Mesoclisis in Old Occitan

Mesoclisis in Old Occitan Xavier Bach (CNRS & Université Toulouse-Jean Jaurès) ABSTRACT This paper delves into the intriguing phenomenon of mesoclisis in Old Occitan. Mesoclisis, the insertion of an object clitic pronoun within the verbal form rather than before or after it, is explored particularly within the context of future and conditional tenses. While mesoclisis is evidenced in both the future and conditional tenses in Old Catalan and Old Spanish, in Occitan, it is exclusively observed in the future tense. Notably, all clitics may feature in this construction, with ahigher prevalence observed in the first person singular, especially with class Iverbs. Furthermore, there appears to be alexicalized association with verbs of speech, likely due to discourse-related factors. This investigation not only sheds light on the differential grammaticalization between the future and conditional tenses but also unveils an intermediate stage in their grammatical evolution, bridging the gap between Catalan and French linguistic contexts. KEYWORDS grammaticalization; future; mesoclisis; Occitan; Romance DOI https://doi.org/10.14712/18059635.2024.1.1 1 INTRODUCTION Mesoclisis is the fact of placing aclitic pronoun (object or indirect object) not before the verb (proclisis) or after the verb (enclisis), but right in the middle of the verb. It is sometimes said that the clitic is placed between the stem and the inflectional ending. Mesoclisis is attested in modern Romance languages only in Portuguese in futures (Crysmann, 1997; Luís & Spencer, 2005), and in Spanish varieties in the imperative (Harris & Halle, 2005; Manzini & Savoia, 2011), but it was also largely present in medieval Romance, including Portuguese, Spanish, and Catalan (Anipa, 2000; Batllori Dillet, 2011, 2012; Bouzouita, 2011; Graham, 2018). In these languages, scholars sometimes talk of an analytic future, because the future seems to appear in such constructions as formed by an infinitive and forms of the auxiliary ‘have’, contrasting with the usual, synthetic forms of the verb. Irefrain from using the term ‘analytic future’ here for reasons explained in section 3.2. In this article, Istudy asmall corpus of known attestations of mesoclisis in Old Occitan. Ianalyse data drawn primarily from the only available searchable electronic corpus for Old Occitan: COM2 (Ricketts, 2005). The corpus collates all verse literature written in Occitan from its origins to the end of the 15th century. As such, it includes the whole corpus of lyric poetry from the troubadours, as well as longer narrative and didactic poems, and theatre. There are obvious limitations with this corpus, some due to the choice of texts included, and some due to the technical apparatus of the corpus itself. The fact that it is restricted to verse can sometimes cast doubts on the reality OPEN ACCESS 8 LINGUISTICA PRAGENSIA 1/2024 of phenomena: were they induced by the necessity to rhyme, or by the fixed syllable length of Occitan verse? Fortunately, at least the second possible doubt can be eliminated in that all mesoclitic forms have the same length in syllables as would have aproclitic and asynthetic verb form. In addition, the majority of attestations are not placed in rhyme position. In order to ascertain that the phenomenon occurred in prose too, Ihave supplemented the corpus with attestations found in one of the translations of the New Testament in prose. The other limitations due to the type of texts are the relative absence of Gascon in the corpus, and the fact that anumber of troubadours included in the corpus wrote in Occitan as their second language, thus opening the door to possible interlinguistic interferences. The second type of limitation is more problematic, the corpus being asimple plain text searchable database, with absolutely no markup beyond proper names: it is impossible to search for agiven tense, or aword class, or even for inflected forms of the same word, only specific strings of texts are accepted (i.e. full words, and words starting or ending with aspecific string of letters, with the possibility of searching for two co-occurring words at some distance). This means that extracting acorpus of mesoclitic futures and conditionals is tricky: one has to search specific endings on one word (-ar, -ir, -re, -er), and pronominal forms co-occurring with it. The very many searches one has to perform result in alot of data noise that then needs to be removed manually. The first section briefly examines mesoclisis in other (medieval) Romance languages. Based on acorpus of more than 200 attestations, Istudy anumber of parameters for mesoclisis in Old Occitan. In asection about the morphology of mesoclisis, Ianalyse first the fact that it is truly acase of clitic placement, not an analytic construction, before ruling out the only three attestations of mesoclisis in the conditional as due to language contact. Ilook at what person and number combinations are attested, noting apre-eminence of 1sg. Ithen examine in turn the lexical verbs that enter this construction, and their inflection class structure. In the next section, devoted to the syntax of mesoclisis, Iexamine what pronouns enter into the construction, before analysing the role played by polarity and the dispreference for clitic first placement. Ialso consider possible cases of aformulaic use of mesoclisis. The last section discusses anumber of analytical results, before Idraw some conclusions. 2 MESOCLISIS IN MEDIEVAL ROMANCE Mesoclisis is attested in medieval Romance languages mostly in the languages of the Iberian Peninsula, including Spanish, Catalan, and Portuguese, in forms of both future and conditional tenses (Crysmann, 1997; Anipa, 2000; Batllori Dillet, 2011, 2012; Bouzouita, 2011; Enrique-Arias & Bouzouita, 2013). For those languages, researchers often use the term ‘analytic future’ for forms of futures with mesoclisis, because here the future seems to be an analytic construction with an infinitive and auxiliary ‘have’, including pronouns in between the two elements, in contrast to the more usual, synthetic future. Ishow below that this term, while it may be appropriate for Spanish, is probably not appropriate for Occitan. OPEN ACCESS XAVIER BACH 9 The alternation between synthetic and analytic forms of futures and conditionals has to do with the etymological origin of the future and conditional in Late Latin, as they formed from aperiphrasis involving alexical verb in the infinitive followed by the auxiliary habere in the present for the future, and in the imperfect for the conditional (see Esher, 2012 for areassessment). For the future, the original periphrasis is: (1) cantare habeo sing.inf have.1sg ‘Ihave to sing > Iwill sing’ The periphrasis originally had ameaning of obligation: Ihave to sing. This meaning implies some futurity of the event being reported, and from there the periphrasis developed into afully fledged future tense (future in the past for the conditional). The process of grammaticalization also implied amore rigid word order with the infinitive coming first, and the progressive phonological reduction of the finite auxiliary inducing aprocess of univerbation: from aconstruction initially made up of two words, it ends up as asingle word inflected for number and person of the subject, with adistinctive stem for the future and conditional. But univerbation was maybe not complete in all cases, in particular those involving aclitic complement when clitics could be placed right after the original infinitive. Cases of mesoclisis would thus be aconstruction that in some ways resisted the grammaticalization of these tenses, and that could have fossilized in an alternate, analytic construction. Mesoclisis has been well studied in medieval Ibero-Romance. There it seems to be arelatively frequent construction, compared to its relative rarity in Occitan sources. Thus, in medieval Spanish, in the sole text of Cantar de Myo Çid, 81.8% of futures are synthetic, and 18.2% are analytic, the proportion of analytic conditionals being slightly lower at 10.8% (Batllori Dillet, 2011, p.7). This construction, in both Old Spanish and Old Catalan, is not compatible with negation (Graham, 2018; Sentí & Bouzouita, 2022). In Catalan as well, mesoclisis is possible with both futures and conditionals. In Old Catalan at least, where the future and conditional seem to be more advanced in terms of grammaticalization than Old Spanish, forms with mesoclitics appear in the same syntactic contexts as forms with enclitics, mainly “i) those that appear with the verb in the first position (P1), ii) paratactic clauses, iii) after avocative, and iv) in coordinated adversative clauses after the conjunction mas or però ‘but’” (Sentí & Bouzouita, 2022, p.61). In the same contexts in Old Spanish, only mesoclitics, not enclitics, appear before the 15th century, which is the main argument that prompts Sentí & Bouzouita (2022) to consider that the future and conditional are less grammaticalized in Old Spanish than in Old Catalan. Ireturn to this point regarding the place of Occitan in acline of grammaticalization in section 5. Some of the contexts for appearance of mesoclisis in Old Catalan do not seem to be perfectly parallel in Old Occitan. In particular, an initial position of the verb in the clause does not seem to prompt mesoclisis in asystematic way as it does in Old Catalan (see section 4.3 below). OPEN ACCESS 10 LINGUISTICA PRAGENSIA 1/2024 3 MESOCLISIS IN OLD OCCITAN: MORPHOLOGY Mesoclisis is also attested in Old Occitan (Jensen, 1994, p.243). The first thing to note is the rather limited occurrence of mesoclisis: the COM2 corpus includes 214 examples, with only three in the conditional, to which one can add about twenty examples found in prose texts, all in the future. Forms with proclitics on the contrary are much more common, probably in the tens of thousands of examples in the COM2 corpus. It thus seems that mesoclisis is ararer phenomenon in Occitan than it is in Old Catalan and Old Spanish (compare, e.g., Batllori Dillet, 2011; Sentí & Bouzouita, 2022). This seems in line with acontinuum of use, noting that Old French has no example of mesoclisis in the documented period: Old Occitan is probably at an intermediate stage in the loss of the possibility of mesoclisis. 3.1 THE PROBLEM OF MESOCLITIC CONDITIONALS In Old Catalan and Old Spanish, mesoclisis is possible with both the future and the conditional. It seems though that mesoclisis is less common with conditionals at least in Old Spanish, although the number of attestations is still relatively high (Batllori Dillet, 2011). The COM2 data on Old Occitan show amarked difference with Catalan and Spanish in that mesoclisis with conditionals is vanishingly rare: out of 214 attestations of mesoclisis in COM2, only three are with conditionals. The three examples are shown in examples (2) to (4): (2) dar la m’ ia, don m’ es bel give 3fsg 1sg cond.3sg of.which 1sg be.3sg nice ‘he would give it to me, which fills me with joy’ (PC 319 003 031) (3) no· us plac per c’ aucir m’ en ia not 2pl please.3sg for that kill 1sg part cond.3sg ‘it does not please you, because it would kill me’ (PC 434a 050 066) (4) sentir n’ ia dolor feel part cond.3sg pain ‘he/she would feel pain from it’ (PRO 1686) All three attestations are with 3sg verb forms, and include aconditional desinence, not aform of auxiliary ‘have’. There may be aslight problem with these attestations. In medieval times, Occitan was alanguage of culture and literature that was used in Occitan speaking areas, but also more widely, particularly in the Iberian peninsula and Italy. This means that anumber of texts collated in the COM2 corpus come from authors whose first language was not Occitan. Interestingly, for all three attestations, the author of the text is known. The first author, Paulet de Marselha, spent considerable periods of time in Catalonia at the Aragon Court, and may have been influenced by Catalan. The other two authors, sometimes considered to be the same person, Cerveri de Girona and Guilhem de Cervera, are Catalan authors writing in Occitan. The fact is that mesoclisis with conditionals is widely attested in Catalan (Sentí & BouOPEN ACCESS XAVIER BACH 11 zouita, 2022). The conjunction of this fact with the life of the authors, and the very low number of attestations in Occitan, seems to point towards acontact phenomenon, where some authors transferred the possibilities of Catalan to their Occitan texts, although it may not have been possible to use mesoclisis with conditionals in Old Occitan. In this context, it seems better to consider that Old Occitan only allowed mesoclisis with the future tense.1 3.2 ATRUE MESOCLITIC The first question one can ask of the data is to know whether the forms show asynthetic future, which finds itself separated by the insertion of amesoclitic, or whether this is still somehow averbal periphrasis composed of an infinitive and the auxiliary ‘have’, which happens to place the clitic pronoun in front of the finite form. It proves difficult to determine which is the case. The first element of proof regards the form of the infinitive with respect to the future stem. In reflexes of the Romance first and fourth conjugations, the identity between the infinitive and the future stem is maintained. But this is not the case in anumber of verbs in the reflexes of second and third conjugations, where the stem of the future and conditional has changed (Esher, 2012). Ishow below in section 3.4 that reflexes of such verbs are not found with mesoclisis in Old Occitan, which could be an indication in favour of either possibility. The second element of proof, in favour of atrue mesoclitic, has to do with the form of the endings of the synthetic future tense as compared with auxiliary ‘have’. As the synthetic future in Romance developed from aperiphrasis including the auxiliary ‘have’ in the present and an infinitive (Esher, 2012), there is still robust identity between forms of the auxiliary in the present tense and future endings in persons of the singular as well as the 3pl (Esher, 2018). This identity is not present in the 1/2pl endings, where futures end in 1pl -em, 2pl -etz, instead of auxiliary form avèm and avètz. The quality of the vowel is different, with amid-closed vowel in the future desinence, but amid-open vowel in the auxiliary present tense, although this fact cannot be used as diagnostic for medieval texts where degree of opening is not noted for mid-vowels. Future tenses with mesoclisis in the 1/2pl are never found with form of the auxiliary ‘have’, but instead only with forms identical to the endings em (9 examples in the corpus) and etz (19 examples in the corpus), as shown in examples (5) and (6): (5) tornar nos n’ em, que ben es de sazo come.back 1pl part fut.1pl because well be.3sg of season ‘we will come back, because it is the season’ (DEB 1842) 1 The situation was different in Gascon, where, as noted by Felip Biu (p.c.), mesoclisis was perfectly possible with conditionals, and is attested in the Récits d’histoire sainte in Béarn. For the most part, the COM2 corpus lacks texts in Gascon. It is also possible that the greater proportion of enclisis in Gascon varieties may favour the occurrence of mesoclisis, including at alater period. Imust for now leave Gascon varieties for future inquiry. OPEN ACCESS 12 LINGUISTICA PRAGENSIA 1/2024 (6) e far n’ etz sotil polvereta (RAC 2374) and make part fut.2pl fine powder ‘and you will make afine-grained powder out of it’ The forms em and etz are attested in isolation in Old Occitan outside the realm of mesoclisis, but they are forms of the auxiliary ‘be’, not of auxiliary ‘have’. In order to consider that future tenses with mesoclisis in the 1/2pl are periphrases, one would have to posit an auxiliary alternation depending on person and number, which makes little sense considering the etymology of the forms. Ithus take it that examples in the 1/2pl instantiate astructure as shown in example (7), with amesoclitic in the middle of the synthetic verbal form, rather than the structure in (8) with aperiphrastic expression including auxiliary ‘have’: (7) STEM + CLITIC + DESINENCE (8) *INFINITIVE + CLITIC + AVER As is shown below, the fact that the stem is always chosen among those which maintain an identity with the infinitive could lead one to think otherwise, but evidence from endings is conclusive. This is amarked difference with anumber of attestations in Spanish, where the finite form of the construction indeed seems to be aform of the auxiliary ‘have’. 3.3 PERSON AND NUMBER All combinations of person and number are attested in the corpus, albeit with an interesting distribution. Examples 9-15 show all these possible combinations, with example 9 and 10 exhibiting a1sg form (in verse and in prose), example 11 a2sg, example 12 a3sg and examples 13–15 displaying the plural persons. (9) Ez ieu dir vos ai la fazon and 1sg say 2pl fut.1sg the way ‘and Iwill tell you how it is’ (JAU 10555) (10) “… far vos ei pescadors d’ omes” do 2pl fut.1sg fishermen of men ‘…and Iwill make you fishers of men’ (Mt 4,19) (11) dis Jaufre, “laissar m’ as dormir? say.3sg J leave 1sg fut.2sg sleep ‘Jaufre says: “will you let me sleep?”’ (JAU 3384) (12) e laissar m’ a tot son estar and leave 1sg fut.3sg all his fortune ‘and he will leave me all his fortune’ (FLA 3380) OPEN ACCESS XAVIER BACH 13 (13) e poish tornar nos n’ em per aquel eish sender and then go.back 1pl part fut.1pl by that awful track ‘and then we will go back along that awful track’ (CCA 200 145) (14) e vos dar n’ etz a cels qu’ en auran and 2pl give part fut.2pl to those that part have.fut.3pl gran mester big need ‘And you will give some of it to those that will have the greatest need of it’ (CCA 200 122) (15) vencutz si er qu’ aucir m’ an li sospire defeated thus be.fut.1sg because kill 1sg fut.3pl the sighs ‘Thus Iwill be defeated because sighs will kill me’ (PC 155 022 013) Table 1 compiles the number of attestations of mesoclisis for each person and number combination, and indicates the percentage of total attestations for which each combination accounts: Person Number % 1sg 113 53.55 2sg 8 3.79 3sg 41 19.43 1pl 9 4.27 2pl 19 9.00 3pl 21 9.95 TOTAL 211 100.00 Table 1.Person and number of future forms with mesoclisis It is very clear from these figures that the majority of examples occur in the 1sg (53.55%). The next main use is in the 3sg, which is more expected considering the narrative nature of alarge number of texts in the corpus. The main use as 1sg may have to do with the lyric nature of anumber of texts, but only 39 out of 113 attestations in 1sg are from lyric texts (i.e. 34%). The 1sg is more common in lyric texts, as it forms 63% of examples in such texts in the corpus of mesoclisis. But this is not enough to explain the whole skewing of the data. On the contrary, the skewing points towards discourse oriented factors in the use of mesoclisis. 3.4 INFLECTION CLASS Another factor that can be examined is the inflection class of the verb entering into afuture construction with mesoclisis. Table 2 synthesizes the findings in this respect, with the total number of examples for each Romance class, as well as the number of different lexemes for each class: OPEN ACCESS 14 LINGUISTICA PRAGENSIA 1/2024 Class N Lexemes Class I 132 60 Class IV 20 10 Class III 59 6 Table 2.Inflection class of verbs in amesoclitic construction in the future tense in COM2 Unsurprisingly, amajority of examples belong to the first, productive, larger class of verbs, for which there is also no stem alternation in the future (Esher, 2012). One can group together verbs of the first and fourth classes, as neither present stem alternations in the future. For both these classes, the mean number of occurrences for an individual lexeme is around 2. This means that mesoclisis is attested for alarge number of lexemes in these classes. The situation is markedly different for verbs of the third class, where examples are limited to six verbs, of which three are attested only once. The bulk of examples involve the verbs far ‘do, make’ (28 times), dir ‘say’ (24 times, including one in the form dizer), and marginally creisser ‘grow’ (3 times). These verbs present an identical characteristic, in that the stem used to form the future and conditional is always identical to the infinitive. That seems indicative of an additional restriction in the use of mesoclisis with futures: only verbs whose future and conditional stem is identical to the infinitive can enter the construction. This makes it rather different from Ibero-Romance equivalents. It also automatically rules out the possibility of mesoclisis being present in compound tenses (anterior future for instance), as auxiliaries do not present afuture stem that is identical to the infinitive. There is one possible counterexample to this rule, in an early text, for the mesoclitic future of the verb dir ‘say’ in the Canso de Santa Fe, where it has the form dizer, as shown in example (16): (16) Dizer vos ei d’ aqelz pagans say 2pl fut.1sg of those pagans ‘and Iwill tell you about those pagans’ (CSF 110) If this is indeed an example of the verb dir, it would be an example of adifferent future stem, which would make up forms of the type *dizerei. Unfortunately, there are no synthetic forms of futures or conditionals for this verb in this text, but there are two attestations of the infinitive, including one as acomplement to amodal voler ‘want’, as shown in examples (17)–(18): (17) Dizer vos voill, enant qe· m paus say 2pl want.1sg before that 1sg.refl stop.1sg ‘Iwant to tell you, before Istop…’ (CSF 454) (18) Ella parled, e saub l’ o dir she speak.pst.3sg and know.pst.3sg 3sg 3nsg say ‘She spoke, and she knew how to tell it’ (CSF 230) OPEN ACCESS XAVIER BACH 15 There is thus, in this early text, some variation as to the form of the infinitive, which could be exploited for metric purposes in these cases. Only one other text in COM2 attests aform of the infinitive in dizer, in the Disticha Catonis (examples 19–21): in this text, there are no infinitive forms of the type dir, and the future of this verb is always built on the form dizer. (19) se ll’ aus dizer bon sen if 3sg hear.1sg say good sense ‘if Ihear him/her speak good sense’ (DIS2T 478) (20) ni ja no· n dizera and.not already not part say.fut.3sg ‘and already he/she will not say anything of it’ (DIS2M 021) (21) no· t dizera de non not 2sg say.fut.3sg of no ‘he/she will not say no to you’ (DIS2T 668) It seems relatively safe to conclude that only verbs with afuture stem identical to the infinitive can enter the mesoclitic construction in Old Occitan. 4 MESOCLISIS IN OLD OCCITAN: SYNTAX Regarding the syntax of mesoclisis, there are three factors to take into account. The first is the type of clitic that can enter into the construction (for which all but one clitic type is attested), and the fact that clitic combinations can also be found. The second one has to do with polarity: there are no examples of negation together with mesoclisis in the future. Finally, it has been suggested that mesoclisis could derive from asyntactic ban on clitics in the first position of the clause, for which evidence has to be examined outside of the mesoclitic construction, namely from cases of proclisis. Iexamine these in turn in the following subsections. 4.1 CLITICS AND GROUPS OF CLITICS Concerning the clitic pronouns which can appear in the mesoclitic construction, there does not seem to be any difference with proclitics: all possible pronouns are attested with mesoclisis, except one, as shown in Table 3, together with the number of occurrences for each. In the majority of cases the pronoun is athird person pronoun. But, as expected from the fact that verb forms are mainly in the 1sg, second person pronouns are also well represented in the corpus (15 for 2sg, 50 for 2pl), clearly taking part in asystem of address from the writer to the reader/listener. Examples (22) to (25) show examples of pronouns in the first and second person in both the singular and the plural. OPEN ACCESS 22 LINGUISTICA PRAGENSIA 1/2024 tional observed in some Occitan varieties (Esher 2012, 2015) might find explanation in such adifferential grammaticalization process. In most (Western) Romance languages, there is asystematic identity between the stem used in the future and the stem used in the conditional, as is the case for example in French (see Esher, 2012). In anumber of varieties of Occitan this identity is broken, in particular by sound change, but asound change that is not general in these varieties, but only limited to the conditional tense. This is the case for r-less conditionals in Occitan: varieties where the future keeps the etymological /r/ of the infinitive but the conditional does not (Esher, 2015). It might be going too far, but such examples are reminiscent of how, in terms of form, further grammaticalization often involves further phonological reduction, of atype that is restricted to the construction at hand and not ageneral sound change in the language. 6 CONCLUSIONS In this article Iestablished anumber of important and novel facts about mesoclisis in Old Occitan, which in turn illuminate the pace of grammaticalization of the future and conditional in Occitan relative to each other, and to future and conditional forms in other Romance languages. The first specific finding is that mesoclisis is only attested in the future, if one excludes marginal cases probably due to the influence of Catalan: in Old Occitan, the conditional seems to grammaticalize more quickly and more fully than does the future, and does not allow mesoclisis. Regarding the morphology of mesoclisis, Ishow that in Old Occitan, the structure is atrue case of mesoclisis, with aclitic fracturing the inflected form, and not an analytical construction involving an infinitive and an auxiliary with the clitic proclitic to the auxiliary (or enclitic to the infinitive), as seems to be the case in Old Spanish for example. The construction is limited to those verbs where the stem of the future is identical to the infinitive. This causes acertain skewing in the inflectional classes that can participate in the construction, mainly the first, open class which does not present stem alternations in the future and conditional, and avery restricted number of verbs in class III. There is aclear bias in favour of the 1sg for this construction: more than half of the corpus attestations are examples of a1sg future. This bias is only partially due to the type of text (lyric poetry), and mostly has to be attributed to one of the possible pragmatic uses of the construction, particularly in narrative: as ameans for the narrator to address the reader/hearer directly, outside of the narration. Regarding syntax, all types of clitic complements are shown to be possible in mesoclitic constructions, except the neutral object clitic o, although it is still unclear whether this is an effect of the small size of the corpus, or whether it is motivated by alinguistic principle. As in Old Spanish and Old Catalan, the construction is never attested with the negative particle non. Iidentify attraction of clitics by negation in Old Occitan as apossible explanation, which will be explored in further work. Ialso critically examine and ultimately reject ahypothesis attributing mesoclisis to the ban on clitics appearing as proclitic with verb initial sentences: although this hypothesis OPEN ACCESS XAVIER BACH 23 partially explains the data in Catalan (Sentí & Bouzouita, 2022), it is shown to be invalid for the Occitan data. The corpus of mesoclisis studied here shows aclear asymmetry in the pace of grammaticalization of the future and the conditional, confirming other studies on these two screeves, and is apossible explanation for the occurrence in modern varieties of stem asymmetries between the future and the conditional, due to amore advanced grammaticalization of the latter. In acomparative view, and for this sole feature, Occitan is shown to be in between Catalan and French in terms of the pace of grammaticalization of the future and conditional, refining the grammaticalization cline suggested in Lamiroy & Pineda (2017) by adding aposition for Occitan. Further analysis of the features used in such studies (auxiliaries, past tenses, mood, demonstratives, etc.) will provide additional confirmation of the position of Occitan along the Romance continuum. REFERENCES Anipa, K. (2000). Astudy of the analytic future / conditional in Golden-Age Spanish. Bulletin of Hispanic Studies, 77(3), 325–338. Bach, X. (2023). Negation in contact: French and Occitan. Transactions of the Philological Society, 120(3), 444–459. Batllori Dillet, M. (2011). Mesoclitic Romance futures and conditionals vs. participle preposing [Conference presentation]. 13th International Diachronic Generative Syntax Conference, Philadelphia. Batllori Dillet, M. (2012). Evidentiality in Old Spanish and Old Catalan: Mesoclitic Configurations [Conference presentation]. The 2nd Cambridge Colloquium on the Histories of the Ibero-Romance Languages— Norman MacColl Symposium 2012, University of Cambridge, Cambridge. Bouzouita, M. (2011). Future constructions in Medieval Spanish: mesoclisis uncovered. In R.Kempson, E.Gregoromichelaki, & C.Howes (Eds.), The dynamics of lexical interfaces (pp.89–130). Stanford: CSLI. Buridant, C. (2019). Grammaire du français médiéval. Strasbourg: SLR/Eliphi. Crysmann, B. (1997). Cliticization in European Portuguese using parallel morpho-syntactic constraints. In M.Butt & T.Holloway King (Eds.), Proceedings of the LFG97 Conference (pp.1–15). Stanford: CSLI. De Mulder, W., & Lamiroy, B. (2012). Gradualness in grammaticalization in Romance. The position of French, Spanish and Italian. In K.Davidse, T.Breban, L.Brems, & T.Mortelmans (Eds.), Grammaticalization and language change: New reflections (pp.199–226). Amsterdam: Benjamins. Enrique-Arias, A., & Bouzouita, M. (2013). La frecuencia textual en la evolución histórica de los clíticos pronominals en español. Iberoromania, 77, 29–46. Esher, L. (2012). Future, conditional and autonomous morphology in Occitan [Unpublished doctoral dissertation]. University of Oxford. Esher, L. (2015). Formal asymmetries between the Romance synthetic future and conditional in the Occitan varieties of the Western Languedoc. Transactions of the Philological Society, 113(2), 249–270. Esher, L. (2018). Implicational relationships between desinences in Occitan imperfect and conditional forms. Lexique, 23, 9–32. Esher, L. (2022). The historical development of imperfect indicative and conditional inflection in Pyrenean Romance. Transactions of the Philological Society, 120(2), 218–245. Graham, L. (2018). An analysis of morphosyntactic variation in the Old Spanish OPEN ACCESS 24 LINGUISTICA PRAGENSIA 1/2024 future and conditional. Journal of Historical Linguistics, 8(2), 192–229. Harris, J., & Halle, M. (2005). Unexpected plural inflections in Spanish : reduplication and metathesis. Linguistic Inquiry, 36(2), 195–222. Jensen, F. (1994). Syntaxe de l’ancien occitan. Tübingen: Niemeyer. Lamiroy, B., & Pineda, A. (2017). Grammaticalization across Romance languages and the pace of language change. The position of Catalan. Lingvisticae Investigationes, 40(2), 304–331. Lema, J., & Rivero, M.L. (1991). Types of verbal movement in Old Spanish: modals, futures and perfects. Probus, 3, 237–278. Luís, A., & Spencer, A. (2005). Aparadigm function account of ‘mesoclisis’ in European Portuguese. In G.Booij & J. v. Marle (Eds.), Yearbook of morphology 2004 (pp.177–228). Dordrecht: Kluwer. Maiden, M. (2018). The Romance Verb: Morphomic structure and diachrony. Oxford: Oxford University Press. Manzini, R., & Savoia, L. (2011). Mesoclisis in the imperative : phonology, morphology or syntax? Lingua, 121(6), 1101–1120. Ricketts, P.T. (2005). Concordance de l’Occitan médiéval (COM2). Turnhout: Brepols. Sentí, A., & Bouzouita, M. (2022). The Syntax of Old Catalan Clitics: Llibre dels Fets. Catalan Journal of Linguistics, 21, 47–77. Xavier Bach CLLE UMR 5263, CNRS & Université Toulouse-Jean Jaurès Maison de la recherche 5 allées Antonio Machado 31058 Toulouse Cedex 9 (France) ORCID: 0000-0002-4389-4871 [email protected] OPEN ACCESS