – Vladimír Mikeš, Prague: Odeon, 1970. 3) The subject’seffort to create or finish something: zařídit (‘to make arrangements’), postarat se (‘to make sure’), dát si práci (‘to make an effort’), dělat (‘to make’), přispět k(‘to contribute to’): (49) Fr. Maître Flinker, reprit-il, je vais vous faire prêter deux apprentis par quelque autre imprimeur de la ville (…). → Mistře Flingere, zařídím, aby vám někdo ztiskařů ve městě půjčil dva učně (…). Literally: Iwill make arrangements so that someone lent to you. Frédéric Tristan, Hrdinné útrapy Baltazara Kobera (Les Tribulations héroïques de Balthasar Kober), transl. Oldřich Kalfiřt, Prague: DharmaGaia – Dauphin, 2003. 4) The logical or unintentional consequence of aprevious state or action: vést k(‘to lead to’), dohánět k(‘to drive sb to smth’): (50) Pt. E sentira a, porventura, essa felicidade, que dão os amores ilegítimos, de que tanto se fala nos romances e em as óperas; que faz esquecer tudo em avida, afrontar amorte, quase fazê-la amar? → Aprožívala snad takové štěstí, které člověk cítí znedovolených milostných vztahů, onichž se tolik mluví vrománech aoperách akteré člověka dohánějí ktomu, aby zapomněl na život, asmrti nejen čelil, nýbrž po ní itéměř toužil? Literally: that drive aperson to forget. 4. causative constructions in romance and their czech respondents 73 Eça de Queirós, Bratranec Basilio (OPrimo Basílio), transl. Zdeněk Hampl, Prague: Odeon, 1989. 4.7.2.4 type 1 – rozplakat type Despite the fact that the Czech prefix rozis often mentioned as asignificant means of expressing causativity, it did not appear very frequently in our data. While this could be attributed to the fact that type 1 is defined in amuch more precise way than the other types (i.e. it is defined by one single prefix; therefore, it does not admit formal variability as the other types do), we claim that the main reason for its low frequency lies in its strong combinatory limitations (see Section. 4.3.1.3). The rozprefix proved to be aproductive causative prefix, that, nevertheless, combined mostly with verbs expressing an emotional reaction such as brečet (‘to cry’), smát se (‘to laugh’), slzet (‘to shed tears’), lkát (‘to wail’); movement such as tančit (‘to dance’), chvět se (‘to tremble’), točit (‘to spin around’), houpat (‘to swing’), třást (‘to shake’), kmitat (‘to wiggle’); or sound such as zvučet (‘to make sounds’), znít (‘to resonate’). 4.7.2.5 type 2 – posadit type and type 6 – způsobit tání type These types appeared the least frequently in our data, which is caused by their narrow definition: they are defined in terms of arather infrequent derivational change (type 2), or in terms of the presence of one particular verbal form, i.e., verbal noun (type 6). Thus, type 2 includes avery limited group of verbs. The low frequency of type 6 can also be influenced by the corpus we used (noun phrases composed of averbal noun are typical for specialist or administrative language rather than for narrative). 4.7.2.6 type 10 – no translation As it follows from the definition, in this type, the causative meaning completely disappeared in the Czech respondent, which gave us no reason to address it in our study. 4.8 conclusions When analysing the Czech respondents of the Romance causative construction, we can observe that it is difficult to precisely determine the proportion of analytic and fusional resources. With reference to Czech causativity, it is better to mention acontinuum. If we conceive the analysed Romance construction as an expression of “pure” causativity, we can also present the respondents as atypological means of express- 80 5.0 introduction The goal of this chapter is to present the most frequent Romance verbal periphrases that express the beginning of aprocess and their Czech respondents, thus aiming to adopt anew perspective on the abstract notion of ingressivity and presenting its possible realisations in the Czech language. The chapter is organised as follows. In Section 5.1, we present some general remarks regarding verbal periphrases in Romance languages. Since our contrastive perspective points out the necessity to analyse the semantic features they express with respect to aspect and Aktionsart, Section 5.2 introduces how these categories are represented in Romance languages and in Czech. Further on, we concentrate on ingressivity and its relationship with these categories. In Section 5.3, we present the results of aseries of corpus analyses that concentrated on Romance ingressive periphrastic constructions and their Czech respondents. Section 5.4 highlights the most important conclusions that can be based on these analyses. 5.1 verbal periphrases in romance Verbal periphrases constitute an important element of the verbal system of all four of the Romance languages studied. In general terms, their main function consists of focusing on aspecific stage of aprocess, presenting it from the perspective of its beginning, its progress or its end.36 From this point of view, these constructions are closely related to aspect and Aktionsart. In the Czech tradition, the notions they express are rather treated within amore specific category called způsob/povaha slovesného děje, which could be translated as the character of verbal action (see Dušková 2012). These categories and their respective position in the Romance and Czech verbal systems will be discussed in Section 5.2. 36 The combinations of modal verbs and the infinitive and even the construction hacer/fare/faire/fazer + infinitive are sometimes also treated as verbal periphrases, see Drzazgowska (2011). However, we will consider as proper verbal periphrases only those constructions that are directly related to the phases of aprocess. For an extended discussion related to hacer/fare/faire/fazer + infinitive, see Čermák – Kratochvílová et al. (this volume). 5. ingressive periphrases in romance and their czech respondents 81 Typically, Romance verbal periphrases display the following structure: semi-auxiliary verb (+ preposition) + non-finite verbal form.37 The resulting constructions can express several tempo-aspectual meanings that can be attributed to one of these six basic types: 1) Imminent future – It is about to rain Es. Está para llover / It. Sta per piovere / Pt. Está para chover 2) Beginning of aprocess – It starts to rain Es. Empieza allover / It. Comincia apiovere / Fr. Il commence à pleuvoir / Pt. Começa achover 3) Process in progress – It is raining Es. Está lloviendo / It. Sta piovendo / Fr. Il est en train de pleuvoir / Pt. Está achover38 4) The end of aprocess – It stopped raining Es. Ha dejado de llover / It. Ha smesso di piovere / Fr. Il acessé de pleuvoir / Pt. Deixou de chover 5) The repetition of aprocess – It started to rain again Es. Ha vuelto allover / It. È tornato apiovere / Pt. Tornou achover 6) Habituality – It usually rains here alot Es. Aquí suele llover mucho / It. Qui suole piovere molto / Pt. Aqui costuma chover muito 5.1.1 approaches to verbal periphrases and the goal of our study Verbal periphrases have been largely discussed in linguistic literature. There are three main approaches which the different authors adopt: 1) The very definition of periphrastic constructions, their distinction from idioms, the combinatory possibilities, the semantics of periphrastic constructions and their inventory. See Drzazgowska (2011), Fente Gómez – Fernández Álvarez – Feijó (1994), Fogsgaard (2002), García Fernández (2006), Gómez Torrego (1988; 1999), Jindrová (2016), Olbertz (1998), Verroens (2011). 2) The position of verbal periphrases within the verbal system, their relationship to aspect and Aktionsart. See Barroso (1994; 1999; 2000), Begioni (2012), Camus Bergareche (2004), Dietrich (1983), Gosseline (2011), Laca (2002), Pešková (2005). 3) Contrastive analyses of verbal periphrases across different Romance languages or the expression of notions expressed by Romance periphrastic constructions in 37 The commonly used French progressive construction corresponding to the English be + -ing form displays amore complex structure constituted by the verb être (‘to be’) followed by en train de (‘in the process of’) + infinitive. In Romanian, acompletion by the subjunctive is also possible, for verbal periphrases in Romanian and their comparison to Spanish, see Topor – Fernández – Vásquez (2006; 2007; 2008). 38 The construction with infinitive is typical for European Portuguese; in the Brazilian variant, the construction Está chovendo with gerund is used. 82 typologically different languages. See Kratochvílová – Jindrová (2017), Głowicka (2013), Luque (2008), Pešková (2018), Sánchez Montero (1993), Topor – Fernández – Vázquez (2006; 2007; 2008), Zieliński (2017). Our approach to the study of Romance verbal periphrases aims to complement the discussion represented by works cited as types b) and c). Leaving aside problems regarding the formal aspects of verbal periphrases and following the general aim of this monograph, we adopt apurely synchronic and contrastive perspective. Our goal is to analyse the most frequently used ingressive Romance periphrases and their Czech translations. Since the Czech language does not possess any clear structural respondent for Romance ingressive constructions, we opt for analysing them within the general categories of aspect, Aktionsart and manner of action, thus defining the relationship between ingressivity and these categories. 5.2 aspect and aktionsart While there is no general agreement regarding the linguistic category that verbal periphrases express, it is commonly assumed that the notions expressed by these constructions are related to the category of aspect and/or Aktionsart. Therefore, acontrastive analysis must bear in mind the inherent differences in the way these categories are expressed in the Romance languages and in Czech. Taking the traditional Comrie’s(1976) approach to aspect as astarting point, we can postulate several key structural differences. 5.2.1 aspect and aktionsart in romance languages When cross-linguistically analysing the category of aspect, Comrie (1976) distinguishes the basic opposition perfective/imperfective (with respective subclasses, such as habitual or progressive). In the studied Romance languages, this opposition is systematically encoded solely in the morphology of past tenses, which differentiate between the processes conceived as awhole (Es. hablé / It. ho parlato (parlai) / Fr. j’ai parlé (je parlai) / Pt. falei)39 and those representing acertain circumstance and conceived as lasting for alonger time with no attention to the beginning or end (Es. hablaba / It. parlavo / Fr. je parlais / Pt. falava). While the category of aspect, in the narrower sense of the term, is expressed morphologically, the category of Aktionsart (inherent/internal/lexical aspect, manner of ac39 Spanish (and, to alesser extent, also Portuguese) also maintain the distinction between the perfective (Es. hablé / Pt. falei) and the perfect (Es. he hablado / Pt. tenho falado); in Italian and French the original perfect (compound) forms tend to substitute the perfective ones in the spoken language, thus neutralising this opposition. 5. ingressive periphrases in romance and their czech respondents 83 tion) has been traditionally conceived in terms of tempo-qualitative meanings resulting from the semantics of averb. This category is thus related to notions such as telic/ atelic, punctual/durative, state/process (see Comrie 1976; Smith 1997; Albertuz 1995). However, in Romance linguistics, Aktionsart is usually understood in abroader sense and is also related to the phrasal characteristics of aprocess, thus including the inchoativity, progressivity (durativity), intromission, termination or iterativity (see Pawlak 2008; Begioni 2012). Given the close relationship between aspectual and Aktionsart-related notions and the limited representation of the category of aspect in Romance languages (in comparison to Slavic languages, see Section 5.2.2), some authors argue against separating these two categories and point out their inherent interconnection (see Fernández Pérez 1993; Gosselin 2011). 5.2.2 aspect and aktionsart in czech The Czech category of aspect is largely comparable to the widely studied Russian one (see Comrie 1976, 125; Smith 1997, 227–259; Croft 2012, 110–114). The Czech verbal system presents a systemic opposition between perfectivity and imperfectivity in all tenses and moods. Perfectivity in Czech is generally expressed through prefixes: Psát (‘to be writing-ipfv’) Napsat (‘to write something-pfv’) Dopsat (‘to finish writing-pfv’) Přepsat (‘to rewrite something-pfv’) Zapsat (‘to take anote-pfv’) Rozepsat (‘to start writing something-pfv’) Připsat (‘to add something in written-pfv’) As can be observed, the Czech perfective prefixes are often polyfunctional and, together with perfectivity, also express other notions related to the tempo-qualitative characteristics of aprocess.40 These characteristics are studied within the category of manner of action (způsob slovesného děje), aterm corresponding to Aktionsart which is, however, used more often in Czech linguistics (see Komárek – Kořenský et al. 1986, 185–187; Karlík – Nekula – Rusínová 1995, 209-213; Karlík – Nekula – Pleskalová 2002, 567-569; Chromý 2018). Nübler (2017) distinguishes the following groups of amanner of action: ingressive, evolutive, delimitative, resultative, terminative, perdurative, finitive, exhaustive, total, saturative, extensive, cumulative, intensive, excessive, semelfactives, momentary, iterative, diminutive, comitative, frequentative, stative, decursive and mutative. 40 For an extended discussion regarding Czech affixes, see Štichauer et al. (this volume). 84 5.2.3 verbal periphrases and the relationship to aspect and aktionsart As can be observed, notions expressed by verbal periphrases, such as the beginning of aprocess, the duration of aprocess or its repetition, correspond to types of Aktionsart / manner of action, as conceived both by the Romance and Czech tradition. However, identifying periphrases entirely with this category is not possible since, when expressed through aperiphrastic construction, these notions are external and subjective, i.e. they do not result naturally from the semantics of averb. Analysing periphrases within the framework of aspect is equally problematic given the possibility to combine, for example, progressivity and perfectivity (Es. Estuvo lloviendo una hora – ‘It rained for an hour’) or termination and imperfectivity (It. Giorno dopo giorno si ripeteva lo stesso: smetteva di piovere e poi tutto cominciava di nuovo – ‘Every day, the same thing used to happen: it used to stop raining and then it all used to start again’). The problematic nature of verbal periphrases generally leads the authors to apprehend the category of aspect in amuch broader sense than it is customary in Czech linguistics and to analyse these constructions as forms of expression of this category, see Barroso (1988; 1994; 2000), Olbertz (1998), Fogsgaard (2002), Camus Bergareche (2004), RAE (2009). Given the inherent structural differences between aspect in Czech and in Romance, we opt for adifferent approach. Following Pawlak (2008) and Zavadil – Čermák (2010), we associate the notion of aspect solely with the perfective/ imperfective opposition. Our claim here is that ingressivity, durativity, iterativity etc. are more closely related to manners of action, extending, however, the definition of this category and conceiving its formal manifestations in terms of acontinuum that includes both internal notions resulting from the very semantics of averb, both the external and explicit expression of these notions. While adopting the traditional definition of the manner of action (henceforth referred to as MoA) in terms of acomplex and heterogeneous category related to the qualitative characteristics of aprocess that allows their division into groups (see Zavadil – Čermák 2010, 314; Nübler 2017), we choose to approach it in highly abstract terms, distinguishing three main levels of its expression: 1) Internal MoA The internal manner of action corresponds with the narrowest understanding of Aktionsart, i.e. with the inherent semantic properties of the processes allowing their classification. 2) Derivative MoA The expression of the tempo-qualitative characteristics through an affix and/or through reflectivity. 3) Analytical MoA The expression of the tempo-qualitative characteristics of aprocess through aproductive verbal or verbo-nominal construction. 5. ingressive periphrases in romance and their czech respondents 85 Further on, we shall concentrate only on one particular type of MoA and will analyse the expression of ingressivity. 5.2.4 ingressive moa Ingressivity (sometimes the terms inchoativity or inceptivity are used in the same sense, see, for example, Barroso 1999; Fogsgaard 2002; Luque 2015) can be defined as the presentation of aprocess from the perspective of its beginning or its initial stage (see Zavadil – Čermák 2010, 316). Comrie (1976, 19–20) analyses ingressivity within the framework of the perfective aspect. Further on, Zavadil – Čermák (2010, 316–319) distinguish four subtypes of ingressivity: 1) imminent ingressivity, i.e. imminently expected process (It’sabout to rain), 2) dispositional ingressivity, i.e. imminently planned process (Iam about to leave), 3) initial ingressivity, i.e. the proper beginning of aprocess (It starts to rain), 4) inceptive ingressivity, i.e. the initial process in asequence of actions (He started by saying his name). We focus solely on type 4), i.e. initial ingressivity, in this chapter. 5.2.4.1 initial ingressivity in romance languages 5.2.4.1.1 derivative ingressive moa in romance Verbal pairs corresponding to opposition ingressive (perfective) / progressive (imperfective) are rather infrequent in Romance languages (especially in comparison to Czech). Begioni (2010, 34) mentions the French prefix enand the Italian prefix ad-, which (when combined with areflexive pronoun) form pairs such as It. dormire, Fr. dormir (‘to sleep’) / It. addormentarsi, Fr. s’endormir (‘to fall asleep’). In Spanish, asimilar pair is created when using the reflexive pronoun: dormir (‘to sleep’) / dormirse (‘to fall asleep’). In Portuguese, acombination of aprefix and asuffix is used: dormir (‘to sleep’) / adormecer (‘to fall asleep’). Given the marginal status of this form of expressing ingressivity, these pairs will not be discussed in this monograph. 5.2.4.1.2 analytical ingressive moa in romance In Romance, the main tool for expressing ingressivity are verbal periphrases. All four Romance languages studied here dispose of aset of partially synonymous constructions that emphasise the beginning of aprocess. Their common semantic feature can be defined as [+ingressive]; however, they differ with regard to their frequency of use, 86 the type of infinitives they typically combine with and the (non-)presence of secondary semantic notions such as [+effort], [+unexpectancy], [+inappropriateness] etc. In general terms, ingressive periphrases in Spanish, Italian, French and Portuguese can be divided into three groups: 1) Ingressive constructions expressing the mere beginning of aprocess These periphrases correspond to the neutral English construction start + -ing. They do not add any specific secondary semantic notion to the representation of aprocess in its initial phase and do not display any considerable combinatory limitations. In all four languages, these periphrases contain asemi-auxiliar that corresponds to the English verbs to start or to begin: Es. Empezar a+ infinitive, comenzar a+ infinitive41 It. Cominciare a+ infinitive, iniziare a+ infinitive Fr. Commencer à + infinitive, commencer de + infinitive Pt. Começar a+ infinitive42 2) Ingressive constructions expressing the beginning of aprocess and the notion of effort on the part of the subject In all the languages studied, there is one commonly used ingressive construction that does not display any significant combinatory limitations although it modifies the basic ingressive notion by adding the semes of [+effort] [+subject’sinterest] [+motivation] [+intentionality], thus making it comparable to the English construction to get down to sth:43 Es. Ponerse a+ infinitive (see García Fernández et al. 2006, 218–223; Zavadil – Čermák 2010, 318; Kratochvílová – Jindrová 2017) It. Mettersi a+ infinitive (see Sánchez Montero 1993, 28; Luque 2008) Fr. Se mettre à + infinitive (see Haton 2005; Pauly 2005; Verroens 2011) Pt. Pôr-se a+ infinitive (see Barroso 2016; Kratochvílová – Jindrová 2017) 3) Highly stylistically marked ingressive constructions In all four languages studied, there is agroup of periphrastic constructions that can be described in terms of the lower overall frequency of use and aspecific semantic feature added to the notion of ingressivity. Sometimes, a clear tendency to prefer 41 In Spanish, there are two more constructions, principiar a+ infinitive and iniciar a+ infinitive, which would fit the description of this group. However, in present-day language, these periphrases are seldom used (see Kratochvílová – Jindrová 2017) and, due to their limited presence in the corpus we work with, will not be analysed here. 42 The Portuguese construction principiar a+ infinitive is similar to its Spanish counterpart (i.e. its use is very limited in present-day language) and it is not analysed here, see Kratochvílová – Jindrová 2017. 43 In general, the notion of [+suddenness] is also present although it is less distinct than with periphrases forming the third group. 5. ingressive periphrases in romance and their czech respondents 87 aconcrete type of infinitive completion can also be observed. Given the huge number of these constructions (that can especially be found in Spanish and in Portuguese) and the limited frequency of their usage, not all stylistically marked periphrastic construction could be studied in this monograph. For the corpus analysis, we chose the following seven constructions that do not display any considerable diatopic restrictions, had afrequency ≥ 15 in our respective subcorpora and are largely discussed in the literature. These constructions were further divided into three subgroups based on the semantic features associated with them. These notions are marked in square brackets; considerable combinatory limitations are specified in curly brackets. [+suddenness] [+unexpectancy] Es. Echarse a+ infinitive {+verbs of emotional reaction} (see Gómez Torrego 1999, 3374) Es. Echar a+ infinitive {+verbs of movement} (see Gómez Torrego 1999, 3347) [+suddenness] [+abruptness] [+previous retention] Es. Romper a+ infinitive {+verbs of emotional reaction} {+verbs of physical activity} {+ verbs of interpretation} {+verbs expressing achange of state} {+verbs associated with meteorological phenomena} (see García Fernández et al. 2006, 230–232) Es. Largarse a+ infinitive (see García Fernández et al. 2006, 183) It. Scoppiare a+ infinitive {+verbs of emotional reaction} (see Bertinetto 2001, 155; Sánchez Montero 1993, 32) Pt. Romper a+ infinitive (see Barroso 1994, 124–125; Kratochvílová – Jindrová 2017) [+suddenness] [+abruptness] [+vehemence] Es. Lanzarse a+ infinitive (see Zavadil – Čermák 2010, 319) As can be observed, the number of periphrases analysed in each language is not the same, we analyse eight Spanish ingressive constructions, four Italian and only three French and Portuguese ingressive periphrases. This discrepancy is given by two factors: 1) The overall number of ingressive periphrastic constructions and their productivity in present-day language differs distinctively according to the specific language. The widest set of constructions, which are also highly productive, can be found in Spanish and Portuguese (see Barroso 1994; 1999, 337; Zavadil – Čermák 2010; Drzazgowska 2011; Jindrová 2016; Kratochvílová – Jindrová 2017). Slightly more limited is the list of Italian verbal periphrases (see Sánchez Montero 1993; Hamplová 1994; Bertinetto 2001) while the most limited set of ingressive constructions can be found in French (see Haton 2005). 2) The lowest number of Portuguese ingressive constructions that could be analysed is given by the limited extension of the Portuguese subcorpus. 88 5.2.4.2 initial ingressivity in czech 5.2.4.2.1 derivative ingressive moa in czech There are several prefixes in the Czech language that can be attributed to an ingressive function. Nevertheless, these prefixes are always polyfunctional and ingressivity is not their unique function. Ingressivity can also inherently combine with other notions (such as avery short duration of aprocess or spatial interpretation in terms of moving somewhere) while the separation of these two notions proves to be impossible. In her extensive study of ingressive prefixes in Czech, Reichzieglová (2010) works with the following set of prefixes that (based on the data provided by three dictionaries) express ingressivity: na-, pro-, roz-, vy-, vzand za-. In the newly published academic grammar of Czech, Štícha et al. (2018), also mention the ingressive interpretation of the prefixes uand po-; on the other hand, they do not consider the prefix naas clearly ingressive. By combining these two classifications, we obtain the following set of Czech prefixes with apossible ingressive interpretation: Prefix rozThe most frequent ingressive suffix in Czech; in Riechzieglová’scorpus-based analysis, it constitutes almost half of all the tokens (cf. 2010, 103). Its usage does not present any considerable combinatory limitation. However, atendency to combine with verbs expressing emotional reaction is clearly observable (cf. Reichzieglová 2010, 103). The ingressive meaning is generally accompanied by reflexivity; non-reflective predicates with rozoften acquire acausative interpretation (see Čermák – Kratochvílová et al. this volume): (1) Pavel se smál. Paul refl laugh-pst.3sg.ipfv ‘Paul was laughing.’ (2) Pavel se roze-smál. Paul refl ingr-laugh-pst.3sg.pfv ‘Paul started to laugh.’ (3) Pavel roze-smál Marii. Paul ingr-laugh-pst.3sg.pfv Mary-acc ‘Paul made Mary laugh.’ Štícha et al. (2018, 1060) also mention the notion of high intensity that can be associated with this prefix in its ingressive interpretation: 5. ingressive periphrases in romance and their czech respondents 89 hořet (‘to be in flames’) rozhořet se (‘to start burning’ – in agiven context, the optional interpretation of progressive growing stronger of the flames is possible). Prefix vyŠtícha et al. consider this prefix to be one of the most polysemantic (cf. 2018, 1067). The ingressive interpretation is associated with its directional function, which relates to moving from one place to another or to leaving aplace. This naturally translates into the frequent combination with verbs of movement: (4) Pavel šel pomalu. Paul go-pst.3sg.ipfv slowly ‘Paul went slowly.’ (5) Pavel vy-šel z domu. Paul ingr-go-pst.3sg.pfv from house-gen ‘Paul went out of the house.’ This prefix can also accentuate the initial phase of an already very short process. Reichzieglová (2010, 108) also mentions the notion of unintentionality from part of the subject: křičet (‘to yell-ipfv’) křiknout (‘to yell-pfv’, i.e. ‘to yell only once and shortly’) vykřiknout (‘to produce ashort yell-pfv’) In our corpus analysis, this prefix was the second most frequent (after the prefix roz-). Prefix zaWhile this prefix is relatively frequent (in Reichzieglová’s(2010) corpus it is in the second position), it cannot express the sole notion of ingressivity, Reichzieglová (2010, 110) mentions a“semantic microstructure” combining the beginning of aprocess with other features. The most frequent notion here is that of avery short duration of the process: smát se (‘to laugh-ipfv’) rozesmát se (‘to start laughing-pfv’ – the laughing took awhile, maybe even progressively became stronger) zasmát se (‘to give ashort laugh-pfv’) Prefix naThis prefix combines ingressivity with semantic features related to the low intensity of the process or its low impact: 96 Literally: he started to read the newspaper. Jorge Amado, Pot (Suor), transl. Otokar Fischer, Prague: Lidové noviny, 1949. In approximately 10.1% of cases, the ingressivity of the original Romance construction was not overtly expressed in the Czech translation. As this is the second most frequent type, the question arises of whether ingressivity is completely absent in the Czech translation or whether it is inherently present in the internal MoA of the verb corresponding to the Romance infinitive, see (14), (15) and (16): (14) Es. Cuando la mujer empezó avolver en sí, Mendibj le tapó la boca y le entregó el papel. → Když žena zase přišla ksobě, Mendíbž jí zacpal ústa apodal jí ten lístek. Literally: when the woman regained consciousness again. Julia Navarro, Bratrstvo turínského plátna (La Hermandad de la Sábana Santa), transl. Vladimír Medek, Prague: Mladá fronta, 2006. (15) It. Mio fratello era nell’ età in cui si comincia aprendere piacere alle letture più sostanziose (…). → Bratr byl ve věku, kdy přicházíme na chuť hutnější četbě (…). Literally: when we develop the taste for more solid reading. Italo Calvino, Naši předkové (Inostri antenati), transl. Zdeněk Digrin – Vladimír Mikeš, Prague: Odeon, 1970. (16) Pt. Estava começando aescurecer quando chegou com seu rebanho diante uma velha igreja abandonada. → Už se stmívalo, když přivedl své stádo ke starému opuštěnéu kostelu. Literally: it was already getting dark. Paulo Coelho, Alchymista (OAlquimista), transl. Pavla Lidmilová, Brno: Jota, 1995. While ingressivity is not explicitly present in either of the examples presented above, the verbs used in the Czech respondents imply achange of state, which is naturally connected to the notion of the beginning of anew process. The verbal periphrases used in the Romance originals accentuate the notion of ingressivity which, nevertheless, is to acertain extent present in the semantics of the auxiliated verb. The Czech respondents do not entirely lack the notion of ingressivity; they solely omit its emphasising. 5.3.2.2 ingressive constructions expressing the beginning of aprocess and the notion of effort by part of the subject In this section, we analyse the following set of verbal periphrases: Es. Ponerse a+ infinitive It. Mettersi a+ infinitive 5. ingressive periphrases in romance and their czech respondents 97 Fr. Se mettre à + infinitive Pt. Pôr-se a+ infinitive Our goal is to determine whether (or how) the semantic notions of [+effort] [+subject’sinterest] [+motivation] that combine with [+ingressivity] are reflected in the Czech translations. The results are resumed in Table 5.2. Tab. 5.2. Czech translations of ingressive constructions expressing the beginning of aprocess and the notion of effort by part of the subject Verb + infinitive Ingressive construction + noun Prefix Other construction No translation Začít Other verb Dát se do Pustit se do Roz-Other prefix Es. Ponerse a (1,232) 587 27 125 48 103 78 63 201 It. Mettersi a (168) 41 0 23 11 21 4 15 53 Fr. Se mettre à (207) 124 0 16 4 29 15 16 3 Pt. Pôr-se a (123) 37 14 4 6 9 14 5 34 Total (1,730) 789 41 168 69 162 111 99 291 % 45.6 2.4 9.7 3.9 9.3 6.4 5.7 16.8 As can be observed, the dominance of the neutral translation type začít + infinitive is less distinctive. While this type of translation does not explicitly reflect the notion of effort from the part of the subject that is associated with these periphrases, its presence can be often found in the very semantics of the infinitive appearing in the construction, which inherently implies the subject’sinvolvement and interest, see (17), (18), (19). (17) Es. En fin, la divergencia se hace superlativa cuando se ponen apensar en los medios que exige una instauración de la paz sobre este pugnacísimo globo terráqueo. → Ksvrchovaným neshodám dochází, začnou-li uvažovat oprostředcích nastolení míru na naší velmi svárlivé zeměkouli. Literally: if they start to think about. José Ortega y Gasset, Vzpoura davů (La rebelión de las masas), transl. Václav Černý – Josef Forbelský, Prague: Naše vojsko, 1993. 98 (18) It. All’inizio aveva accumulato idee, poi si era messa ariempire quaderni d’appunti. → Na začátku nahromadila nápady apak začala zaplňovat sešity poznámkami. Literally: she started to fill the notebooks with notes. Alessandro Baricco, City (City), transl. Alice Flemrová, Prague: Volvox Globator, 2000. (19) Fr. De temps en temps, il s’approchait de notre groupe, nous écoutait et quand je me mettais à raconter mes histoires françaises, il me fixait d’un air méfiant. → Občas se knašemu hloučku přiblížil, chvíli nás poslouchal, akdyž jsem začal vykládat svoje francouzské historky, upřel na mě podezíravý pohled. Literally: when Istarted to narrate my French tales. Andrei Makine, Francouzský testament (Le Testament français), transl. Vlasta Dufková, Prague – Liberec: Paseka, 2002. Once again, the type labelled as no translation was the second most frequent. As in the case of the neutral constructions analysed in Section 5.3.2.1, the absence of any explicit expression of ingressivity was often associated with verbs expressing achange of state or a momentary action, i.e. verbs containing inherent ingressivity (20): (20) Pt. Ah, porque não me disse? – e pôs-se alimpar as mãos ao avental. → “Proč jste to neřekl hned?” – aotřela si ruce do zástěry. Literally: she wiped her hands on the apron. Fernando Namora, Muž smaskou (OHomem Disfarçado), transl. Pavla Lidmilová, Prague: Svoboda, 1979. Nevertheless, the missing explicit notion of ingressivity in many cases also resulted from the strengthening of the secondary semantic notions of effort expressed by these periphrases. This kind of translation is represented by (21) and (22): (21) Es. Después se pusieron afumar hombro contra hombro, satisfechos. → Pak se opřeli jeden druhému orameno aspokojeně pokuřovali. Literally: they were happily puffing away. Julio Cortázar, Nebe, peklo, ráj (Rayuela), transl. Vladimír Medek, Prague: Mladá fronta, 2001. (22) It. Mi sono messa acorrere verso la foresta come non avevo mai corso prima d’allora, e ho continuato finché sono crollata perterra; (...). → Hnala jsem se klesu jako ještě nikdy vživotě apak jsem běžela dál, dokud jsem nepadla; (...). 5. ingressive periphrases in romance and their czech respondents 99 Literally: Irushed towards the wood. Sebastiano Vassalli, Nespočet (Un infinito numero), transl. Kateřina Vinšová, Prague – Litomyšl: Paseka, 2003. The Czech respondent of the neutral Spanish verb fumar (‘to smoke’) in (21) contains ahigher level of expressiveness since the verb pokuřovat (in contrast to the neutral kouřit – ‘to smoke’) also implies certain pleasure and involvement in the action of smoking, thus roughly corresponding to the English verb ‘to puff’. In (22), the neutral Italian verb correre (‘to run’) is translated by the Czech verb hnát se (‘to rush’, ‘to dash’), which in its semantics contains the notion of [+effort] [+subject’sinterest] [+motivation]. 5.3.2.3 ingressive constructions expressing the beginning of aprocess and the notions of suddenness and unexpectedness In this section, we analyse the Spanish constructions echarse a+ infinitive and echar a+ infinitive. While the additional semantic notions expressed by these periphrases are the same (see García Fernández et al. 2006, 121–126), they differ in the combinatorics since echarse acombines mostly with verbs of emotional reaction while echar acombines with verbs of movement (cf. Gómez Torrego 1999, 3374). The infinitive completions with f ≥ 10 that were found in our corpus are resumed in Tables 5.3 and 5.4. Tab. 5.3. Infinitive completions of echarse a. Echarse a+ Frequency reír (‘to laugh’) 212 (60.6%) llorar (‘to cry’) 196 (27.4%) temblar (‘to tremble’) 11 (3.1%) andar (‘to walk’) 11 (3.1%) correr (‘to run’) 10 (2.9%) Tab. 5.4. Infinitive completions of echar a. Echar a+ Frequency correr (‘to run’) 124 (48.0%) andar (‘to walk’) 118 (39.9%) caminar (‘to walk’) 16 (5.4%) volar (‘to fly’) 12 (4.0%) The limited combinatorics of these constructions also influenced the Czech translations resumed in Table 5.5. 100 Tab. 5.5. Czech translations of ingressive constructions expressing the beginning of aprocess and the notions of suddenness and unxepectancy Verb + infinitive Ingressive construction + noun Prefix Other construction No translation Začít Other verb Dát se do Pustit se do Propuknout v Roz-Other prefix Es. Echarse a (350) 24 0 78 4 3 188 23 21 9 Es. Echar a (296) 2 0 26 0 0 76 118 51 23 Total (646) 26 0 104 4 3 264 141 72 32 % 4 0 16.1 0.6 0.5 40.9 21.8 11.1 4.9 The dominant type of translation was the prefix rozfollowed by the translation type marked as other prefix (generally corresponding to the prefix vyhere). With the construction echarse a+ infinitive, the analytic type dát se do + noun was also frequently used (approximately 22.3% of all respondents). The strong dominance of these respondent types is not surprising since they correspond extremely accurately to all notional and combinatory characteristics of the analysed constructions. All three types imply asudden beginning of aprocess. Rozand dát se do prefer the combination with verbs or nouns corresponding to emotional reaction (plakat ‘to cry’, brečet ‘to cry’, smát se ‘to laugh’ / pláč ‘crying’, brek ‘crying’, smích ‘laughter’), in its ingressive interpretation the prefix vygenerally combines with verbs of movement (see Section 5.2.4.2). Therefore, the prototypical respondents found in the analysed concordance can be represented by (23), (24) and (25): (23) Es. Se revelaba sólo amedias, fugazmente, en un juego exasperante de sombras chinescas, pero al despedirse, cuando ella estaba apunto de echarse allorar por hambre de amor, le entregaba una de sus prodigiosas cartas. Objevoval se jen napůl, prchavě, vnesnesitelné stínohře, která ji přiváděla kzoufalství, ale při loučení, kdy se zhladu po lásce už už chtěla rozplakat, předal jí jeden ze svých úžasných dopisů. Literally: she was about to start crying (plakat = ‘to cry’, rozplakat se = ‘to start crying’). Isabel Allende, Dcera štěstěny (Hija de la fortuna), transl. Monika Baďurová, Prague: BB Art, 2004. (24) Es. Ruibérriz se echó areír ahora, el labio vuelto y la dentadura relampagueante como si le estallara. Ruibérriz se dal do smíchu, horní ret se mu ohrnul azuby se blýskaly, jako by metaly blesky. 5. ingressive periphrases in romance and their czech respondents 101 Literally: he gave himself into laughter (he started to laugh). Javier Marías, Vzpomínej na mě zítra při bitvě (Mañana en la batalla piensa en mí), transl. Marie Jungmannová, Prague: Argo, 1999. (25) Es. La muchacha echó aandar. Dívka vykročila. Literally: the girl made first steps (she started to walk towards somewhere). Camilo José Cela, Úl (La colmena), transl. Alena Ondrušková, Prague: Odeon, 1968. 5.3.2.4 ingressive constructions expressing the beginning of aprocess and the notions of suddenness, abruptness and previous retention In this section, the following set of periphrases will be analysed: Es. Romper a+ infinitive Es. Largarse a+ infinitive It. Scoppiare a+ infinitive Pt. Romper a+ infinitive All these constructions are highly stylistically marked and, consequently, less frequently used than the periphrases analysed in previous sections. The semantic notions Tab. 5.6. Czech translations of ingressive constructions expressing the beginning of aprocess and the notions of suddenness, abruptness and previous retention Verb + infinitive Ingressive construction + noun Prefix Other construction No translation Začít Other verb Dát se do Pustit se do Propuknout v Roz-Other prefix Es. Romper a (58) 5 0 10 0 6 27 3 6 1 Es. Largarse a (36) 5 0 7 3 0 15 1 3 2 It. Scoppiare a (16) 0 0 6 0 2 3 2 3 0 Pt. Romper a (15) 0 0 11 0 2 2 0 0 0 Total (125) 10 0 34 3 10 47 6 12 3 % 8 0 27.2 2.4 8 37.6 4.8 9.6 2.4 102 associated with these constructions naturally result in their frequent combination with verbs of emotional reaction, thus expressing the subject’sintention to suppress or slow down the manifestation of his feelings and the abrupt outbreak. While in English, similar notions can be attributed to the idiom ‘to burst into tears’, Czech offers asemi-productive ingressive construction propuknout v+ noun, which also combines with verbs of emotional reaction and displays the same semantics as the Romance constructions analysed here (see Section 5.2.4.2.2). However, despite the existence of this apparently ideal Czech respondent, the Czech respondents, as resumed in Table 5.6, display great variability. As in the case of echarse a+ infinitive, the most frequent respondent type was the prefix roz-and the construction dát se do + noun. The construction propuknout v+ noun appeared more frequently than with echarse a+ infinitive although its overall frequency proved to be relatively low. 5.3.2.5 ingressive constructions expressing the beginning of aprocess and the notions of suddenness, abruptness and vehemence While the Spanish ingressive periphrases lanzarse a+ infinitive points out notions similar to those presented in Section 5.3.2.4, such as suddenness and abruptness, we analyse it separately to find out whether the substitution of the seme [+previous retention] with the seme [+vehemence] has any consequences on the Czech respondents. The results are summarised in Table 5.7. Tab. 5.7. Czech translations of ingressive constructions expressing the beginning of aprocess and the notions of suddenness, abruptness and vehemence Verb + infinitive Ingressive construction + noun Prefix Other construction No translation Začít Other verb Dát se do Pustit se do Vrhnout se do/na Roz-Other prefix Es. Lanzarse a (27) 8 3 0 3 2 1 3 3 4 % 29.6 11.1 0 11.1 7.4 3.7 11.1 11.1 14.8 The analysis reveals that the presence of [+vehemence] has strong consequences on Czech respondent types. While the notions of sudden beginning and abruptness (potentially accompanied by the previous retention) expressed by the periphrases analysed in Sections 5.3.2.3 and 5.3.2.4 found their prevailing respondents in the prefix roz- (eventually the prefix vyin the case of verbs of movement) and the construction dát se do + noun, lanzarse a+ infinitive lacks any clearly dominant Czech respondent. The notion of vehemence and an abrupt or sudden beginning expressed by the Spanish construction finds its theoretical respondent in the Czech constructions 5. ingressive periphrases in romance and their czech respondents 103 vrhnout se do/na. Nevertheless, this type of translation was found only twice in our corpus. The remaining respondents oscillated between the complete omission of any notions apart from the mere beginning of aprocess (26) and an explicit expression of vehemence which also supposed the lack of [+ingressivity] (27): (26) Es. Casi temí que de entre la jungla de la habitación pudiera surgir un borzog y se lanzara amorderme las pantorrillas . Skoro jsem dostal strach, že zté džungle vyskočí borzog azačne mě hryzat do lýtek. Literally: he starts to bite my calves. Pablo Tusset, Nejlepší loupákův zážitek (Lo mejor que le puede pasar aun cruasán), transl. Ondřej Nekola, Prague: Garamond, 2007. (27) Es. Es como si de pronto se hubiera lanzado arecobrar el tiempo perdido conmigo. Jako by si honem chtěla vynahradit čas, který se mnou ztratila. Literally: as if she quickly wanted to compensate for the time. Arturo Pérez Reverte, Kůže na buben (Piel del tambor), transl. Vladimír Medek, Prague: Euromedia Group, 2004. 5.4 concluding remarks The analyses presented in this chapter revealed two systemic similarities between Romance languages and Czech. 1) The tendency to combine ingressivity with other qualitative features referring to the beginning of aprocess Both in Romance languages and in Czech, aclear tendency to combine [+ingressivity] with other semes can be observed. In Romance, the secondary notions result from the original semantics of the semi-auxiliary verb (for an extended discussion, see Kratochvílová – Jindrová 2017) and are expressed through the periphrastic construction. In Czech, the accumulation of several aspectual-qualitative characteristics is typical for prefixes; nevertheless, the analyses proved that the verbo-nominal constructions dát se do / pustit se do + noun are also highly productive in the present-day language. 2) Combinatory tendencies/limitations Both Czech prefixes and verbo-nominal ingressive constructions display certain combinatory preferences. It is interesting to observe that these limitations are often comparable to those found with Romance periphrastic constructions. In Romance languages, there is aset of ingressive expressions that often combine with verbs of emo- 104 tional reaction (Es. echarse a/ Es. romper a/ It. scoppiare a+ infinitive).54 In Czech, the tendency to combine with similar verbs can be observed with the prefix rozand the verbo-nominal constructions dát se do and propuknout v+ noun. Atendency to combine with verbs of motion can also be found both in Spanish (echar a+ infinitive) and in Czech (the prefix vy-). The key structural differences can be summarised in three points: 1) Formal expression of ingressivity While in the Czech language ingressivity and the category of MoA, in general, is typically associated with prefixes, the analyses prove that this category has, in fact, three major forms of expression: the neutral construction začít + infinitive, verbal prefixes (especially roz-) and the verbo-nominal constructions dát se do and pustit se do + noun. The heterogeneity of Czech respondents presents the category of ingressivity in anew light. We conclude that the notion of beginning aprocess is distributed through the whole Czech verbal system and is systemically expressed both analytically (verbal and verbo-nominal constructions) and synthetically (prefixes). These observations could lead to ageneral question of whether identifying the category of MoA solely with prefixes in Czech is appropriate and whether notions such as ingressivity, iterativity and frequentativity should not be comprehended rather as abstract and polyfacetic categories that display alarger scale of systemic expressions. An analysis of the concurrence of word-formatting and analytical (lexical) resources could provide new insights into the category of MoA in Czech. 2) Productivity of highly stylistically marked constructions The analyses have also revealed that both in Romance languages and in Czech, the beginning of aprocess is acommonly expressed semantic feature that has, in all the analysed languages, its preferred means of expression. Nevertheless, their set differs notably not only according to the specific Romance language but also when contrasting the highly stylistically marked Romance constructions and their Czech respondents. Despite the fact that the Czech language offers atheoretically ideal counterpart for the constructions Es. romper a/ Es. largarse a/ It. scoppiare a/ Pt. romper a+ infinitive (i.e. the verbo-nominal construction propuknout v+ noun) and for the Spanish periphrasis lanzarse a+ infinitive (i. e. vrhnout se do/na + noun), the frequency of these translations was very limited, and the Romance constructions were generally translated with aconstruction displaying alower degree of stylistic markedness. When analysing ingressivity in Czech as awhole and considering all its above-mentioned systemic expressions (začít + infinitive, verbo-nominal constructions, prefixes) as respondents of 54 The Portuguese periphrasis desatar a+ infinitive displays similar combinatory preferences; however, it was not analysed here due to its low frequency in the corpus. In French, there is afrequently used expression éclater de rire (‘to burst into laughter’), which, nevertheless, combines solely with rire. Therefore, we consider it to be an idiom and not aproper verbal periphrasis. 5. ingressive periphrases in romance and their czech respondents 105 Romance ingressive periphrases, we conclude that their set and their usage is more comparable to Italian or French than to Spanish or Portuguese. 3) Inherent MoA and its explicit emphasising The high frequency of cases where the Czech respondents of Romance ingressive constructions did not express ingressivity at all or lacked any explicit expression of secondary semantic notions attributed to the original periphrasis, suggests that Czech displays agreater sensitivity to meanings related to MoA that are, nevertheless, inherent and are contained solely in the semantics of averb. This observation supports our original claim that Aktionsart in the narrowest sense of the term is inherently connected to other expressions of MoA and that the category of the manner of action should be analysed in abroader sense and should not be identified with one single expression tool. 112 gerunds are taken into account, since they convey the “adverbial subordination” typical for converbs. 6.2.2 semantic interpretation of the (adverbial) romance gerund The scale of meanings conveyed by the Romance adverbial gerund (and the Czech transgressive) is very large (König – Auwera 1990, 342; Halmøy 2003a etc.), ranging from the accompanying circumstance to the manner, means, cause, temporal meaning (repère temporel, see Gettrup 1977), concession and condition. Moreover, these meanings often overlap (e.g. manner/means, cause and temporal meaning of anteriority, etc.) and the overall meaning of the gerund may remain vague.64 Halmøy (1982, 2003a) suggested to systematise the gerundial meanings in two groups: on the one hand, the “Type A”, based on the (chrono)-logical relationship between the gerund and the main clause and including the temporal meaning, the cause, the condition and the concession; on the other hand, “Type B”, conveying pure accompanying circumstance (circonstance concomitante). Kleiber (2007b, 117 or 2009, 19) later stated that the French gérondif does not convey any of these meanings and he reduced the meaning of the French gérondif on the interpretative instruction, which is very close to Haspelmath’sdefinition of the converb given above: “associer sur un mode subordonné ou circonstanciel le procès du SG [syntagme gérondif] à la prédication principale”. Factors influencing the interpretation of the gerund are grammatical, syntactic, semantic and pragmatic (König 1995, König – Auwera 1990, 337 or Nádvorníková 2012) and involve not only the gerund but also the main verb.65 For example, at the morphological level, the compound form of the gerund conveys the meaning of anteriority, often triggering causal interpretation (the cause preceding the consequence). Similarly, the conditional interpretation of the gerund is often based on the conditional form of the main verb. One of the main syntactic factors involved in the semantic interpretation of the gerund is its position vis-à-vis the main verb: in Portuguese, for example, gerunds in the anteposition often convey the temporal meaning of anteriority, whereas the postposed forms are more related to the posteriority. In French, the anteposition may have the same effect as in Portuguese although the French gerund, even in the postposition, never conveys posteriority (see the constraint of en in 6.1) – this meaning is expressed by the other V-ant form, the present participle. König (1995, 69) illustrates the complexity of the interpretation of contextual converbs in the example of concession, which is based on inferences, i.e. on pragmatic Karlík, Nekula – Rusínová 1995, 336-337). However, Dvořák (1970, 37–45), in his diachronic study points out that 30% of transgressives in the 17th century, i.e. before the normative intervention, were non-coreferential. 64 Moortgat (1978, 157) considers the French gerund a“semantic chameleon” (see also Halmøy 2003a). 65 Kleiber (2009) points out that the semantic interpretation of the gerund is based on the lexical and aspectual meanings of the gerund and the main verb. 6. the romance gerund and its czech respondents 113 factors (p & q, if p then not-q). Due to the complexity of its interpretation, the concessive meaning of the gerund is signalled by lexical means (in French the adverb tout preceding the gerund,66 in Portuguese mesmo/embora or in Spanish by aun and in Italian by pur). Nevertheless, the causal relationship also involved in the concessive meaning, is not indicated lexically, but is deduced on the basis of shared knowledge (cf. the meaning of manner in Fr. il est parti en claquant la porte / Es. se fue cerrando la puerta / ‘he left slamming the door’ and causal meaning in Fr. il aréveillé son petit frère en claquant la porte / Es. Despertó a su hermanito cerrando la puerta ‘he woke his little brother by slamming the door’, see Halmøy 2003a, 88 or Nádvorníková 2013a).67 The shared knowledge is also necessary for the interpretation of the meaning of manner and of the specific meaning of “equivalence” (Type A’, Halmøy 2003a, 100), e.g. Fr. il acommis une erreur en se mariant ‘he made amistake in getting married’ (the main clause is are-interpretation of the gerund; similarly in Es. Esta mañana ha caído estrepitosamente el mercado americano, confirmándose así los pronósticos de la prensa – ‘This morning, the American market fell spectacularly, thus confirming the prognostics of the press’). Dvořák (1978, 29–30) identified asimilar meaning of the Czech transgressive. Finally, the meaning of the gerund may be indicated by aclose lexical relationship between the main verb and the gerund: in this very specific case, the gerund conveys the manner of the realisation of the main process, e.g. Fr. dire (‘to say’) – chuchoter (‘to whisper’) (dit-elle en chuchotant – ‘she said whispering’, Cs. zašeptala), Fr. arriver (‘to arrive’) – courir (‘to run’) (il est arrivé en courant – ‘he came running’, Cs. přiběhl; Es. salir (‘to leave’) – correr (‘to run’) (salir corriendo – ‘run away’, literally: ‘leave running’, Cs. utéct) etc. Halmøy (2003a, 104) identified this type (Type B’) for the verbs of speech and the verbs of movement in French. In the case of verbs of movement, this type of gerund illustrates the typological characteristics of Romance languages pointed out by Talmy (2000, 102), based on the means of expressing the Manner and the Path of the movement. In fact, all four Romance languages under scrutiny in this study are verb-framed languages, conveying the Path of the movement by the finite verb (arriver, salir etc.) and the Manner by the gerund (en courant, corriendo etc., op. cit., 222 or 114). Czech (and English), on the contrary, belong to the satellite-framed languages, conveying the Path by the satellite (particles in English – away, off, out, etc. and prefixes in Czech – při-, od-, vyetc.) and the manner of movement of the finite verb (English) or the verb root (in Czech). We expect that the analysis of the Czech respondents of the Romance gerund will reflect this typological difference. 66 The adverb tout preceding the gerund in French emphasises the simultaneity of both processes; if they are incompatible, the resulting meaning may be adversative or concessive (Nádvorníková 2007 and 2012). 67 Since the meanings of manner and means are closely interconnected (see Nádvorníková 2013a), we put them in our analyses in one group (Manner), see Section 6.5.1. 114 6.3 typology of czech respondents of the romance gerund As non-finite verb forms, converbs are an important means of syntactic condensation: they convey the meaning corresponding to afinite clause in fewer words and make the sentence structure more compact (see Mathesius 1961, 146). Thus, converbs may be placed in the middle of the scale of syntactic condensation (Vachek 1955 or Nosek 1964): between the finite clauses in asubordinate or acoordinate structure on the one hand, and the nominalisations, such as NPs or PPs, on the other. However, as suggested above, the Czech converb, transgressive, almost disappeared from contemporary use and is considered bookish (the present transgressive) or even archaic (the past transgressive). In other words, the potential systemic respondent of the Romance gerund is no more available in Czech. Consequently, we expect that most of the respondents of Romance converbs will belong to the other parts of the scale of syntactic condensation conveying the same meaning. In both cases, the implicit meaning of the Romance gerund may be rendered more explicit (by aconjunction introducing the finite clause or apreposition in PPs). The classification of the Czech respondents of the Romance gerunds takes into account in the first place their position on the scale of syntactic condensation: finite verbs (in coordinate or subordinate structures), non-finite verb forms, and nominalisations. The syntactic and semantic specification of finite clauses is then given by the conjunctions introducing these structures (protože – ‘because’; když – ‘when’;68 a– ‘and’;69 tak – ‘so’, že – ‘that’ etc.); juxtaposition (JUXT), i.e. asyndetic coordination, was classified as aspecial category. We considered a special category also for the change in the hierarchy of clauses (the original subordinate non-finite clause becomes the main clause, and the original main clause is rendered as asubordinate one).70 The semantic specification of nominalisations is especially given by the preposition (in the case of PPs, e.g. při příchodu – ‘upon arrival’) or by the case (the instrumental). The translation respondents reflecting the typological difference between Czech and Romance related to the way of expression of the Manner and Path of the movement (see Section 6.2.2) were put in aspecial category (see the analysis in 6.5.2.3). Finally, particular attention was focused on the potential systemic respondent of the Romance gerund, the transgressive (see Section 6.5.2.3). The analysis of Czech transgressives as respondents of Romance gerunds was completed by a backward analysis, focused on the Romance respondents of the transgressive (in translated as 68 We are aware of the fact that the conjunction když ‘when’ is semantically vague (polysemous), as it introduces subordinate clauses conveying not only temporal meaning but also condition or cause. In fact, this property corresponds perfectly to the semantic vagueness of the Romance gerund (see above). 69 We also took into account the adverbs specifying the relationship between the two clauses coordinated by a/ and, especially the adverbs přitom or zároveň (meaning ‘at the same time’). 70 In some cases, the original main clause is rendered completely implicit, especially in introductory clauses to direct speech, e.g. Fr. fis-je en baissant la tête ‘Isaid, hanging my head’ > Cs. sklonil jsem hlavu – ‘Ihung my head’ (see Nádvorníková, forthcoming b). 6. the romance gerund and its czech respondents 115 well as non-translated texts, see Chlumská 2017). In this way, it is possible to find out whether the Romance gerund represents the dominant respondent of the Czech transgressive or whether other types of respondents prevail. 6.4 data elaboration and quantitative analysis of the romance gerund The analysis of Romance gerund carried out on the InterCorp parallel corpus was focused on the adverbial (converbal) use of this form and on the typology of its Czech respondents (see Section 6.5). Nevertheless, the data extracted from the corpus provided some interesting findings concerning the comparison of the frequency of the gerund in the four Romance languages. Regular expressions used in the corpus took into account the potential variants of gerunds: clitics (postposed in Italian, Portuguese and Spanish and placed between the preposition and V-ant in French, see Section 6.1; and in Table 6.1 for the queries) and potential extraction noises (non-gerundival expressions in -ndo, such as Es. cuando – ‘when’). From this overall number of occurrences (see line 2 in Table 6.1), asample for Tab. 6.1. Occurrences of the Romance gerund in the InterCorp parallel corpus Frequency of gerund Fr. Es. It. Pt. 1 Size of the subcorpus (tokens) 1,533,451 9,326,150 1,631,204 1,485,541 2 Total No of concordance lines 2,64171 68,52672 8,91473 12,92474 3Size of the sample analysed manually – with noises 2,641 3,000 2,534 2,539 4No of gerunds (including periphrases) – without noises 2,363 2,971 2,322 2,422 5No of gerunds (without periphrases) 2,362 1,965 1,857 1,991 6 No of adverbial gerunds 2,362 1,561 1,857 1,448 71 [word="(E|e)n"] [word=".*ant"]; [word="(E|e)n"] []{1,2} [word=".*ant"] – the results obtained with this query contained 20% of noises. No occurrence of gerund including three positions between the preposition and the V-ant was found. 72 [word=".*ndo(me|te|le|lo|la|se|nos|os|les|los|las(lo|la|le(s)?)?)?" & tag!= "NP" &tag!= "NC" &tag!= "ADJ" & tag!= "ORD" &tag!="NMEA" & tag!="ADV" & word!="(.|!)?[Cc]u[aa]ndo" & word!= "ando|entiendo|extiendo|comprendo"] 73 [word=".*ando|.*endo" & !word="(q|Q)uando"]; [word=".*(a|e)ndo(ce|ci|gli|glie|la|le|li|lo|me|mi|ne|se|si|te|ti|ve|vi)"]; [word=".*(a|e)ndo((ce|ci|gli|glie|la|le|li|lo|me|mi|ne|se|si|te|ti|ve|vi)(ce|ci|gli|glie|la|le|li|lo|me|mi|ne|se|si|te| ti|ve|-vi)"] 74 [word=".*ando|.*endo|.*indo|.*ondo” & word!="[Qq]uando"] 116 manual analysis was extracted, in which we observed syntactic and semantic types of Romance gerunds and the corresponding Czech respondents (see line 3 vTable 6.1). The samples extracted for manual analysis contain from 2,500 to 3,000 occurrences (see line 3 in Table 3.1); in French, all the occurrences were analysed (compare lines 2 and 3 in Table 6.1). After the elimination of extraction noises, the resulting samples for manual analysis are slightly smaller (see line 4 in Table 6.1). For our study, focused on the adverbial uses of the gerund, we eliminated periphrastic gerunds (see Line 5 in Table 6.1 and Graph 6.1 above) and the non-adverbial uses of this form (see Line 6 in Table 6.1 and Section 6.4.2 below). Given the differences in the size of the four subcorpora (see the introductory chapter in this volume and line 1 in Table 6.1), the absolute frequency of the gerund in our corpus is the highest in Spanish. Nevertheless, abetter overview is obviously provided by relative frequencies (ipm – see Graph 6.1): their comparison reveals that the frequency of the gerund is the highest in Portuguese, followed by Spanish and Italian. Graph 6.1 shows the comparison of the relative frequencies of the Romance gerund in its periphrastic and non-periphrastic uses. It demonstrates that in Spanish and Portuguese, the general relative frequencies of the gerund are comparable. Nevertheless, the relative frequency of the periphrastic uses of the gerund is higher in Spanish than in Portuguese.75 This may be caused by the fact that since the 18th century, the gerund has been progressively replaced in European Portuguese by prepositional use of infinitive (a+ Infinitive) in some functions, namely in periphrastic constructions, e.g. Pt. estar fazendo / estar afazer (‘to be doing’). In future research, acontrastive analysis 75 The most frequent periphrases in the Spanish subcorpus involve the verbs estar (439 occurrences of 1,006) and seguir (202 occurrences). We also include among the periphrastic uses of the Spanish gerund in Spanish the semi-periphrastic constructions, such as empezar/comenzar ‘start’ + gerund – ‘start/begin by doing sth’. Graph 6.1. Relative frequency of the periphrastic and non-periphrastic gerund in French, Spanish, Italian and Portuguese 6. the romance gerund and its czech respondents 117 of the periphrastic gerund in Spanish, Portuguese (and Czech) may be of interest, but in this study, we restrict ourselves to non-periphrastic uses of the gerund to allow for the comparison with the cross-linguistic category of the converb. In Italian, the total relative frequency of the gerund is lower than in Spanish and Portuguese although the periphrastic uses of the gerund still represent 20% of this number (in Spanish, this proportion is 33% and in Portuguese only 18%). In French, the periphrastic use of the gerund is attested by only one occurrence of 2,363,76 which basically means that the French gerund is limited to non-periphrastic use. Graph 6.1 also shows that the frequency of the gerund is three times lower in French than in the other three Romance languages, which may be caused both by the quasi-absence of the periphrastic use of this form in contemporary French as well as by the concurrence of the present participle and the absence of non-adverbial uses of the gerund (see below).77 Thus, among the four Romance languages under investigation in this study, French appears to be the most specific in terms of frequency and use of the gerund. Nevertheless, the global figures provided in Graph 6.1 have to be specified in two aspects: first, related to the composition and size of the corpus used in this research (see Section 6.4.1), and second, regarding the syntactic functions of the non-periphrastic gerund (see Section 6.4.2 and Table 6.2). 6.4.1 factors influencing the frequency of the romance gerund The relative frequency of the gerund shown in Graph 6.1 may be particularly influenced by the fiction text type that prevails in our corpus. For example, adetailed analysis of the relative frequency of the gerund in the different texts of the Italian subcorpus reveals the lowest relative frequency of the gerund in two non-fiction texts (Giorgio Agamben – Mezzi senza fine, 3,453 ipm and Giuliano Procacci – Storia degli Italiani, 2,945 ipm; cf. 5,100 ipm in the whole Italian subcorpus in Graph 6.1). In the French subcorpus, we observed alower frequency of the gerund not only in non-fiction78 but also in fiction emulating spoken language (five of Asterix’sadventures, on average 539 ipm, and anovel by Ferdinand Céline – D’un château à l’autre, only 346 ipm; cf. 1,571 ipm in the whole French subcorpus in Graph 6.1).79 This tendency is corrobo76 The progressive periphrase aller (en) V-ant (see 6.2.1): La chair du requin aspirait les épices, des odeurs de coquillages allaient en montant. → Žraločí maso vsakovalo koření, vůně škeblí byla stále výraznější. Patrick Chamoiseau, Solibo Ohromný (Solibo le Magnifique), transl. Růžená Ostrá, Brno: Atlantis, 1993. Literally: the mussel odour was still stronger. 77 For this reason, all the occurrences of the French gerund available in the corpus were analysed (in the other three Romance languages, only samples were used). 78 Georges Duby – Dames du XIIe siècle (1,057 ipm) and Albert Camus – Carnets II (essays), 846 ipm and Antoine de Saint-Exupéry, Lettre à un otage, 610 ipm. 79 Moreover, as shown in Nádvorníková (2012), the text type influences not only the frequency of the gerund but also the proportions of its semantic types (in fiction, the meaning of accompanying circumstance prevails, whereas in non-fiction, the meaning of manner/means is the most frequent), see below for more details. 118 rated by recent research conducted on aspoken corpus (the gerund in spoken French is rare and often non-coreferential, see Escoubas-Benveniste 2013). In consequence, future research into gerunds should also investigate the frequency and use of this form in other text types. In Spanish and in Portuguese, another factor comes into play: the variety of the language. In non-European varieties of these languages, the relative frequency of the gerund is higher than in the European ones. In Spanish, for example, the first five ranks in the frequency list are occupied by texts written by Juan Carlos Onetti (Uruguay, 13,095 ipm), Alejo Carpentier (Cuba, 11,913 ipm), Juan Rulfo (Mexico, 11,857 ipm), Luis Sepúlveda (Chile, 11,657 ipm) and Jorge Zúñiga Pavlov (Chile, 10,667 ipm); cf. 7,274 ipm in the whole Spanish subcorpus in Graph 6.1. In Portuguese, the gerund is frequent not only in texts written by Brazilian authors (such as Jorge Amado) but also in novels by one European author (Eça de Queiroz), writing in the 19th century, when the Portuguese gerund was not yet so much affected by the competition with the construction a+ infinitive (see above and also Kratochvílová – Jindrová et al. this volume). Finally, it is also necessary to point out that in fiction, special idiolects of authors may considerably modify the frequency of the gerund. For example, in Spanish and in Portuguese, avery high frequency of the gerund in texts whose authors prefer writing in long, complex sentences (e.g. Teolinda Gersão in Portuguese, 14,576 ipm, cf. 8,300 ipm for the whole Portuguese subcorpus in Graph 6.1)80 can be observed. Moreover, two texts written by the same author may display considerable differences in the frequency of the use of the gerund, depending on the specific style of the text (e.g. the frequency of the gerund is 8,362 ipm in the novel Sogni di sogni by Tabucchi, but only 4,343 ipm in another novel by the same author – Il gioco del rovescio). For this reason, we systematically indicate the name of the author and the title of the text in the examples provided in the main part of our study (see 6.5).81 6.4.2 syntactic functions of the romance non-periphrastic gerund As mentioned in Section 6.2.1, the Romance gerund in its non-periphrastic use can fulfil not only the function of an adverbial modifier, typical for the converb but also other syntactic functions. However, in the samples manually analysed in our research (see line 5 in Table 6.1), the adverbial modifier represents the dominant function of the gerund in the four Romance languages under scrutiny: 80 In the French subcorpus, we observed apotential effect of the interference of the mother tongue of the author: the highest relative frequency of the gerund was identified in the novel Le Testament français written by Andreï Makine (5,184 ipm), whose mother tongue is Russian, where the converb деепричастиеis very frequent. 81 Despite the specificities given by the composition of the corpus, the data appears to be sufficiently reliable. For example, in the subcorpus of the French corpus FRANTEXT limited to novels published after 1950 (24 million tokens), the relative frequency of the gerund is 1,640 ipm; in the InterCorp French subcorpus, this number is comparable – 1,571 ipm (see Graph 6.1). 6. the romance gerund and its czech respondents 119 Tab. 6.2. Syntactic functions of the non-periphrastic Romance gerund Syntactic function of the non-periphrastic gerund Pt. Es. It. Fr. Converbal Adverbial modifier 1,448 72.73% 1,561 79.44% 1,822 98.12% 2,297 97.21% Absolute construction 169 8.49% 24 1.22% 20 1.08% – – Non-converbal Attributive 228 11.45% 133 6.77% – – – – Independent predicate 95 4.77% 88 4.48% – – – – Predicative 35 1.76% 153 7.79% – – – – Other 16 0.80% 6 0.31% 15 0.81% 65 2.75% Total 1,991 100% 1,965 100% 1,857 100% 2,362 100% The functions of the Romance gerund resumed in Table 6.2 can be divided into converbal and non-converbal. Apart from the typical and dominant function of the adverbial modifier (more than 70% in Portuguese and in Spanish, and more than 90% in Italian and in French), we can also consider as converbal the use of gerund in an absolute construction (see the same approach in Haspelmath 1995, 87) since it corresponds to the category of adverbial subordination (with semantic vagueness), see the following example in Italian: (1) It. Cominciavano i duelli, ma già il suolo essendo ingombro di carcasse e cadaveri,cisimuovevaafatica,(…).→Došlo na souboje, ale země užbylatak posetá harampádím amrtvolami, že byl každý pohyb těžký,(…). Literally: the ground was already. ItaloCalvino,Naši předkové (Inostri antenati), transl. ZdeněkDigrin – Vladimír Mikeš, Prague: Odeon, 1970. The gerund in an absolute construction can also be used as pragmatic marker conveying comments upon the main clause, e.g. in the so-called incisos in Spanish: (2) Es. Durante algunos minutos, que me parecieron eternos y que después, recordando todo el asunto, advertí de que efectivamente lo fueron, pensé en cuál debía ser mi comportamiento. → Pár minut, které mi připadaly nekonečné avlastně takové ibyly, když si to teď vybavuju, jsem přemýšlel, jak bych se měl zachovat. Literally: when Inow recall it. Jorge Zúñiga Pavlov, La Casa Blů (La Casa Blů), transl. Dita Grubnerová, Prague: Garamond, 2006. 120 In our corpus, the use of the gerund in an absolute construction is more frequent in Portuguese than in Spanish and in Italian (8% and 1% respectively)82 although this result may be influenced by the difference in size and composition of the subcorpora and is to be verified in further research. Unlike the gerund in an absolute construction, the category of “independent predicate” does not correspond to converbal use any more: although the gerund retains its (vague) adverbial meaning, it is not subordinated to afinite verb and is characterised by astrong narrative dynamism. Typically, there are several gerunds cumulated in one (complex) sentence. The non-subordinate character of this use of the gerund is reflected by its most typical respondent in Czech: independent finite clauses: (3) It. Saltandogli d’intorno, e correndogli sotto le zampe, e volandogli al di sopra, e pungendolo da tutte le parti; e non lasciandogli posa, e guizzandogli davanti, e riapparendogli da due lati quasi contemporaneamente fino amoltiplicarsi alle sue pupille e farlo impazzire, come se non un solo Nino gli fosse contro, ma cento. → Běhal by kolem něho, podlézal pod ním apřelétal nad ním, bodal ho ze všech stran anepopřál mu oddechu, tu by stál čelem kněmu apak zas na jedné avtu ránu na druhé straně, až by se vjeho zřítelnicích zmnožil apilot by začal šílet vdomnění, že proti sobě nemá jednoho, ale sto Ninu. Literally: he would run around him, he would slip under him and fly over, he would bite him and he would not give him amoment of peace, suddenly, he would appear in front of him. Elsa Morante, Příběh v historii (La storia), transl. Zdeněk Frýbort, Prague: Odeon, 1990. This use of the gerund is close to the “narrative converb” defined by Haspelmath – Nedjalkov (Haspelmath1995, 58;Nedjalkov1995,106–110 and 6.2 in this chapter) and similar to the narrative use of the present participle in French. Typically, non-converbal uses of the gerund are attributive and predicative. Although previously condemned by the norm (see Section 6.2.1), these functions are quite well attested to in our data in Portuguese and in Spanish – see Table 6.2 and the following example for Spanish: (4) Es. El ejemplo de las esposas de los militares actuando en vez de sus maridos fue rápidamente imitado. → Příklad manželek jednajících za své důstojnické chotě se brzy rozšířil. Literally: wives acting-adj. Isabel Allende, Paula (Paula), transl. Anežka Charvátová, Prague: Slovart, 1998. 82 In Italian, absolute constructions are considered formal and are restricted to written language. 6. the romance gerund and its czech respondents 121 Czech respondents of these uses of the gerund reflect their adjectival character: active participles in –ící (as in (4)), or subordinate attributive or predicative clauses. Finally, the category “other” in Table 6.2 especially includes the lexicalised forms of the gerund (e.g. en attendant – ‘in the meanwhile’ or en passant – ‘by the way’).83 The results presented in Table 6.2 confirm the difference between Spanish and Portuguese on the one hand, and Italian and especially French on the other. Typically, non-converbal, adjectival uses (attributive or predicative) are attested only in the former; the use of the gerund as independent predicate and in absolute constructions are attested in Spanish, Portuguese and Italian, but not in French. In French and in Italian, an adverbial modifier (including absolute constructions) represents the overwhelming majority of the occurrences of the gerund. Therefore, the gerund in Italian and in French may be considered as representing the category of the monofunctional, canonical converb (see the classification in Nedjalkov 1995, 104sq.), whereas in Portuguese and Spanish, the gerund is closer to the category of bi-functional, potential quasi-converb (ibid.). However, in the four Romance languages, the dominant use of the gerund is adverbial (in absolute or non-absolute construction), which will be investigated in the next section. 6.5 the adverbial romance gerund and its czech respondents This section, representing the core of our research, presents the semantic types of the adverbial Romance gerund – in absolute as well as non-absolute constructions (Section 6.5.1) and puts them in correspondence with their respondent types in Czech (Section 6.5.2). 6.5.1 semantic types of the romance adverbial gerund and the czech transgressive As suggested in Section 6.2, the Romance gerund belongs to the category of the contextual converb, i.e. its semantic interpretation depends on the grammatical, syntactic, semantic and pragmatic factors given by the context. For this reason, the quantification of the semantic types of the Romance gerund is very difficult – not only the meaning usually remains vague but the semantic types often overlap (especially the categories of CIR-accompanying circumstance and Manner, or TEMP-temporal and CAUSE).84 83 For the lexicalization of the gerund en passant, see Stosic 2012. 84 In our analyses, we distinguished the categories of X (representing the pure case of the meaning, e.g. CIR) and X+ (representing the meaning overlapping with another one, e.g. CIR+, combining the meaning of accompanying circumstance with the Manner). For the sake of simplicity, we did not include these categories in Table 6.3. 128 The last meaning of the gerund is signalled lexically: by the adverbs mesmo or embora in Portuguese:89 (20) Pt. É preciso não relaxar nunca, mesmo tendo chegado tão longe (...). → Nikdy nesmíme polevit, ikdyž dojdeme tak daleko, (...). Literally: even when we get that far. Paulo Coelho, Alchymista (Alquimista), transl. Pavla Lidmilová, Prague: Argo, 2005. (21) Pt. E asua intervenção parecia ter apenas opropósito de restabelecer orespeito que cada um devia aos camaradas, embora sabendo que obrilho das suas palavras provocaria nos outros um ressentido amargor de inferioridade. → Asvým zásahem jako by nesledoval jiný cíl než znovu nastolit vzájemnou úctu, přestože ví, že lesk jeho slov vyvolá vdruhých záštiplnou trpkost méněcennosti. Literally: even though he knows. Fernando Namora, Muž smaskou (OHomem Disfarçado), transl. Pavla Lidmilová, Prague: Svoboda, 1979. by the adverb aun in Spanish: (22) Es. Una fuerza instintiva e irrefrenable me impulsaba y habría continuado solo aun sabiendo que un turbio destino (y tal vez la muerte) me aguardaban. → Hnala mě instinktivní anezkrotná síla abyl bych pokračoval, ikdybych věděl, že mě čeká temný osud (snad ismrt). Literally: even if Iknew. Eduardo Mendoza, Pravda opřípadu Savolta (La verdad sobre el caso Savolta), transl. Petr Koutný, Prague: Odeon, 1983. by the adverb pur in Italian: (23) It. Il professor Broderfons, pur ammettendo la correttezza della mia osservazione, non riconosce ad essa alcun significato particolare. → Profesor Broderfons, ikdyž připustil správnost mé připomínky, jí nepřisuzuje žádný zvláštní význam. Literally: even if he admitted. Alessandro Baricco, Oceán moře (Oceano mare), transl. Miloslava Lázňovská – Alice Flemrová, Prague: Eminent, 2001. 89 We identified even one case of mesmo followed by the preposition em: Vejo que osenhor não riu, mesmo em tendo vontade. → Vidím, že jste se nezasmál, přestože jste chtěl. João Guimarães Rosa, Velká divočina (Grande Sertão), transl. Pavla Lidmilová, Prague: Mladá Fronta – Dauphin, 2003. Literally: even though you wanted. 6. the romance gerund and its czech respondents 129 and by the adverb tout in French (although most of the constructions tout + gerund do not convey concession but emphasise the simultaneity of the two processes, see (7)): (24) Fr. Tout en lui conservant pour l’éternité une fidélité muette, je m’estimais libéré de lui dès lors que des lecteurs s’en étaient emparés. → Třebaže jsem jí zůstával navždy věrný, cítil jsem se svobodně, teprve když se jí zmocnili čtenáři. Literally: even though Iremained devoted to her forever. Pierre Assouline, Zákaznice (La Cliente), transl. Lubomír Martínek, Prague: Prostor, 2000. It is worth noting that the concession is almost non-existent in the Czech transgressive (see Dvořák 1978), very probably due to the absence of an explicit lexical signal of this complex meaning. The Czech respondents of the Romance gerund, including the transgressive, will be investigated in detail in the next section. 6.5.2 czech respondents of the romance gerund In our research, we divided the Czech respondents of the Romance gerund into three main groups corresponding to the three levels of the scale of syntactic condensation: finite verb (specified in acoordinate or asubordinate clause), nominalisation (PP or NP, adverb, etc.) and non-finite verb forms (especially the transgressive). Table 6.4 shows the proportions of these categories: Tab. 6.4. Types of Czech respondents of the adverbial Romance gerund (PP – prepositional phrase, NP-instr – nominal phrase in the instrumental case, Tg – transgressive, Inf – infinitive) Type of respondent in Czech Fr. It. Es. Pt. Coordinate finite clause 985 41.70% 1,089 58.67% 903 56.97% 885 54.70% Subordinate finite clause 591 25.02% 335 18.05% 207 13.06% 232 14.34% Nominalisations (PP, NP-Instr, adverb) 372 15.75% 179 9.64% 137 8.64% 125 7.73% Non-finite verb form 56 2.37% 51 2.75% 140 8.83% 220 13.60% Other 358 15.16% 202 10.88% 198 12.49% 156 9.64% Total 2,362 1,856 1,585 1,618 130 The category of “Other” contains, on the one hand, missing respondents (misaligned segments, zero translations etc.), and special translational solutions on the other (esp. modulations90 ou dépouillements91). However, due to the high quality of the corpus (translations as well as alignment, see Rosen – Vavřín 2012), the frequency of misaligned or missing segments is very low. As for the special translational solutions, they may be interesting for further research in translation studies but in acontrastive analysis, they are not relevant as only recurrent translation respondents (see Krzeszowski 1990, 27) can reveal the structural, systemic similarities and differences between the languages (see the introductory chapter in this volume). Consequently, the category of “Other” will not be taken into account in the following explanations – with one exception only: the category of verbs of movement conveying the Manner, see Section 6.5.2.3). The most striking finding revealed in Table 6.4 is the overwhelming proportion of finite respondents of the Romance gerund in Czech, and very low frequency of the other types, especially the non-finite one. In fact, in the four Romance languages, the finite respondents (together with the coordinate and the subordinate ones) represent approximately 70% of all the occurrences analysed in our research.92 Therefore, the Czech language, in comparison with the four Romance languages, shows aclear tendency to verbal (finite) expression (such as e.g. Norwegian in comparison with German, see Fabricius-Hansen 1998 and 1999). This tendency is most likely caused by the strong stylistic markedness of the Czech converb, the transgressive. Other explanations are also possible, for example, Vachek (1961, 43), observing the same tendency in Czech in comparison with English, explains this by the typological differences between the two languages, stating that “there is certain interdependence between the analytical language structure and the reduced dynamism of the finite verb in English and on the other hand, the synthetic language structure and the strong dynamism93 of the finite verb in Czech”. In what follows, we will introduce adetailed analysis of the three types of Czech respondents, particularly with respect to the semantic types of the Romance gerund they correspond to. Particular attention will be paid to the transgressive (Section 6.5.2.3), the potential (converbal, nonfinite) systemic respondent of the Romance gerund. 90 E.g.Aquelle perversion obscure avez-vous cédéen fournissantà l’humanité, de votre plus belle plume, un acte d’autoaccusation d’une transparence aussi criante?→Jaká ničivá zvrácenost vás přimělasepsatpro lidstvo nejčistším stylem, jakého jste schopen, tak křiklavě průhledné sebeobvinění? Literally: to write. Amélie Nothomb, Vrahova hygiena (Hygiène de l’assassin), transl. Jarmila Fialová, Prague – Liberec: Paseka, 2001. 91 E.g. Un trio se donna des gifles et la jeune fille chanta en s’accompagnant au luth. → Trojice herců se fackovala aděvče zazpívalo sloutnou. Literally: with. Frédéric Tristan,Hrdinné útrapy Baltazara Kobera (Les Tribulations héroïques de Balthasar Kober), transl. Oldřich Kalfiřt, Prague: DharmaGaia – Dauphin, 2003. 92 Asimilar proportion of finite respondents was also observed by Malá – Šaldová (2015) in translations of English participial adjuncts in –ing into Czech (73%). 93 By the term dynamism of the finite verb Vachek means the tendency of the language to convey linguistic contents using finite forms, in opposition to non-finite ones. 6. the romance gerund and its czech respondents 131 6.5.2.1 finite verbs as respondents of the romance gerund Finite respondents of Romance gerunds may be divided into two major types: finite verbs in coordinate and subordinate clauses. The finite respondents represent 66.72% of occurrences in French, 76.72% in Italian 70.03% in Spanish and 69.04% in Portuguese. The analysis of the semantic types of the gerunds corresponding to these main types of Czech respondents revealed aclear correspondence between the gerund conveying the meaning of accompanying circumstance (Type B identified by Halmøy 1982 and 2003a, see Section 6.2.2) and the coordinate finite clause in Czech and astrong correlation between aspecific adverbial meaning of the gerund (Type Aaccording to Halmøy 2003a) and the subordinate clause. In fact, the adverbial relationship of the gerund conveying the meaning of accompanying circumstance to the main verb is only vague and corresponds to the basic interpretative instruction defined by Kleiber (2007b, 117 or 2009, 19 and Section 6.2.2 above): the two processes are only juxtaposed, co-occurring. On the contrary, gerunds conveying specific adverbial meanings necessitate the explicitation of the logical relationship between the two processes by a(subordinating) conjunction. In what follows, we intend to analyse the specific types of these two major categories and find to what extent they respect the original meaning of the Romance gerund. 6.5.2.1.1 coordinate finite clause as arespondent of the romance gerund The overwhelming majority of coordinate clauses as respondents of Romance gerunds are related to the other clause (corresponding to the original main clause) by the conjunction a/and; see (6) and the following example: (25) Es. Campillo lo miraba ahora con fijeza, fruncido ligeramente el ceño, tamborileando con los dedos sobre el brazo del sillón. → Campillo na něho teď hleděl upřeně slehce svraštělým obočím abubnoval prsty na opěradlo křesla. Literally: and he drummed his fingers. Pérez-Reverte, Arturo, Šermířský mistr (El maestro de esgrima), transl. Bronislava Skalická, Prague: Alpress, 1998. The cases of asyndetic relation (juxtaposition) were also placed in this category – see (5) and (7). This type of respondents (coordinate and asyndetic clause) represent approximately one-half of the respondents of the Romance gerund in our corpus (the least in French). The advantage of the coordinate clause as arespondent of the Romance gerund is the semantic vagueness of the relationship between the two clauses, which corresponds perfectly to the Romance gerund of this semantic type. The most frequent specification of this meaning in Czech is the adverb přitom/at the same time, explicitating the simultaneity of both processes (see (7)). However, the relation of 132 coordination/juxtaposition places the two clauses on the same level of importance, which does not correspond to the meaning of the converb, conveying aprocess considered to be secondary, circumstantial. In some cases, the translators try to retain the hierarchy of processes by changing the order of the clauses: the respondent of the converb is placed in the first position, and the respondent of the main clause is placed at the end of the sentence in aclearly rhematic position: (26) Fr. Là-dessus, elle apris le tisonnier pour soulever le couvercle de ma cuisinière et elle ajeté votre lettre dedans, en la froissant en boule, (…). → Potom vzala pohrabáč, nadzvedla poklop na kamnech, zmačkala dopis do kuličky (…). Literally: she crumpled the letter into abowl. Sébastien Japrisot, Příliš dlouhé zásnuby (Un Long dimanche de fiançailles), transl. Veronika Sysalová, Prague: Euromedia Group, 2005. Another potential semantic shift caused by the translation of the Romance gerund by acoordinate clause concerns the temporal relationship between them: in fact, if the two coordinate verbs are perfective in Czech, the meaning of simultaneity may be turned into succession. This type of shift is particularly frequent in introductory clauses: (27) Es. —Ya sé, ya sé aquién se parece —sonrió feliz, mostrando aLituma el alto de revistas multicolores. → „Aha, už to mám, komu je podobný!“ Šťastně se usmál aukázal Litumovi štos obrázkových časopisů. Literally: and he showed to Lituma. Mario Vargas Llosa, Tetička Julia azneuznaný génius (La tía Julia y el escribidor), transl. Libuše Prokopová, Prague: Mladá Fronta, 2004. 6.5.2.1.2 the subordinate finite clause as arespondent of the romance gerund As mentioned above, the subordinate finite clause is the most frequent respondent of gerunds conveying not asimple accompanying circumstance but aspecific adverbial meaning (temporal, causal, conditional, etc. – Type Aidentified by Halmøy 1982, 2003a). This type of respondent for example, in French, represents between 13.06% and 25.02% of the respondents of the Romance gerund (the least in Spanish and the most in French). This type of respondent explicates the semantic type of the gerund by asubordinating conjunction. Nevertheless, the most frequent conjunction introducing this type of respondent, když (‘when’), to acertain extent maintains the semantic vagueness of the gerund since it can convey both the temporal meaning (simultaneity – (12) 6. the romance gerund and its czech respondents 133 or anteriority – see (13)) as well as causal or condition nuances (such as (13), (16) or (17) or note 37). The temporal meaning is also rendered in Czech by more specific conjunctions than když, such as zatímco or jak (meanwhile, conveying simultaneity, see (28) and note 43) or jakmile (as soon as, conveying immediate anteriority – see (14)). (28) It. Uscendo dalla cucina incontrammo Aymaro. → Jak jsme vycházeli zkuchyně, potkali jsme Aymarda. Literally: as we were leaving the kitchen. Umberto Eco, Jméno růže (Nome della rosa), transl. Zdeněk Frýbort, Prague: Odeon, 1988.94 After the temporal conjunctions (including the polysemic když), the second most frequent specific semantic type rendered by subordinate finite clauses is Manner/ Means, introduced in Czech by the compound subordinators tak, že (‘so that’) and especially tím, že (‘by’) (see (11)): (29) Fr. Elles savent reproduire artificiellement n’importe quelle phéromone: passeport, piste, communication… juste en mélangeant judicieusement des sèves, des pollens et des salives. → Dovedou uměle vytvořit jakýkoli feromon: vstupní, stopovací, komunikační… prostě tím, že dovedně míchají šťávy, pyl asliny. Literally: by mixing skilfully. Bernard Werber, Mravenci (Les Fourmis), transl. Richard Podaný, Prague: Euromedia Group – Knižní klub, 2005. Subordinate finite clauses introduced by the conjunctions conveying the meaning of Manner/Means are quite frequent. For example, in French, they represent 17% of this type of respondent (together with the temporal conjunctions, they represent 85% of the subordinate clauses corresponding to French gerund). The same meaning can also be rendered by anoun phrase (NP in the instrumental case) although this type of respondent is limited by syntactic constraints and especially by the number of gerund complements (see below Section 6.5.2.2). Explicitation of the causal relationship by the conjunction protože/poněvadž (‘because’) is rare, as it is usually rendered by the polysemic conjunction když/when (see (13) or (16)). This type of respondent often corresponds to the gerund expressed by astatic verb conveying emotions (30) or by verba opinandi (31): 94 Or in French, with the conjunction zatímco (‘while’) as arespondent in Czech: En attendant les brioches, ils s’échangeaient puces, poux, morpions, gales… → Zatímco čekali na briošky, vyměňovali si blechy, vši, filcky, svrab… Literally: while they were waiting. Ferdinand Louis Céline, Od zámku kzámku (D’un château l’autre), transl. Anna Kareninová, Brno: Atlantis, 1996. 134 (30) It. Ancor bambina, mia madre restò incinta di me, — raccontava Torrismondo, — etemendo le ire dei genitori quando avessero appreso il suo stato, fuggì dal castello reale di Scozia e andò vagando per gli altopiani. → „Má matka nosila mě pod srdcem ještě jako dívka,“ vyprávěl Thorismund, „aprotože se obávala, že by ji rodiče zahrnuli hněvem, kdyby zjistili její stav, prchla ze skotského královského zámku atoulala se po horských pláních.“ Literally: because she was afraid. Italo Calvino, Naši předkové (Inostri antenati), transl. Zdeněk Digrin – Vladimír Mikeš, Prague: Odeon, 1970. (31) Pt. E o bom Ferrão sorria, sabendo que, sob aquela ferocidade de ímpio obtuso, havia um santo coração... → Adobrák Ferrão se usmíval, poněvadž věděl, že pod divokostí toho zavilého bezbožníka tepe šlechetné srdce... Literally: because he knew. José Maria Eça de Queiroz, Zločin pátera Amara (OCrime do Padre Amaro), transl. Zdeněk Hampl, Prague: SNKLU, 1961. In comparison with the other Romance languages, the causal relationship conveyed explicitly is even rarer among respondents corresponding to the French gerund because this form in French is considered incompatible with static verbs (see Halmøy 2003a); these verbs are used more in the form of the present participle; thus, the corresponding forms in (30) and (31) in French would be craignant and sachant and not ?en craignant and ?en sachant). The remaining semantic types – condition and concession – are usually rendered by the corresponding specific conjunctions in Czech: kdyby (‘if’) – see (18) in Pt. and (19) in Fr., and třebaže and ikdyž, meaning ‘although’ (signalled by mesmo or embora in Pt., see (20) and (21), by aun in Spanish, see (22), by pur in Italian, see (23) and by tout in French, see (24) in Section 6.5.1. To summarise, most of the subordinate finite clauses explicate the vague meaning of the non-finite Romance gerund (with the exception of the polysemic conjunction když), restraining it to one interpretation only. Asimilar effect is observed on the other extremity of the scale of the syntactic condensation, in nominalisations. 6.5.2.2 nominalisations as respondents of the romance gerund Nominalisations represent the second most frequent respondent on the scale of syntactic condensation corresponding to the Romance gerund, after the coordinate and the subordinate clauses (about 10%; the most are in translations from French and the least are in translations from Portuguese). The three most frequent types of nominalisations correspond to the three most frequent meanings of the Romance gerund: 6. the romance gerund and its czech respondents 135 s(‘with’) + NP The PP introduced by the preposition s(‘with’) conveys the meaning of accompanying circumstance, e.g. súsměvem (‘smiling’ It. sorridendo, see (8); Fr. en souriant; Es. sonriendo; Pt. sorrindo or ‘laughing’/ridendo etc.).95 This type of respondent is particularly frequent in introductory clauses, since it renders, in acondensed way, acircumstance of the reported speech:96 (32) Pt. — E estou também com vontade de ir rezar unia estaçãozinha para aliviar cá por dentro —ajuntou, suspirando. → „Achci se tam také pomodlit, aby se mému srdci trochu ulevilo,“ dodala spovzdechem. Literally: with asigh. José Maria Eça de Queiroz, Bratranec Bazílio (OPrimo Basílio), transl. Zdeněk Hampl, Prague: Státní nakladatelství krásné literatury, hudby aumění (SNKLHU), 1955.97 při (‘with’, ‘by’) + NP This type of PP usually corresponds to the gerund conveying the temporal meaning. In the four Romance languages, one of the most frequent gerunds having this respondent in Czech is při pohledu na ‘looking at’ – Fr. en regardant, Es. mirando, It. guardando, Pt. olhando: (33) Fr. Et tes amis seront bien étonnés de te voir rire en regardant le ciel. → Tvoji přátelé se budou strašně divit, až tě uvidí smát se při pohledu na nebe. Literally: with alook at the sky. Antoine de Saint Exupéry, Malý princ (Le Petit prince), transl. Zdeňka Stavinohová, Prague: Albatros, 1989. However, other verbs are also possible in this meaning: ‘running’ (It. correndo – Cs. při běhu), ‘saying that’ (Fr. en disant – Cs. při těch slovech) etc.98 95 E.g. It. “Perché”, gli risponde tuttavia l’altro, ridendo, “la bellezza era un trucco, per farci credere al paradiso, quando si sa che tutti noi siamo condannati fino dalla nascita.” → „Proč…?“ „Protože,“ odpoví mu král se smíchem, „krása je jenom obyčejný trik, abychom uvěřili, že je nějaký ráj, když každý naopak ví, že jsme od narození odsouzeni.“ Literally: with laughter. Elsa Morante, Příběh vhistorii (La storia), transl. Zdeněk Frýbort, Prague: Odeon, 1990. 96 For example in Italian and in French, this type of respondent represents about 30% of all the nominalisations. 97 Nevertheless, this type of respondent is not limited to the introductory clauses, cf. Émerveillés, les indigènes suivent longtemps les bateaux en chantant et en dansant au son des tambourins. (Davidson, Sur les traces d’Alexandre le Grand, 2002) → Žasnoucí domorodci sledovali dlouho lodě se zpěvem atancem za zvuku bubínků. Literally: with song and dances. Marie Thérèse Davidson, Po stopách Alexandra Velikého (Sur les traces d’Alexandre le Grand), transl. Vladimír Čadský, Prague: Knižní klub, 2005. 98 The temporal meaning is (less frequently) rendered also by the PP v(‘in’) + NP, e.g. Fr. en dormant – Cs. ve spánku (‘in sleep’), It. conversando – vhovorech (‘in conversations’) or za + NP (Fr. en marchant – za chůze (‘in walking’)). 136 NP-Instrumental This type of respondent is typical for the meaning of Manner/Means, see (10) for Spanish or the following example: (34) It. E quante ore ho trascorso afissare il bianco di un foglio di pergamena, pensando aciò che avrebbe potuto prendere vita su quel foglio, se Velthune avesse voluto aiutarmi… → Akolik hodin jsem strávilspohledem upřeným na bílý list pergamenu přemýšlením otom, co by se na tomto listě mohlo zrodit, kdyby mi Velthune chtěl pomoci… Literally: by thinking. Sebastiano Vassalli, Nespočet (Infinito numero), transl. Kateřina Vinšová, Prague – Litomyšl: Paseka, 2003. In the case of the instrumental, the noun is usually averbal noun in –ní, retaining the verbal meaning (cf. přemýšlením ‘by the thinking’ (34), odhalením ‘by the revelation’ in (10), or uškrcením ‘by the strangling’ in the note 50). Nevertheless, other nouns are also acceptable (e.g. zradou ‘by the treason’ in (10), láskou ‘by the love’, popisem ‘by the description’, diskusí ‘by the discussion’ etc.). The NPs in the instrumental keep the subordinate character of the gerund. However, their use in Czech is limited in two aspects: it is not able to render long gerundival clauses (the number of complements of verbal nouns in Czech being limited) and for some verbs, the corresponding verbal noun is not available in Czech. The last type of respondent of the Romance gerunds in Czech, the transgressive, does not have these constraints. Nevertheless, due to its stylistic properties, it is the least frequent from all the three members of the scale of syntactic condensation examined in this study. 6.5.2.3 non-finite verb forms as respondents of the romance gerund Among the three non-finite verb forms available in Czech, the transgressive is the most frequent among the respondents of the Romance gerund. This prevalence of the transgressive (in comparison with the other non-finite verb forms) reflects its converbal character (see Section 6.2). However, as shown in Table 6.4, it represents the least frequent type of Czech respondent of the Romance gerund in our corpus. This very low frequency of the transgressive among the respondents of the gerund is provided by its very low frequency in contemporary Czech in general. In our corpus, the occurrences of the transgressive are limited on the one hand by its stylistic specificity (the present, i.e. the imperfective transgressive is considered bookish; the past, i.e. the perfective transgressive is even archaic), and on the other hand by the overwhelming majority of only one meaning – the accompanying circumstance. Due to the stylistic specificity, the transgressive especially occurs in texts with specific, e.g. historical, stylisation. In the Italian subcorpus, for example, 59% of all the oc- 6. the romance gerund and its czech respondents 137 currences of the transgressive corresponding to the gerund come from only one text: Inostri antenati by Italo Calvino: (35) It. Dall’olmo, sempre cercando dove un ramo passava gomito agomito con irami d’un’altra pianta, si passava su un carrubo, e poi su un gelso. → Zjilmu, hledaje vždy místo, kde větev světvemi sousedního stromu se proplétala, na rohovník přelezl aposléze na morušovník. Literally: searching. Italo Calvino, Naši předkové (Inostri antenati), transl. Zdeněk Digrin – Vladimír Mikeš, Prague: Odeon, 1970.99 By using the transgressive, the translators in (35) intend to render the archaistic stylisation of the original. Similarly, in the Portuguese subcorpus, most of the transgressives corresponding to Portuguese gerunds are attested in texts written in the 19th century by Eça de Queiroz (this fact explains the high frequency of non-finite verb forms in translations from Portuguese, see Table 6.4): (36) Pt. São omelhor bocadinho deste vale de lágrimas – interrompeu com fatuidade oSavedra, dando palmadinhas sobre o estômago. → Jsou nejchutnějším soustíčkem vtomto slzavém údolí,“ přerušil ho ješitně Savedra, poplácávaje se po břiše. Literally: smacking his belly. José Maria Eça de Queiroz, Bratranec Bazilio (OPrimo Basílio), transl. Zdeněk Hampl, Prague: SNKLHU, 1955.100 Another factor influencing the frequency of transgressives in translation is the idiolect of the translator and the date of creation of the translation. For example, most of the occurrences of the transgressive corresponding to the French gerund are found in atranslation first published in 1965 (both occurrences of the past transgressive in translations from French come from this text): 99 The only occurrence of past (perfective) transgressive corresponding to the Italian gerund comes from the same text: E spartendo davanti asé le foglie ognuno dal ramo in cui stava scese aquello più basso, verso il ragazzo col tricorno in capo. → Akaždý, rozhrnuv před sebou listí haluze, na které seděl, na nižší větev slezl, blíže kchlapci střírohákem na hlavě. Literally: having pulled. Italo Calvino, Naši předkové (Inostri antenati), transl. Zdeněk Digrin – Vladimír Mikeš, Prague: Odeon, 1970. 100 Cf. Asimilar historical stylisation in the following text: Là-dessus il voulut me mettre dehors en invoquant l’heure tardive et son sommeil troublé. → Načež mě chtěl zase vystrnadit ven na déšť, odvolávaje se na pozdní hodinu asvůj přerušený spánek. Literally: invoking. André Pieyre de Mandiargues, Vlčí slunce (Soleil des loups), transl. Ladislav Šerý, Prague: Reflex, 1992. 144 6.6 conclusion The aim of this chapter was to analyse the Romance gerund and its Czech respondents and, on the basis of the cross-linguistic notion of the converb used as tertium comparationis, to find to what extent the Romance gerund corresponds to its potential systemic counterpart in Czech – the transgressive. From the morphological point of view, these non-finite verb forms are different: the Romance gerund, resulting from the Latin ablativus gerundii, is non-congruent whereas the Czech transgressive agrees with its controller in gender and in number. Despite this, all these forms are considered as converbs. From the syntactic point of view, they may be considered the middle member of the scale of syntactic condensation, between the finite verb (in acoordinate or asubordinate clause) and nominalisations. We first focused our analysis on the Romance gerund only – its frequency, syntactic functions and semantic interpretation. The analysis of the frequency and the syntactic properties of the gerund in the four Romance languages revealed important differences between the Spanish and Portuguese gerunds on the one hand and the Italian and, especially, the French form on the other. In fact, in Spanish and in Portuguese, the gerund is used not only in its adverbial converbal function (including absolute constructions) but adjectival uses (attributive as well as predicative) are also well attested while both languages (especially Spanish) make extensive use of the gerund in verbal periphrases. On the contrary, in Italian and French, the adjectival uses of the gerund are excluded and moreover, in French, the gerund is limited to the non-absolute adverbial use and the only verbal periphrase involving this form (aller (en) –ant) is extremely rare. As for the Czech transgressive, it seems closest to the French gerund (with respect to the syntactic properties): Tab. 6.7. Syntactic properties of the gerund in four Romance languages and of the Czech transgressive Romance gerund Es. Pt. It. Fr. Cs. non-converbal verbal periphrases + + + (+) – adjectival use + + – – – converbal (adverbial) absolute constructions + + (+) – (+)106 non-absolute + + + + + The differences in the syntactic functions of the gerund strongly influence the overall frequency of the gerund in the four languages. In Portuguese and Spanish, the relative frequency of the gerund (in all the uses together) is 8,300 and 7,274 ipm respectively; in Italian, the frequency is lower but still comparable (5,100 ipm) but in French, the relative frequency of the gerund is only 1,572 ipm. 106 According to Dvořák (1978), absolute uses were well attested up to the 17th century in Czech, even though in contemporary grammars, they are non accepted. 6. the romance gerund and its czech respondents 145 Observing the differences in the use of gerunds in the four languages, we suggested that in accordance with the classification by Nedjalkov (1995, 104sq.), the gerund in Italian and in French may be considered as a“monofunctional, canonical converb” (as well as the Czech transgressive) while in Spanish and in Portuguese, the gerund belongs more to the category of polyfunctional, potential quasi-converbs. The analysis of semantic types of converbal uses of the gerund (absolute as well as non-absolute) revealed, on the contrary, striking similarities between the four forms – and also the Czech transgressive (converb). In the five languages, the dominant meaning is the accompanying circumstance, corresponding to the basic interpretative instruction of the gerund (at least 40% of all the occurrences). However, this tendency also reveals the potential limitation of our research: in fact, this meaning is typical for narrative texts, as it allows for the expression of two co-occurring processes without astrict logical relationship. Since our corpus contains mostly fiction (see Nádvorníková this volume), the predominance of this meaning is inevitable. Therefore, future research, aimed at amore complex analysis of the question, should also include other text types, especially non-fiction. The second similarity in the semantic interpretation of the four Romance gerunds and the Czech transgressive was the ranking of the remaining meanings: in all the five languages, the second most frequent semantic type is temporal, followed by manner/ means and cause. The remaining semantic types (condition and concession) are rare. The extremely low frequency of concessive meaning can be explained by the high cognitive effort necessary for its decoding. The thorough analysis of the semantic types of the Romance gerund confirms that this form belongs to the contextual converb category, as already suggested by Haspelmath (1995): its meaning remains vague and is given by the context. On the basis of the research conducted by (Dvořák 1978) in Czech, the same confirmation may be given for the Czech transgressive. Contextual factors participating in the interpretation of the Romance gerund, as well as the Czech transgressive, are multiple: the aspectual and semantic relationship between the gerundival verb and the main verb, mode of the main verb (especially the conditional), position of the form vis-à-vis the main clause (the anteposition facilitates the temporal or causal interpretation) etc. The explicit lexical signals of the meaning are limited to the temporal meaning of the immediate anteriority (em in Portuguese and in Spanish) and especially to the concession (tout in French, aun in Spanish, mesmo or embora in Portuguese and pur in Italian). The lack of an explicit lexical signal may explain the quasi-absence of the concessive meaning in the Czech transgressive. In the second part of our study, we investigated the potential correlations of the semantic types of the Romance gerund and the types of its Czech respondents. Since the potential systemic counterpart of the Romance gerund, the transgressive, is rare in contemporary Czech and considered very formal and bookish (the imperfective form, conveying simultaneity) or even archaic (the perfective form, conveying anteriority), we expected that the other members of the scale of syntactic condensation will take its place. This hypothesis was confirmed only partially since the overwhelming majority 146 of the respondents of the Romance gerund belong to only one category: finite clause (coordinate clauses being twice or even three times more frequent than subordinate ones). Nominalisations represent only aminor part of the respondents. This result confirms the tendency of Czech for explicit verbal expression, which has also been observed in previous studies. The research also revealed astrong correlation between the semantic type of the Romance gerund and the type of its Czech respondent: the basic meaning of pure accompanying circumstance is dominantly rendered in Czech by acoordinate finite clause (with the conjunction a‘and’), whereas the gerunds conveying more specific adverbial meanings usually have the corresponding adverbial subordinate clause as arespondent. In contrast with specific subordinating conjunctions, the coordinating conjunction aretains the vague semantic relationship between the clauses in Romance although the coordination modifies their hierarchy, as it replaces the subordination, typical for converb, and puts both clauses on the same level of importance. As expected, the non-finite respondent of the Romance gerund, the transgressive, was very rare. The final research focused specifically on the transgressive and its respondents in Romance showed that with the exception of French, the gerund is effectively the dominant respondent of the transgressive. In French, the gerund is strongly in concurrence with the other V-ant form – the present participle. However, these results are limited in several aspects: the low frequency of the transgressive in our corpus and its limitation to only one semantic type (accompanying circumstance), the potential influence of authors’ idiolects (in translations from Czech) and the translators’ strategies (in the opposite direction of translation) etc. The limitations of the research carried out on the transgressive are also applicable to the whole research presented in this study. As mentioned in the introductory chapter in this volume, the subcorpora of the four Romance languages under investigation are of different sizes, and in numerous aspects are not representative, especially with regard to the variety of text types. Future research into the Romance gerund and its respondents in Czech, founded on corpora containing not only fiction but also an important sub-corpus of non-fiction or journalistic texts, might bring interesting insights into the different uses of Romance converb. From the contrastive point of view, the present study also revealed the necessity to examine the converb (not only the Romance gerund but also the Czech transgressive) with respect to its valeur in the system of the other non-finite forms (infinitives and participles). 7. formal expressions vs abstract linguistic categories 147 7. formal expressions vs abstract linguistic categories: coming to terms with potential (non-volitional) participation, iterativity, causation, ingressivity and adverbial subordination petr čermák dana kratochvílová olga nádvorníková pavel štichauer 148 7.0 introduction Throughout the present monograph, we have analysed five different phenomena that can be found in Spanish, Italian, French and Portuguese. In abstract terms, these phenomena could be defined as an expression of potential (non-volitional) participation, repetition (iterativity), causation, beginning of an action (ingressivity) and adverbial subordination (in the broadest sense of the term). In the Romance languages studied, all these phenomena dispose of ameans of expression that is typically associated with them, i.e. expresses these notions in their “purest” form while also being highly productive and frequent. The potential (non-volitional) participation is expressed through the suffix –ble/-bile/-vel, iterativity through the prefix re-/ri-,107 causativity is expressed through the construction hacer/fare/faire/fazer + infinitive, ingressivity through awide range of partially synonymous verbal periphrases and non-finite adverbial subordination is typically marked by the gerund. In the Czech language, the above-mentioned notions are coded in adifferent manner and their prototypical Romance forms of expression do not always find aclear systemic Czech counterpart. We can imagine ascale ranging from an apparently perfect analogy between the Romance expression and Czech (both in terms of the notions typically attributed to the expression and its formal manifestation), through partial correspondence (either in terms of non-corresponding secondary notions attributed to the expression and/or its formal expression) to an apparently missing form of systemic expression. This initial schema based on Romance and Czech grammars is represented in Table 7.1. The objectives of our study can be subsumed into the following points: 1) While not being our main goal, the decision to consider the above-presented phenomena as generally Romance and put them into contrast with Czech, required at least abrief comparison of their functions across the Romance languages under scrutiny and to pinpoint some general differences. 107 In the case of iterativity, the prefix re-/rishares afunction with the iterative verbal periphrasis Es. volver a+ infinitive, It. tornare a+ infinitive, Pt. voltar a+ infinitive. However, there is no similar verbal periphrasis in French. 7. formal expressions vs abstract linguistic categories 149 2) The main objective of our study was acorpus-based analysis of the Czech respondents of the Romance linguistic phenomena. This focussed primarily on the question as to whether the existence of apartial or apparently absolute systemic Czech counterpart automatically means that this counterpart will be the predominant respondent in the corpus and as to whether Romance phenomena that do not pos sess any clear Czech systemic counterpart have any dominant Czech respon dent(s), which can be structurally defined. 3) The analysis of the Czech respondents and the secondary notions they expressed also enabled us to reformulate some of the original assumptions regarding the semantics of the analysed Romance phenomena. 4) We could evaluate the exploitation possibilities and limitations of parallel corpora and the possible contribution of corpus-based analyses to the discussion regarding the nature of aconcrete language phenomenon. In the following sections, the above-presented points are discussed in greater detail with reference to the conclusions we were able to make in light of the conducted corpus analyses. 7.1 correspondences of the analysed phenomena across romance languages Adata-based comparison among all four of the languages studied proved to be difficult due to the differences in the size of the respective subcorpora with which we worked (see Nádvorníková this volume). The limited amount of Italian, French and Portuguese data (in contrast to the considerably larger Spanish subcorpus) turned out to be especially relevant when analysing the complex words (Štichauer et al. this volume), Tab. 7.1. Systemic counterparts of the analysed Romance phenomena Romance expression Czech systemic counterpart Notional correspondence Formal correspondence Frequency and combinatory correspondence suffix –ble/-bile/-vel suffix -telný yes yes ? gerund transgressive partial partial no ingressive verbal periphrases prefixes partial no ? hacer/fare/faire/ fazer + infinitive prefix rozpartial no no prefix re-/ri-no systemic correspondence 150 where no reliable quantitative analysis in terms of affix frequencies and their combinatorics could be made. However, the data also revealed an interesting frequency mismatch worth exploring in afuture study. In fact, we noted that the overall frequency of the prefix rewas considerably higher in French than in the other Romance languages. In the absence of afurther in-depth study, we can only guess that this might be due to the absence of aproductive iterative verbal periphrasis in French (as opposed to Spanish, Italian and Portuguese, where verbal periphrases of this type are commonly used, see Kratochvílová – Jindrová et al. this volume). Nevertheless, based on the corpus data, we can conclude that in terms of combinatory possibilities and overall frequency, there are considerable differences between hacer/fare/faire/fazer + infinitive on the one hand and the ingressive verbal periphrases and gerund on the other. While in Čermák – Kratochvílová et al. (this volume, Section 4.6), we observed that the causative construction displays similar combinatory possibilities in all four languages studied and the possible differences can be considered to be only isolated phenomena, the analyses made by Kratochvílová – Jindrová et al. (this volume, Section 5.2.4.1.2) and Nádvorníková et al. (this volume, Section 6.2.1) prove key structural differences in the usage of both the ingressive verbal periphrases and the gerund. These differences can be defined in terms of the lower general frequency of the phenomenon in question in Italian and French and its very high frequency of use in Spanish and Portuguese. While apparently unrelated, these observations might point towards adifferent behaviour of non-finite verbal forms in the analysed languages. Analyses presented in this monograph indicate aclose relationship between the usage of verbal periphrases and the gerund. As observed by Kratochvílová – Jindrová et al. (this volume), the set of French periphrastic constructions is considerably smaller than the Spanish and Portuguese one (Italian being in the middle between these two poles). The general preference for expressing the manner of action in forms other than periphrastic construction is closely related to the limited periphrastic usage of the Italian and especially the French gerund (see Nádvorníková et al. this volume, Section 6.2.1), thus influencing the general lower frequency of its usage (see Nádvorníková et al. this volume, Section 6.4). In terms of the semantic notions attributed to the phenomena in question, the largest differences can be observed in the case of ingressive periphrastic constructions, especially when referring to notions other than the mere beginning of an action. Unlike Italian and French, Spanish and Portuguese dispose of alarge set of stylistically marked verbal periphrases that underline notions such as [+unexpectancy], [+sheer energy], [+inappropriateness] etc. (see Kratochvílová – Jindrová this volume; Jindrová 2016, Kratochvílová – Jindrová 2017). However, an exhaustive comparison between Spanish and Portuguese proved to be impossible, especially due to the limited extension of the Portuguese subcorpus. Such acomparison requires aconsiderably larger set of data, as proven by Kratochvílová – Jindrová (2017), who analysed the differences between Spanish and Portuguese ingressive periphrases using the considerably larger CORPES XXI, CETEMPúblico, Araneum Hispanicum Maius and Araneum Portugallicum Maius corpora. 7. formal expressions vs abstract linguistic categories 151 7.2 czech respondents of the analysed phenomena vs systemic counterparts As can be observed in Table 7.1, the suffix -ble/-bile/-vel was the only analysed phenomenon with aclearly defined Czech counterpart (the suffix -telný), which apparently corresponds to the Romance element both in the form (suffix) and the semantic features (potential participation). While this respondent type, indeed, proved to be the most frequent (see Štichauer et al. this volume, Sections 3.4.2 and 3.6), it was used only in approximately 58% of all analysed translations (see Table 3.2). This suggests differences both in the combinatorics of the Czech and Romance suffix (asystemic comparison is impossible due to the limited amount of data contained in InterCorp) and, perhaps more importantly, in the behaviour of the abstract category of potential (non-volitional) participation in Romance and in Czech, its definition and formal manifestation. The question of how precisely it is possible to define semantic notions attributed to an affix becomes even more important when analysing the prefix re-/ri-, which lacks any clear Czech counterpart. The great heterogeneity of the Czech respondents and the dominance of respondent types where iterativity either resulted from alarger context or was apparently not expressed at all (see Štichauer et al. this volume, Sections 3.5.2 and 3.6) give rise to questions regarding not only the organisation of the iterativity category in Czech but also the combinations of iterativity with other notions in the matrix of the Romance prefix on one hand and the possible lexicalisation of aprefixed word, i.e. the semantic emptiness of the prefix, on the other. It is interesting to observe that very similar problems, i.e. the inherent presence of the notion attributed to the Romance phenomenon in the very semantics of the Czech respondent or in the meaning of an utterance as awhole rather than in aconcrete formal respondent and the combination of the notions traditionally attributed to the phenomenon in question with others that proved to be hard to define, also became an important topic when analysing the causative constructions. While the prefix roz-, which is generally considered to be aprototypical means of expressing causativity in Czech, proved to be rather amarginal respondent of hacer/fare/faire/fazer + infinitive, in approximately 60% of cases, causativity was not overtly expressed in the Czech respondent and resulted either from the meaning of averb or from syntax. The third most frequent Czech respondent is an analytic causative construction which, nevertheless, always includes secondary notions such as [+forcing], [+command], [+allowance] etc. (see Čermák – Kratochvílová et al. this volume, Section 4.7), which, on the contrary, were not overtly expressed in the Romance original and resulted from the context. However surprising, probably the greatest systemic similarities between Romance languages and Czech can be found when analysing the ingressive verbal periphrases and the Czech respondents. Prefixes, such as roz-, vyor za-, which are generally considered ingressive, were not the clearly dominant respondent type and analyses revealed that ingressivity in Czech is also systematically coded through verbal and verbo-nominal constructions, see Kratochvílová – Jindrová et al. (this volume). However, 152 the Czech respondents of concrete periphrases could generally express not only the beginning of an action but also secondary semantic notions, such as [+quick beginning], [+previous retention], [+energy] etc., which can also be attributed to Romance periphrases. The analyses revealed that both the Romance languages and Czech tend to express these notions cumulatively, both the Romance languages and Czech also dispose of aset of partially synonymous expressions that accentuate different facets of the beginning of aprocess. In the analysed Romance languages, these constructions display large formal similarities (the construction of asemi-auxiliary verb + infinitive); in Czech, the forms of expressing ingressive MoA are less coherent (prefixes and verbal or verbo-nominal constructions), nevertheless, the tendency to express the initial stage of aprocess through arelatively clearly defined set of productive linguistic features can also be observed. On the other hand, especially with regard to Czech, where ingressivity is traditionally associated solely with prefixes, the analyses clearly show that, just like in the case of causativity, iterativity and the expression of action carrier, identifying the analysis of the beginning of an action solely with one formal manifestation, clearly impedes us from viewing the category in its complexity. Finally, in light of the presented analyses, we can state that observations regarding the deep and complex nature of the category in question, which were made with reference to potential (non-volitional) participation, iterativity, causation and ingressivity, also apply when referring to the gerund. Despite the large systemic similarities between the Romance gerund and the Czech transgressive (see Nádvorníková et al. this volume), the stylistic features attributed to the Czech transgressive, such as “archaic” and “obsolete”, made this apparently ideal typological counterpart appear very rarely in the corpus (even in literary texts that constituted the main part of our data). If restricted to its non-periphrastic, i.e. converbal use, the Romance gerund shows striking semantic similarities in the four Romance languages under scrutiny in this study (see Section 7.1). If we base the definition of the main function of the gerund on its most frequent kind of usage, we can identify it with ahighly abstract notion of adverbial subordination, more concretely with the expression of accompanying circumstance. Being the most frequent Czech respondent of this type of gerund, acoordinate clause with the conjunction a(‘and’), it might seem tempting to conclude that Czech respondents do not reflect the main syntactic feature of gerund (i.e. subordination), thus changing the relationship between the main process (expressed through afinite verbal form in Romance) and its accompanying circumstance (expressed through the gerund). However, leaving aside the purely formal syntactic features of aRomance sentence with agerund and its most frequent Czech respondent, i.e. acoordinate clause, we can also observe that ais the most frequent and, consequently, also the most neutral, conjunction in the Czech language. Returning to the question of the semantic properties of the Romance prefix re-/riand its semantic non-transparency (possibly even emptiness) in some contexts, we might ask whether the conjunction adoes not serve in many contexts as aneutral way of connecting two verbal contents without overtly pointing out the hierarchical relationship between them. In this way, we can also conclude that the notion of accompanying circumstance, which is explicitly ex- 7. formal expressions vs abstract linguistic categories 153 pressed through the form of the gerund in Romance, is once again present in the very semantics of the Czech respondent or in the context. 7.3 exploiting the parallel corpus in search of language universals and abstract categories The analyses presented throughout this monograph and the results lead us to the conclusion that, even when counting on relatively small data-sets for all analysed languages (with the possible exception of Spanish), asystemic contrastive analysis of concrete language phenomena and their Czech respondents can offer interesting insights both when concentrating on the typology of Czech translations and when observing the semantic features of the Romance construction in question, in the light of its Czech respondents. All the presented analyses clearly demonstrate that notions attributed to the Romance phenomena under scrutiny are very common in language and, often, the Czech speaker does not even realise their presence (for example, in the case of inherent iteratives and causatives or in the case of the commonly used ingressive prefix za-). On the other hand, the analyses also reveal asimilar tendency in the case of the Romance phenomena we analysed. The large amount of non-transparent uses of the prefix re-/ri-, the combination of stylistically neutral ingressive verbal periphrases with averb that clearly expressed the beginning of an action in its internal MoA or the high frequency of the circumstantial gerund that was translated through aneutral coordinate construction, suggest that an overt expression of iterativity, ingressivity and adverbial circumstance is, actually, redundant in these cases and might be explained on the grounds that these forms are often lexicalised or considered aneutral form of expression rather than amarked emphasising of the above-mentioned notions. We consider the observed non-transparency of the analysed categories (both in Romance languages and in Czech) probably the most important general conclusion that can be drawn from our study. Parallel corpora proved to be auseful tool for revealing non-transparent, non-systematic or covert expressions of potential (non-volitional) participation, iterativity, causation, ingressivity and abstractly conceived adverbial circumstance. In this aspect, the presented analyses clearly shed new light on the nature of the categories the analysed phenomena express and on the organization of these categories both in Romance and in Czech, which proves to be much more complex, more abstract and less delimited by the formal manifestation than it is generally assumed. While our analyses concentrated on morphology and morphosyntax, the categories under scrutiny proved to also be connected on apurely semantic level (for example, in the case of inherent iteratives or causatives) and to hypersyntax and pragmatics (for example, in the case of causative and circumstantial relationship or the expression of the potential participation, which resulted from the context of the analysed utterance rather than from one concrete element).