Contemporary Croatian future II: a tense or a mood?

Summary: This paper seeks to determine the place of contemporary standard Croatian future II in the verbal system and to define its meaning more accurately. What applies to Standard Croatian is assumed to apply to the majority of Š tokavian and Č akavian dialects as well. While the primary focus of this work is modern future II, some of its older meanings are mentioned briefly in Section 4. Modern Croatian future II is peculiar in several ways, due in part to its unclear place in the verbal system: some linguists treat it as a tense (i. e. as an indicative tense), others as a mood (i. e. as a non-indicative tense). This paper attempts to resolve that disagreement by comparing future II with future I in similar contexts. The difference bet-ween the two future forms is difficult to interpret as a difference between categories but, in any case, it is modal rather than temporal. In that sense, future II functions as an ‘ oblique ’ mood in relation to future I, which belongs to the indicative. Whereas future I refers to a future process anchored in the present, future II refers to one that is not. The anchoredness of a future process in the present is understood chiefly as a relatively concrete concept: one that increases the conviction that a process will be realised. The non-indicativity of future II is also corroborated by the situation in languages that use the subjunctive to express the future II meanings. Future II as an oblique mood, whether formed using the l-participle or the infinitive, is a logical consequence of its development from earlier indicative forms.


Introduction
The main aim of this paper is a better understanding of the use and meaning of future II, and thus a more accurate categorisation of future II as a verb form.The topic is Standard Croatian (and, generally, Štokavian and Čakavian) future II in its modern meaning.Its earlier standard and some dialectal meanings, although they are not the main topic, will be addressed briefly in Section 4. The term future II is also used to describe certain non-Slavic verb forms and some Slavic forms that do not belong to (Modern) Štokavian or Čakavian.The question posed in the title of this paper does not apply to such forms. 1 Future II is a fairly unusual form in Croatian, with its syntactic restrictedness being perhaps its most notable characteristic.Namely, unlike the other finite verb forms, future II is used almost exclusively in subordinate clauses, and only in certain types.In brief, it can be said that future II occurs in conditional 2 and restrictive (Milošević 1970: 128) relative clauses, whereby the term relative clause is considered also to include adverbial clauses of manner, quantity, place, and time. 3When the subordinate clause contains future II, the main clause nearly always4 expresses a future situation.The main clause normally contains a future I form, but the imperative, present, perfect or aorist with a future meaning can also occur, and the verb can even be omitted.The use of future II in certain other types of subordinate clauses is rare and atypical (cf.Milošević 1970: 119-121).In independent clauses, future II occurs almost only with the word možda 'perhaps' in Standard Croatian (Kovačević 2009).
Future II is also peculiar morphologically.In Standard Croatian, the future II active is composed of the thematic 5 present of the verb biti 'be' and the l-participle (e. g. budem činio, do-1 1SG SG. .FUT FUT2 2).In certain dialects, it can be composed of the thematic present of the verb biti and the infinitive (e. g. budem činiti).Future II is normally imperfective and, in many dialects, it is exclusively or almost exclusively imperfective. 6This applies, to various degrees, to the entire Štokavian and Čakavian area (cf.e. g.Vuković 2014: 82-83;Grickat 1956: 90, 96;Kravar 1978: 23;Stevanović 1989: 654;Riđanović 2012: 323;Čirgić et al. 2010: 127).Since the meaning of modern perfective future II can be conveyed by the perfective present (which also has various other meanings), the latter is normally used instead of perfective future II.When discussing the meaning of future II in the remainder of this paper, I usually also refer to the meaning of the perfective present in the corresponding usage, although I do not indicate that specially.The imperfective present active, however, cannot replace the imperfective future II active.
The sole exception from the rule mentioned in the previous paragraph is the verb biti 'be'.Namely, its future II form (budem bio) is quite uncommon and the function of both perfective and imperfective future II of this verb (along with some other, present functions)7 is usually performed by the biaspectual thematic present budem.Due to this semantic exceptionality, the form budem can be considered to be both a present and a future II form. 8What has been said about the present budem also applies to the participial passive formed from it.Namely, both the passive forms of perfective verbs, such as bude sagrađen (build-3 3SG SG. .PFV PFV..PASS PASS), and the passive forms of imperfective verbs, such as bude građen (build-3 3SG SG. .IPF IPF..PASS PASS), function as both the present and future II.Whereas perfective passive forms such as bude sagrađen can in fact be considered to be present forms when occurring in contexts typical of future II, because the perfective present active also functions as future II, there is probably no point in considering imperfective forms such as bude građen that occur in the same contexts to be present forms.The form budem is described by all grammars as a present form only. Passive forms like bude (sa)građen, however, are considered to be future II by most grammars.There are two exceptions to this generalization: Babić et al. (1991: 719), where forms such as bude građen are considered to be both present and future II, and Tanasić (2005: 452-453), where such forms are considered to be present, and only the less common forms such as bude bio (sa)građen are interpreted as future II.This phenomenon was noticed and ap-propriately described more than half a century ago by Milošević (1971).This issue has also been discussed by Benić & Moretti (2020: 197-198).
Besides being peculiar syntactically and morphologically, future II is also unusual semantically.The meaning of this verb form is fairly elusive, so it is categorised as a tense (i.e. as an indicative tense) by some linguists and as a mood (i.e. as a non-indicative tense) by others.An attempt to ascertain a solution to the problems associated with the meaning of future II will be made in the rest of this paper.
2 Future II as a problematic category Before addressing the main topic of this article, a number of other facts, mainly related to the temporal value of future II, need to be clarified.Croatian future II is a problematic category chiefly due to its peculiar meaning.This is also implied by the fact that several terms are used in Croatian to describe it.The most common ones are: futur drugi (the second future), futur egzaktni (based on Latin futurum exactum 'completed future') and predbuduće vrijeme (the anterior future tense).Occasionally, other terms are also suggested (see Sections 3.1 and 4).From those available, the most appropriate neutral label for forms such as budem činio is futur drugi (rendered here as future II).Other Croatian variants of this term, such as drugo buduće vrijeme (the second future tense) are also suitable, albeit more awkward.The labels futur egzaktni and predbuduće vrijeme, however, are misnomers.
The term futur egzaktni, specified (usually as a secondary term) by a number of grammars, originated from Latin grammar.Namely, Latin verb tenses were divided originally into tempora actionis imperfectae (imperfective action tenses) and tempora actionis perfectae sive exactae (perfective or completed action tenses), cf.Kühner & Holzweissig (1989: 653).The former are the tenses formed from the present stem, which correspond to the Croatian imperfective aspect to some extent, and the latter are those derived from the perfect stem, which correspond partially to the Croatian perfective aspect.In accordance with this, the term futur egzaktni would mean more or less the same thing as perfective future.Latin future II (i.e. futurum exactum or future perfect) is formed from the perfect stem and is perfective, i. e. it generally corresponds to the Croatian perfective present used in contexts typical of future II, whilst Croatian future II usually corresponds to Latin future I.As mentioned before, Croatian future II is mostly imperfective, and the label futur egzaktni, which implies the opposite, should be avoided.
Apart from being perfective, Latin future II also expresses anteriority.Probably because of this trait of Latin future II, Croatian future II is sometimes incorrectly described as an anterior future tense (predbuduće vrijeme) too.Future II is defined as a form that expresses imperfective anterior future meanings, for example, by Silić & Pranjković (2000: 92).The same grammar (p.194) claims that future II is always relative because it denotes a future action that precedes another future action or is (less commonly) simultaneous with it.As a predominantly imperfective form, however, future II is very frequently, and perhaps mainly, simultaneous, and can also be anterior or posterior, as will be demonstrated later.Menac-Mihalić (1989: 101) applied the term predbuduće vrijeme (anterior future tense) to the dialectal form occurring in the main clause in the following example from the dialect of Murter.
(1) Kad se onȁ vrãti, o n će vȅć bȉti lȅga.return-3 3SG SG. .PRS PRS..PFV PFV lie down-3 3SG SG. .FUT FUT..PRF PRF..PFV PFV 'When she returns, he'll already have gone to bed.'This is the Romance-type anterior future, used predominantly in the main clause in dialectal croatian.The label anterior future tense is more suitable for this form than for future II.One, however, has to be careful when applying particular terms to various forms with a future component because there is a fair number of such forms which all have similar meanings.Cf. the following short list: 1. Croatian future II (characteristic of relative and conditional clauses; predominantly imperfective, not a relative tense); 2. Latin future II or futurum exactum (characteristic of relative and conditional clauses; perfective/anterior); 3. the Romance-type anterior future or future perfect (in dialectal Croatian9 characteristic of mainclauses; denoting an anterior future action or the simultaneous result of an action); 4. the deferred-cognition evidential (characteristic of if-clauses; referring to a future realisation that a process occurred in the past, e. g.OCS ieda bǫdetъ sъlъgalъ 'in case [he] will have lied' from Whaley (2000: 110) 10 ; 5. the future-in-the-past (a process occurring in the future relative to a reference point that is in the past). 11 Although Croatian future II is still sometimes referred to as the anterior future tense or is considered to be that, the majority of linguists and grammarians are well-aware of the fact that it is very often simultaneous.Croatian grammars, however, do not note that future II can also be posterior.This is unusual, given the fact that the posterior use of future II has been mentioned in a number of academic works (starting from Grickat 1956: 90-91) and in some Serb(o-Croat)ian grammars (cf.Vuković 2014: 88).The temporal meaning of future II can be explained briefly as follows.
Firstly, although it occurs mainly in subordinate clauses, future II is predominantly an absolute tense.Moreover, it is no more relative than future I.The following examples demonstrate that the reference point from which the temporal meaning of future II is determined is the time of utterance, rather than the time expressed by the main clause.Namely, while future II refers, in all these examples, to a process that lies in the future with respect to the moment of utterance, its relation to the tense of the main clause differs in various examples: it is anterior in (2), simultaneous in (3), and posterior in (4) and ( 5).
(2) Ako danas budem radio tri sata duže, sutra ću se cijeli dan odmarati.Secondly, imperfective future II may be used in the subordinate clause to express a process simultaneous with that of the main clause, as well as one that is anterior or posterior to it.Perfective future II and the corresponding present, however, are used mainly to express a process that is either anterior or posterior to that of the main clause.Although simultaneity is considerably more typical of the imperfective aspect, it can be expressed marginally by the perfective aspect too, as in ( 7) and (8).In (7), the processes are not simultaneous but they are perceived as such.In (8), however, the processes are definitely simultaneous: as the fruit loses water, it becomes shrivelled.While, as noted, future II has long been recognised as capable of expressing posteriority, several issues concerning its posterior usage remain unresolved.Namely, posterior future II has been exemplified in the literature only by temporal clauses. 12ntrigued by that, Vuković (2014) analysed examples from the Croatian National Corpus (Hrvatski nacionalni korpus) and reached to the following conclusions.In temporal clauses, future I does not occur, so future II can express an anterior, simultaneous or posterior action, i. e. future II does not have a relative temporal meaning in such clauses.In conditional clauses, however, both future II and future I can occur.In this regard, according to Vuković (2014: 89), there exists a purely temporal complementary distribution between future II and future I in conditional clauses, whereby future II indicates anteriority and simultaneity, while future I indicates posteriority.Although Vuković was familiar with works that identified the existence of more refined semantic differences between the two future forms, 13 he considered his own explanation both simpler and more plausible.And yet, this explanation is simply incorrect.The fact that the difference between future I and future II in ifclauses does not lie in the sequence of tenses, is demonstrated clearly by the follo-wing examples.In (9), future I indicates an anterior process, and in (10) Such temporal relations are rarer than those noted by Vuković, but not at all unusual.The claim that Vuković makes about the future forms in more or less typical relative clauses is similar to his claim about the future forms in conditional clauses (Vuković 2014: 90-91).This claim, however, applies to relative clauses even less than it does to conditionals.The posteriority of future II in such clauses is confirmed by (5) above.However, the anterior use of future I in typical relative clauses, as well as the fact that the difference between the two future forms is not temporal in essence 14 , is illustrated by ( 11) and ( 12), both from Grickat (1956: 91) Obviously, there is no temporal difference between the two future forms in these examples, nor even a stylistic and/or aspectual distinction such as the one between the Standard Croatian perfect, aorist, and imperfect tenses. 15Namely, they rather seem to differ in mood, and if that is true, it makes no sense to consider future II an indicative tense.
3 Future II as a mood In the preceding sections, future II has been defined as a finite verb form that refers to absolute future time and does not have a relative future meaning, although it occurs almost exclusively in subordinate clauses.Another question that needs answering is whether this is an indicative form or not, i. e. to put it simply, is it a tense (i.e. an indicative tense) or a mood (i.e. a non-indicative tense).In this paper, future II is considered to be a non-indicative mood.The difference between indicative future I and non-indicative future II is that future I refers to a future process anchored in the present, whilst future II refers to one that is not.The anchoredness of a future process in the present is understood chiefly as a relatively concrete concept: one that increases the conviction that a process will be realised.The difference between future I and II will be addressed in greater detail in the following sections.The thesis that future II is a mood is supported by the interpretations offered by certain earlier linguists, as well as by a more meticulous analysis of the relationship between future II and future I in similar contexts, and finally, by some typological parallels.

The attitudes of other linguists
Although it is not dominant today, the interpretation of future II as a mood has a relatively long tradition.Peti (1985: 114) says that future II has been included amongst verb moods by some, but unfortunately, he mentions no names.I will list here some of these authors, who I have managed to identify without a detailed scrutiny of the bibliography.
In his grammar -Cassius ([1604]1977: 95, 99) -Bartol Kašić mentions the future indicative (future I) and future subjunctive (Croatian future II formed using the lparticiple).I am not aware, however, whether the interpretation of future II as subjunctive is Kašić's original idea or, which is more likely, it might be that Latin future II was referred to as future subjunctive in the Latin grammars used by Kašić, so he primarily temporal.The exceptionally complex relationship between these verb forms, however, is not the topic of this article.
Contemporary Croatian future II applied the same term to the 'Illyrian' future II. 16Veber (1871: 71-73, 169) has the potential mood, which includes the perfect, present, and future tenses.The perfect potential is Modern Croatian conditional II, the present potential is conditional I, and the future potential is imperfective future II and the corresponding17 perfective present.While such a division appears systematic, it cannot be justified because the use of the conditional differs considerably from that of future II.
In Stevanović (1957-58: 39-40), future II is considered to be a mood.What is normally labelled the conditional in Croatian studies, is referred to as the potential by Stevanovićas well as by some other Serbian authorswhilst, in his opinion, the term conditional should instead be used to refer to future II.This attitude was maintained until the last edition of Stevanović's grammar (Stevanović 1989: 693-694) although, for tradition's sake, he still used the term future II and discussed this form in the section dedicated to tense.Stevanović explains the need to include future II amongst moods by the fact that, in subordinate clauses, future II indicates an unrealised action, whereas in independent clauses, it expresses doubt or assumption.Such an explanation is not particularly convincing, because every typical future, and not just future II, indicates a process that has not been realised at the moment of utterance.The words doubt and assumption probably do not describe accurately the meaning of future II in independent clauses either, but the fact is that future II occurs in independent clauses that involve a neutrally potential18 process, whilst future I clauses involve processes that are generally probable, i. e. it is assumed that they will be realised.Milošević (1970: 152) also considers future II to be a tense, although she does not reject the validity of certain reasons for interpreting it as a modal form.Future II is considered to be a mood and is discussed in the section dedicated to mood by Stanojčić & Popović (1992: 92, 387, 389-390).

The contrast to future I
The non-indicative meaning of future II, and its meaning in general, is most obvious when future II is contrasted with future I in similar contexts, chiefly conditional and relative clauses.The differences between the two future forms in conditional and relative clauses have been addressed, as has already been mentioned, in earlier works.Comparing the examples adduced earlier in this paper as ( 11) and ( 12), Grickat (1956: 91) claims that future I expresses a higher degree of certainty that a process will be realised than future II does in the same context.Additionally, she notes that the noun čovek 'man' is referential with future I in ( 11) and non-referential with future II in (12).Milošević (1970: 118) says the following about the difference between the two future forms: "Ako je u U rečenici futur I, onda je uslov budućoj radnji sadašnja uvjerenost subjektiva o (iz)vršenju buduće radnje (sam stav sigurnosti prema realizovanju buduće radnje koji je izvor indikativnog značenja futura I-og); a ako je u rečenici futur II, onda je buduće realizovanje uslov glavnoj radnji."[If the if-clause contains future I, the condition for the future action is the present conviction of the subjective19 that this future action will be realised (the very certainty that the future action will be realised which is the source of the indicative meaning of future I); and if the clause contains future II, the future realisation [of the action] is the condition for the main action.]Although these works do not determine the differences precisely and contain unresolved issues, they are definitely on the right track.

Conditional clauses
The difference between the two future forms when they occur in conditional clauses is most easily interpreted, on one hand, as a difference between factual and hypothetical conditional clauses and, on the other, as a difference between various types of hypothetical conditional clauses.
The difference between factual and hypothetical conditional clauses is illustrated by the following examples, in which (13) demonstrates a factual conditional clause and ( 14) a hypothetical conditional.The classification of conditional clauses into these two types is based on their meaning, cf.Bhatt & Pancheva (2006: 671-672) and Pekelis (2017: § 3.1).Hypothetical clauses are the most common type of conditionals.Various classifications of hypothetical conditional clauses according to the probability of the condition are found in the literature, and they often contradict each other.Although a more detailed discussion of the classification of hypothetical conditionals would take us too far from the topic of this paper, it is worth mentioning that conditional clauses with future II, such as ( 14), have a potential meaning and that they express neutral potentiality. 20Factual conditionals can be regarded as a marginal type of conditional clauses.Factual conditionals imply that a non-speaker considers their proposition 21 to be true or at least claims that it is true, whereas hypothetical conditionals do not involve such an implication.In other words, in ( 14), it is completely unknown whether it will rain on Friday or not, whilst (13) implies that someone claims that it will rain on Friday (the addressee or, more likely in this particular case, meteorologists).In factual conditional clauses, future II cannot occur and future I can.In hypothetical conditional clauses, however, both future forms can occur.
The difference between the two types of hypothetical conditional clauses is illustrated by the following examples.20 The traditional term used in Croatian linguistics to describe such clauses is realne (real); however, they do not express 'reality', because it is unknown whether the condition will be realised or not, i. e. it is equally probable that it will be realised and that it will not.The terms potential, dubitative and counterfactual conditionals are more accurate than the traditional terms, respectively, real, potential and irreal conditionals.Most conditional clauses express irreality.Hypothetical conditionals that could be considered to express reality are, actually, borderline cases between conditional and temporal clauses.For more or less similar attitudes, cf.Švedova (1980: § 3000), Annear Thompson et al. (2007: 255-256), Pekelis (2017: § 3.1.1.).21 I.e. the idea that would be obtained if the protasis were transformed into an independent clause, e. g.U petak će padati kiša 'It will rain on Friday' in the case of ( 13) and ( 14).
The conditional clause in (15) can be interpreted as both factual and hypothetical.The factual interpretation presupposes that the addressees (whether singular or plural) claim that they will drink coffee.In the hypothetical interpretation of the same clause, however, the speaker only assumes that the addressees might want/ intend to drink coffee.In ( 16), however, nothing is said about the addressees' claims or intentions.In other words, in the case of ( 16), only the future will show whether the addressee(s) will drink coffee or not.
To put it shortly, in the case of future I clauses, the future event is anchored in the present in a way, which is not the case with future II clauses.This anchoredness is manifested in two ways: it is either a non-speaker's opinion/claim (factual conditionals) or the speaker's assumption about a prerequisite for a future event (hypothetical conditionals).This prerequisite is normally the intention of the subject, but it can also be represented by any other circumstances that may lead to the realisation of a process in the future.Future II in conditional clauses, however, indicates a potential process acting as the condition for the realisation of the apodosis.It is a process about whose probability nothing is said by future II itself,22 but it is left to the future to show whether the process will be realised or not.This type of potentiality can be described as neutral potentiality.
The anchoredness of the condition depends largely on the speaker's conception.For this reason, in fact, both ( 13) or ( 15) and ( 14) or ( 16) may be used in the same circumstances, depending on the importance that the speaker attaches to another's claim or intention.In hypothetical conditionals, however, future II cannot occur if the condition for the apodosis is not the event itself, but rather a prerequisite for the event, as in example (17), which is about an advance payment. 23In factual conditionals, future II cannot occur simply by definition.
(17) Ako će mi oni odvesti sve ovo smeće, ja ću im platiti.drive away-3 3PL PL. .FUT FUT..PFV PFV pay-1 1SG SG. .FUT FUT..PFV PFV 'If they'll drive all this rubbish away for me, I'll pay them.'This clause may be understood as factual, or as hypothetical but close to factual.In the case of the hypothetical interpretation, the speaker assumes that they may be willing to drive the rubbish away and (s)he will pay them if that willingness is confirmed.
Anchoredness in the present, or its prerequisite, is generally more important in the case of posteriority than in the case of anteriority and simultaneity.Namely, in the latter case, the relation between the condition and the consequence is usually direct.So, if example (17) were about a payment following the completion of the work, the willingness to drive the rubbish away would normally be relatively unimportant: the speaker will pay once (s)he knows that the work has been done.In that regard, posteriority is indeed more commonly expressed by future I, and simultaneity (and anteriority) by future II.Consequently, in examples in which the simultaneous interpretation is impossible and for which there is no detailed context, future I in the protasis may be understood automatically as posterior, and future II as anterior.
( As shown by ( 13), however, anchoredness can also be important in the case of anteriority (and simultaneity).On the other hand, if there is no anchoredness in the present, future II can also denote posteriority.In these cases, we are dealing with 'retroactive conditions', as in the following example.Although the relation of such conditions with the consequence is not direct, there is no anchoredness in the present either, and the consequence arises from the information about the condition that is obtained in the period between the moment of utterance and the realisation of the consequence.

Relative clauses
In the majority of Standard Croatian relative clauses, both future I and future II can occur.Future II, however, is not used in non-restrictive relative clauses and future I does not occur in one type of restrictive relative clauses: temporal clauses.The following is a comparison of the use of future I and future II in entity relative clauses.In Croatian, these are the restrictive relative clauses that involve relativisers such as koji 'who/which/that', tko 'who', što 'what/which/that', which refer to entity. 24What applies to entity relative clauses, is also true, mutatis mutandis, of other restrictive relative clauses: qualitative, quantitative, spatial and temporal.
The difference between relative clauses can be illustrated by the following examples.
( Example (21) may be interpreted in two ways.In the first interpretation, the noun phrase majstor koji... 'the tradesman who...' is non-referential.In that case, the relative clause defines the type of tradesman: the expression the tradesman who will do your tiling is similar to the tiler (as opposed to some other type of tradesman, e. g. a painter).The tradesman's identity is irrelevant, but it is assumed that there exists a tradesman who will tile the addressee's walls.In the second interpretation, the noun phrase majstor koji... is referential, i. e. the tradesman's identity is more or less defined at the moment of utterance.The definition expressed by the relative clause is basically the same as in the first interpretation: the tradesman who will do your tiling is similar to that tiler of yours.However, since the identity of the tiler who will do the tiling for the addressee is known, the relative clause identifies the tradesman as an individual and may contrast him to other individuals, i. e. other relatively defined tradesmen.Example ( 22) has only one possible interpretation.The noun phrase majstor koji... is nonreferential here.Unlike the first interpretation of example (21), however, in (22) the tradesman is not defined generically, but rather as an individual.This is deferred definition: The tradesman's identity is unknown at the time of utterance, but the relative clause represents the situation that will determine his identity in the future.
Relative clauses that employ future I, therefore, generally provide direct generic and direct concrete identifications of the parameter,25 whereas those involving future II provide indirect concrete identification of the parameter that represents the expressed or unexpressed antecedent (sc.or postcedent) of the relativiser. 26arginally, future I clauses also provide somewhat indirect concrete identification.These are clauses with an expressed or unexpressed non-referential antecedent27 in which future I indicates a prerequisite for a future process: generally the will/intention/willingness of the subject.In Standard Croatian, this is a marginal phenomenon, but it is more common in certain local dialects.For example, in the Central Čakavian dialect of Kukljica on the island of Ugljan, the use of future I and future II differs in the following examples.In ( 23), a degree of willingness to care for the mother in the future is presumed to exist in the present, whilst in (24) it is irrelevant whether such willingness exists in the present or not.
Due to their mutual similarity, examples ( 21) and ( 22) illustrate the difference between future I and future II in relative clauses very well.The above-mentioned difference between the three types of relative clauses, however, is probably better illustrated by the following examples.In some cases, paraphrase in the form of a conditional clause may bear less resemblance to the relative clause of indirect identification (cf.[30]), but certain similarity between them still exists.
(29) Ako ti X/Y/Z majstor bude lijepio pločice, neka ti zatvori te rupe.stick-3 3SG SG. .FUT FUT2.2.IPF IPF tiles-ACC ACC close-3 3SG SG. .IMP IMP..PFV PFV 'If tradesman X/Y/Z does your tiling (in the future), let him fill in those holes for you.' Relative clauses of direct identification, however, cannot be paraphrased as typical hypothetical conditional clauses.In the case of direct generic identification, the closest type of paraphrase is usually the one that involves the expression takav da 'such that', as in (30).A clause of direct concrete identification, however, may be paraphrased fairly closely by a factual conditional clause, as in ( 31), although paraphrasing by factual clauses is sometimes unsuccessful.Relative clauses of indirect identification also differ from relative clauses of direct identification in their tendency to involve inversion and insertion, as well as in that they can include relativisers with the particle god '-ever'.Examples such as ( 27) thus cannot contain future I, but only future II.
Direct relative clauses that involve future I are anchored in the present, whilst indirect relative clauses involving future II are not.Namely, in future I direct relative clauses, a certain type of parameter or the parameter itself (entity, quality, quantity, etc.) which can be considered to be real in the present is described through its future role.In the case of future II indirect relative clauses, however, such a type of parameter respectively the parameter itself appears not to exist in the present.In other words, in indirect relative clauses, i. e. in relative clauses that contain future II, the starting point is not the parameter (of which the future situation is characteristic), but rather the future situation itself, which should define the value of the said parameter.Accordingly, future I occurring in relative clauses is similar to future I in Croatian generally, i. e. although it is in the presupposition, it possesses a degree of 'indicative assertionality' (conviction/assumption/prerequisite in the present), which future II occurring in relative clauses does not have.This does not mean that future II indicates the same degree of potentiality in a relative clause as it does in a conditional clause.Future II relative clauses are most often considered more likely to be realised than not.However, due to the value of the parameter that is yet to be defined, the proposition is less well-founded, so future II relative clauses are close to some extent to Standard Croatian relative clauses in the optative.The fact that the degree of probability in relative clauses that employ future II or the corresponding present is not particularly distant from neutral potentiality in hypothetical conditional clauses after all, at least in some contexts, is demonstrated by examples such as the following.Finally, even when future I occurs in relative clauses that are 'somewhat indirect', it presupposes an intention in the present, so these cases too involve the anchoredness in the present, which does not exist in the case of future II.
Both direct and indirect relative clauses are characterised by various tenses in both the relative and the main clause.Certain differences exist in this usage too and I will mention only those that are typical of relative clauses involving a future form.Determining the value of the parameter is simultaneous with the time of speaking in future I direct relative clauses, whereby the parameter is often a referent known at the moment of utterance (direct concrete identification).In that regard, future I is equally common in the relative clause in combination with both non-future and future tenses in the main clause.Examples involving a future tense in the main In the case of indirect relative clauses that include future II, however, the value of the parameter is determined subsequently, through a future situation about which the speaker and the addressee may never know anything.In that regard, future II is common in the relative clause only in combination with a future situation in the main clause, as can be seen in the examples adduced thus far.This is expected too, because it is uncommon to talk of a role played in the present or in the past by a parameter whose value is yet to be determined in a future situation.Although uncommon, however, it is still possible, as is demonstrated by the following examples.Because of the semantics of the hypothetical period, however, future I is more or less equally marginal in the protasis as future II, when the apodosis contains a present or past tense.

Typological parallels
The fact that Croatian future II occurs exactly in restrictive relative clauses and in hypothetical conditional clauses is also interesting from a wider perspective.Namely, the future is often expressed in a specific way in other languages too in similar contexts. 29For example, to express the meaning of Croatian future II in such contexts, certain languages use the 'ordinary' future (e. g.Russian), others use the present indicative (e. g.English and German), and some use forms that are interpreted as subjunctive.Portuguese, for example, uses a form labelled the future subjunctive in a similar way as future II and the corresponding present are used in Croatian. 30imilarly, the Ancient Greek equivalents of these two forms in conditional and relative clauses are the present subjunctive (imperfective) and the aorist subjunctive (perfective). 31The clauses in question involve conjunctions combined or fused with the particle ἄν.In such conditional and relative clauses, the Greek present/aorist subjunctive normally corresponds to Croatian future II/present when the main clause describes a situation unfolding in the future (in German grammars of the Greek language such clauses are referred to as eventuell/Eventualis32 ), and if the main clause describes a situation that includes the moment of utterance, such conditional and relative clauses denote the present iterative.The iterative is associated with contexts characteristic of future II by reduced concreteness.The subjunctive occurring in conditional and temporal clauses also corresponds to Croatian future II in Modern Greek.What is more, the relation between the future and the subjunctive in Modern Greek conditional clauses is similar to that between future I and future II/ present in Croatian (Hedin 2000: 332-333).This relationship, however, is not completely identical, and besides, the future also occurs in temporal clauses in Modern Greek, albeit relatively rarely.For the future in conditional and temporal clauses in some languages, cf. the table in Hedin (2000: 331-332).
In the cited article, Hedin refers to Bybee & Dahl (1989), a typological paper that analyses the results of the research in the use of verb tenses in 50 languages.This work also notes the peculiarity of referring to future time in various languages in clauses involving the conjunctions whatever, if, even if, and when. 33In the context of these conjunctions, the authors (Bybee & Dahl 1989: 93-94) have noted two deviations in future time reference with regard to the situation in the usual contexts, such as independent clauses.These deviations are: 1) a higher ratio of inflectional34 forms and 2) the predominance of non-future verb forms, i. e. those forms that are not specialised for future time reference.The authors assume that future-tense marking is redundant in clauses that involve the afore-mentioned conjunctions.On the other hand, they consider the inflectional forms to be more suitable than the periphrastic ones for the contexts in which they are redundant, because they are more grammaticalised and more desemanticised.Exactly in this way they explain the more frequent occurrence of inflectional forms with these conjunctions.Based on the insights and assumptions presented, the authors have drawn two more conclusions: 1) in the clauses involving these conjunctions, we are dealing with "pure future time reference, without a sense of prediction"; 2) since the future tense is relatively rare in such clauses, simple future time reference is not its central meaning.These claims by Bybee and Dahl involve a number of problematic elements.Firstly, future-tense marking is not redundant in clauses employing the conjunctions whatever, if, and even if; it can be considered redundant only in the case of the conjunction when.For non-redundancy of future tense marking in if-conditionals cf. ( 41 Redundancy, therefore, will not be the reason for a higher frequency of inflectional forms in the afore-mentioned contexts.This is implied by another context discussed in Bybee & Dahl (1989), in which the future is indeed redundant.This is the complement clause that accompanies the verb promise, in which the ratio of inflectional verb forms in the questionnaire is lower than with the afore-mentioned conjunctions.Secondly, I could agree with Bybee and Dahl in that the clauses in question essentially involve the future without a sense of prediction because I, too, have spoken about potentiality and neutral potentiality.However, I do not consider the expression pure future time reference to be clear or appropriate.Accordingly, I do not regard as clear the claim that simple future time reference is not the central meaning of "future grams".A more or less typical future tense, such as Croatian future I, which is grammaticalised to a considerable extent, refers to future time, with certain peculiarities, in a similar way as the preterite refers to the past, and the present tense to the present.In an independent affirmative clause, all these tenses denote a statement, and thus a relatively high degree of conviction (unless the clause contains words that cancel it, such as perhaps or possibly).Namely, just as nobody can be certain of a future situation because they cannot know what will happen, similarly we often cannot be certain of a present or past situation either.The degree of certainty, thus, may be even lower in (42) than in ( 43).All these statements involve a degree of conviction.Such conviction is characteristic of the indicative, whether in the perfect or future tense.If, however, a language or dialect has a special form (along with the future indicative) peculiar in that it denotes the future without conviction (i.e. as potentiality/neutral potentiality) or special future situations that are abstract due to the unidentified value of the parameter, that form cannot possibly be an indicative tense in that language.As for the answer to the question what exactly kind of mood should future II be regarded as if it is not indicative, I do not consider it particularly important, all the more so because such an answer is difficult to provide.Should we start from the morphosyntactic features of future II, we could interpret it as subjunctive since it mainly occurs in subordinate clauses.If we start from its semantic features, which would be much more desirable, terms such as neutral potential and inassertive come to mind, none of which is completely satisfactory for a number of reasons.What is much more im-portant than an accurate name for future II as a mood is the awareness that the crucial difference between the two futures is neither aspectual nor temporal, but rather modal: the difference between the anchored and unanchored future.As for the name of this form, the term future II can still be used until a better one has been coined.

A glimpse into diachrony and the conclusion
Some light is shed on future II itself as well as on the peculiarity of referring to future time in contexts typical of future II by a short diachronic overview of the development of the Croatian future II forms.As we know, there are two forms referred to as future II in Croatian: the dialectal one, with the infinitive, and the standard one, with the l-participle, which is also characteristic of most local dialects.Both future II forms are complemented by the perfective present, and the difference between the future II forms and the perfective present in their respective contexts is primarily aspectual.
The development of the usage of future II formed with the infinitive differs considerably from the development of future II with the l-participle.The geographical area in which future II with the infinitive has been used in modern times is relatively poorly defined.It is certain that this type of future II exists35 in part of the Čakavian dialect group (e. g.Ugljan and Sveti Filip i Jakov [my data], Božava, Vrgada, Susak [Menac-Mihalić 1989: 90]) and in some Štokavian dialects (the Western dialect [Šimundić 1971: 198;Lisac 2003: 57], the East Bosnian dialect [Lisac 2003: 85], the East Herzegovina-Krajina dialect [Lisac 2003: 103]).A considerably wider area has been defined by Kravar, according to whom (Kravar 1978: 22), the 'budem + infinitive' future can be heard "u čitavoj Dalmaciji, po Hrvatskom primorju i Istri s pripadajućim otocima, u Lici, Bosni-Hercegovini i Slavoniji" ["in all of Dalmatia, in Hrvatsko primorje and Istria with the associated islands, in Lika, Bosnia and Herzegovina, and Slavonia"].Kravar, however, did not provide any examples or references to a source of this data, and besides, more than half a century has past since the publication of his work, so the area in which the 'budem + infinitive' future may be used should definitely be investigated more thoroughly.This future form, according to Kravar (1978: 23), is mainly imperfective and rarely perfective in modern dialects.A short (incomplete) overview of the bibliography on the Croatian 'budem + infinitive' future has been provided by Whaley (2000: 29-35).The 'budem + infinitive' form used to occur in independent clauses too, i. e. functioning as the 'normal' future, but it is not clear whether it can still be used like that in any dialect. 36Rösler (1952, as  cited in Whaley 2000: 29) claims that the 'budem + infinitive' future is more common in independent clauses in Čakavian than it is in Štokavian.However, Rösler has not provided any examples obviously, and the quoted work does not make it clear whether this claim is about modern Čakavian dialects or about older Čakavian texts.Whaley (2000: 30) attempted to find a modern Čakavian example, but the one selected is inappropriate. 37Namely, the author cites an example from Kalsbeek (1998: 284) nȉč ne buȏde takȍ čȕt 'there won't be anything to hear this way', in which the verb bȉt 'be' (i.e. buȏde) is not an auxiliaryin any case, not the auxiliary used in forming the future tensebut rather a peculiar modal verb in the present tense with a future meaning. 38The use of the 'budem/budu + infinitive' future as the normal future tense (especially in independent clauses) in the older language is wellattested.A large number of examples of such usage, starting from the 16 th century, have been provided by Daničić (1882: 354-358), and two examples have been added by Kravar (1978: 20). 39he 'budu/budem + infinitive' future was once used in a large part of Croatia and beyond, and unlike today, it was often perfective. 40It was first attested in Croatian in the second half of the 14 th century41 and had definitely existed in spoken language for a while before its earliest attestation in the written record (Whaley 2000: 33).The 'budu + infinitive' future spread to Croatia from the West Slavic area42 , followed somewhat later by the future formed with the verb htjeti 'want', which arrived from the Eastern South Slavic region (Vela 2018: 75-76).Shortly afterwards, these forms supplanted the older ways of expressing the future to a certain degree: the perfective present and the 'ingressive verb + imperfective infinitive' constructi-ons.In Čakavian, in fact, the 'budem + infinitive' future did not displace the perfective present; instead, it gradually came to be formed mainly from imperfective verbs (cf.Whaley 2000: 33).Finally, in a few centuries, the htjeti-future basically spread into the contexts typical of modern future I, whilst the 'budem + infinitive' future and the perfective present maintained the 'function of the imperfective and perfective future II'.Along with the use typical of modern future I and future II, yet another very common use of the 'budem + infinitive' future existed in the earlier stage of the language: the use in volitive type clauses. 43This use can still be heard, albeit rarely, cf.example (44) from the Kukljica dialect.Future II formed with the l-participle evolved from the Common Slavic deferredcognition evidential, cf.Whaley (2000: 114), where this form is referred to as the future perfect.This was a form with a relatively complex and therefore unstable meaning.It denoted a situation that lies in the past relative to the moment of speaking but which becomes known in the future (as in ako budem lagao 'if it turns out that I've lied'). 44The original meaning of the deferred-cognition evidential was preserved definitely into the 20 th century in certain dialects and perhaps even up to this day in some. 45This was also the basic meaning noted for the 'budem + l-participle' form by Maretić (1899: 614), who referred to this form as the drugo složeno prošlo vrijeme (the second compound past tense).The deferred-cognition evidential, however, began to develop various other meanings relatively early (cf.Kravar 1960: 31-36;Daničić 1882: 359-360), and since the 16 th century, the usage typical of modern future II, as described in section 3.2, has also been attested.This is its only use that can be considered standard nowadays, and the situation was more or less the same about seventy years ago (Grickat 1956: 91-92).The meaning of future II is very close to that of the deferred-cognition evidential, as the examples demonstrate, and since the 'budem + l-participle' form with this meaning certainly had little competition in the east, this is its most common meaning today.Unlike future II, however, 43 The term volitive type clauses is used here to refer to complement clauses of desiderative, manipulative, and some other predicates, and to adverbial final clauses.For examples, see Daničić (1882: 357-358).44 For the meaning of the deferred-cognition evidential, see also Whaley (2000: 109-110).45 Cf.Lisac (2003: 85).An interesting opposition occurring in the East Bosnian dialect has been noted here between the deferred-cognition evidential (budem + l-participle), future I (formed with the verb htjeti 'want'), and the perfective present and imperfective future II (budem + infinitive).
Contemporary Croatian future II the deferred-cognition evidential is more often formed from perfective verbs, as expected, but the function of perfective future II was already performed by the perfective present, so the modern participial future II form became mainly imperfective, just as the infinitival one is.During its evolution into future II, the deferredcognition evidential lost its preterite meaning (the moment of becoming aware in the future was replaced by the situation itself), but it has preserved, to a greater or lesser extent, its future neutral potentiality meaning.This is another fact that implies the incorrectness of the understanding of future II as an anterior future tense, and the correctness of regarding future II as a non-indicative mood.Except in Kajkavian dialects, the 'budem + l-participle' form occurs quite marginally as the normal future tense.The future II form with the l-participle, which used to be more typical of Eastern Central South Slavic, began to displace rapidly the western Čakavian and Štokavian future II form with the infinitive relatively recently, with the standardisation of the Croatian language (Kravar 1978: 23).
It is evident that Croatian future II, which is considered to be a non-indicative form here, has developed from the indicative.The future II form with the infinitive used to be the normal future tense, whilst the future II form with the l-participle used to represent the deferred-cognition evidential.Even if the indicative nature of the deferred-cognition evidential could be disproved to some degree,46 the indicative character of the old 'budem + infinitive' future is indisputable.The development of the non-indicative from the indicative is not a particularly unusual phenomenon, and in the case of a tense with a future component, it can be explained easily.A simple tense that has both an indicative and a non-indicative meaning loses the former one for one reason or another.For example, in many Čakavian dialects (e. g. on the island of Ugljan) and in some Štokavian dialects (see Maretić 1899: 618), the verbal form of the type bȉće lȅga, which is mentioned in Section 2 as the (Romancetype) anterior future tense, functions exclusively as a mood used to express an assumption or prediction (the meaning as in: he will (probably) have gone to bed). 47he Italian anterior future also has the non-indicative epistemic meaning (Renzi et al. 1991: 124, 128, 133), and in some Čakavian and Štokavian dialects, this is the only meaning that has been preserved, or the only meaning they borrowed.In the case of future II, however, the indicative meaning has been taken over by other forms which had either existed before or appeared later on.
It remains to establish why the new future with the verb htjeti 'want' has not spread into all future contexts of the perfective present and the 'budem + infinitive/lparticiple' form.Despite the claim by Bybee and Dahl, as has already been said, the reason for that will not be the redundancy of future time reference, because this is characteristic only of relatively few uses of future II and the corresponding present.The desemanticisation mentioned by Bybee and Dahl, however, does play a certain role.Namely, it is expected that, during its gradual expansion through the Štokavian and Čakavian areas, the htjeti-future had a largely prospective and intentional meaning.Since prospectivity and intentionality are completely excluded from the meaning of future II, the htjeti-future has not spread into the contexts typical of future II.The old perfective present had a fairly broad meaning, which means, in other words, that it was largely desemanticised.For this reason, it was displaced easily from the typical future usage by the new htjeti-future, but was completely suitable for marginal usages, such as the volitive type clauses and the contexts typical of future II.The old 'budem + infinitive' future also had a broad meaning, which is evidenced by its use in volitive type clauses. 48The desemanticised status of the deferred-cognition evidential and its compatibility with the contexts typical of future II have already been discussed.The prospectivity of the htjeti-future largely faded away later on, but by that point, the future I vs. future II opposition had already emerged.This in a way petrified opposition, however, by no means implies that "simple future time reference" is not the central meaning of futures such as Croatian future I, as Bybee and Dahl would probably conclude.It is, in fact, the meaning of future II which is largely marginal with respect to the typical usage of the future tense and, as has already been said, it is actually non-indicative, i. e. it lacks the sense of assertionality/conviction that is generally characteristic of the indicative.In that regard, it is not unexpected that some new future may not spread into future II contexts, even if it would not be prospective.