JAIS INTERNET THE ROLE OF THE B-PREFIX IN GULF ARABIC DIALECTS AS A MARKER OF FUTURE , INTENT AND / OR IRREALIS

This paper presents arguments for a re-analysis of the b-prefix in Gulf Arabic dialects. Similar to several other dialects, Gulf Arabic possesses a bprefix that is inserted before the p-stem (prefix form) of the verb. However, the Gulf Arabic b-prefix differs substantially from the one encountered in other Arabic dialects. According to most previous studies, the Gulf Arabic b-prefix encodes future tense or intentive mood or a combination of these. Based on a thorough survey of the use of this particle in modern speech, I suggest that it is used in Gulf Arabic today as a generalized marker of the irrealis mood rather than being limited to function as a future/intentive marker. Futurity is one––but not the only or necessarily the most important one––of its connotations. Meanwhile, another marker, rāḥ, emerges as an obvious future marker in some parts of the dialectal area.

Introduction 1 The Levantine and Egyptian dialects of Arabic possess a b-prefix that is inserted before the p-stem (prefix form) of the verb to form the indicative mood and/or cursive 2 , progressive or habitual aspect.In the dialect of 1 This study constitutes a part of a larger project on the typology of verb syntax and tense/mode/aspect morphology in Gulf Arabic, 'Verb Syntax in Gulf Arabic Dialects', funded by the Bank of Sweden Tercentenary Foundation.Initial reports from this study have been presented at various occasions (Persson  2006a, 2006b, 2008a, 2008b, 2008c).I would like to thank my Omani colleagues at the GAP institute in al-Ain/Buraimi who patiently answered my many questions, and curriculum director Tim Peverill who proofread an earlier draft of this article and contributed with his expertise in both Nizwa dialect and English language, as well as my employers who have supported me in every way and worked my teaching schedule around my research needs.The project would also not have been possible without the help of a large number of former GAP students who are now working in various areas of the Gulf and who introduced me to their local friends, colleagues and neighbours.Through them I received access to people's homes and obtained unique interviews with local women.
2 A generalized, cursive aspect expresses a universal truth or ongoing event want'.These were included in this study for comparison with the bprefix.
The study is based on a corpus of twenty-three hours of recorded authentic speech.The interviews took place mostly in people's homes during 2006-7, and were recorded by myself and by local staff. 5ranscriptions of database material have been made with the purposes of this study in view.This article deals with syntactical and morphological traits in language, but not with phonology.The database covers dialect areas that differ from each other phonologically, for example, in the way certain letters are pronounced.Such differences have not been consistently noted in the transcriptions.Rather, to some extent, an attempt has been made towards a more unified representation of words to make it easier for readers who are not so well acquainted with the phonological peculiarities of each of the Gulf dialects to recognize words and read the examples given.
Tense, mood and aspect in Gulf Arabic 6 A general comparison between classical Arabic and modern Arabic dialects reveals that tense plays a more prominent role in the dialects and that some dialects have developed fairly comprehensive systems of tense/mood/aspect markers. 7According to Eksell, there is a general tendency to renew the ancient cursive aspect.This, she says, is accomplished by developing particular markers distinguishing the progressive/progressive-habitual, the imminent/volitive and the future.The old, unmarked imperfect form then contrasts with these to denote the cursive aspect in general. 8In, for example, the urban Levantine and Egyptian dialects, elaborate temporal systems have been created.Thus, 5 Gulf Arabic as defined in the project comprises the dialects of southern Iraq, Kuwait, Bahrain, Qatar, eastern Saudi Arabia, Oman and the UAE. 6For a definition of tense, mood and aspect as used here, see Comrie 1976,  3, 1985, 6, 9 and Palmer 2001, 1. Palmer gives a concise definition of TMA saying that 'Tense, rather obviously, is concerned with the time of the event, while aspect is concerned with the nature of the event, particularly in terms of its 'internal temporal constituency' (Comrie 1976, 3).Modality is concerned with the status of the proposition that describes the event'.Cf. also Bybee et al.  1994, 176-181; Givón 1995, 112; Ingham 1994, 117f; Trask 1993, 92, 174f; Van Valin and LaPolla 1997, 41.  7 Comrie 1976, 80 n.1; Cuvalay-Haak 1997, 217-224, 238f; Holes 2004,  217ff, 226ff; Kaye and Rosenhouse 1997, 291, 304f; Mitchell and El-Hassan  1994. 8Eksell 2006, 79.

PRE-PAPER JAIS INTERNET
in Syrian dialect we have a basically temporal tripartite system with past; present; future where the past is signified by the s-stem (suffix form) of the verb or the active participle; the present by the p-stem of the verb; and the future by the particle rāḥ combined with the p-stem of the verb.
For compound tenses various forms of kāna, 'to be', is used as an auxiliary.In addition, there are examples of modal particles such as the prefix ʿam(ma)which is added to the p-stem of the verb to express the progressive.The already mentioned b-prefix in this dialect signals indicative mood and has also been suggested to have aspectual functions. 9 The dialects of the Arabian Peninsula, however, are comparatively poor in terms of modal or temporal markers. 10Ingham says, describing the Najdi dialect of Saudi Arabia that: When contrasted with the modal auxiliary verbs of English, […] Turkish or […] Swahili, which are morphologically and distributionally definable classes of elements, the Najdi modals appear as a rag-bag of elements derived from verbs, adjectives and primitive particles and rather vaguely differentiated from certain other elements on the periphery of the group. 11cording to Ingham, Najdi Arabic, like Classical Arabic, is an aspectcentred system.Although he does recognize sporadic signs of a developing tense-based system within the main structure, he devalues any overt tense markers that may appear in the dialect by believing them to stem largely from direct or indirect influence from other Arabic dialects. 12he temporal and modal system of Gulf Arabic appears to be quite rudimentary.Comrie holds (classical) Arabic to be a language with 'combined Tense/Aspect oppositions' considering the distinction between 'imperfect' and 'perfect' in Arabic to signal both aspect and relative tense. 13Ingham, in his study on the Nadji dialect in Saudi Arabia, describes Arabic as 'a language of the type showing Aspect with tense implications'. 14Holes, writing about the conservative verb morphology of Bedouin dialects and referring to data from Abboud  (1964), comments that no system of modal particles has developed in the 9 Ebert 2000, 776f;Eisele 1990, 184-190;Holes 2004, 226ff, 366ff. 10 Brustad 2000, 241;Holes 2004, 218f, 226, 236f;Ingham 1994, 87,117ff. 11 Ingham 1994, 117. 12Ibid., 87ff, 109. 13Ingham 1994, 87.dialect of the Nadjdi town of Haᵓil.15However, concerning the urban Gulf Arabic dialects, he remarks that: The Bedouin-descended dialects of the seaports of southern Iraq and the Gulf littoral are an interesting typological compromise between the reductional tendencies of the Levant cities and the conservatism of inner Arabia, which again reflect the sociolinguistic history of their speakers. 16les goes on to say that a partial system of mood particles has developed in the town dialects in Kuwait and Bahrain for the expression of future intention.His conclusion is, though, that mood marking in the Bedouin dialects and particularly those of the Arabian Peninsula is less fully evolved. 17Brustad, finally, concludes that the dialects she studies, including the Kuwaiti dialect, are basically aspectual, and especially so in narratives.Basic time-reference is established around the moment of speech but other devices such as adverbs and context are used to change the temporal reference point.A special group of verbs that she calls 'time verbs' with kāna as its central member have as their primary function to establish time-reference. 18evious studies on the Gulf Arabic b-prefix and other suggested markers for future tense The Gulf Arabic (GA) b-prefix has held a prominent position in what has previously been written concerning future tense in GA.In both older and more recent research on Gulf Arabic the b-prefix is presented as a marker of future tense. 19Johnstone considers it to have future meaning with a sense of volition in Kuwaiti, Bahraini and Qatari dialects.He also mentions its use to signal future tense in the dialects of the Trucial Coast. 20Qafisheh mentions the verb baġa 'to want' with a volitive meaning, but does not seem to include the b-prefix in his description of

PRE-PAPER JAIS INTERNET
Gulf Arabic grammar. 21Brocket finds the b-to be used for pure future tense--without any implication of will or intent--in his recordings from 1979 and 1980 of the dialect in the Omani coastal town al-Khabura.He further makes the rather categorical statement that the use of this b-prefix is 'restricted to future meaning' and that future tense is almost always marked by this prefix in his data. 22Brustad says that the Kuwaiti bseems to indicate both future and intentionality, at least in the speech of one of her informants.She also reports one instance where she interprets the use of b-as marking intention in opposition to another particle (rāḥ) denoting future tense. 23A differentiation in the usage of b-and rāḥ, though a somewhat different one, was noted by Johnstone in his data from Kuwait and Bahrain.There, he found that b(i)-usually has the sense of volition whereas rāḥ usually indicates intention. 24Al-Maʿtūq, also describing a Kuwaiti dialect, recorded a use of the b-for the near future. 25Holes relates that the verb yabi, 'want', (in the form yabba) is used together with the p-stem of the verb in the Shiite dialects of Bahrain (Baḥārna) to express both 'proximate intention' and 'wanting.'In Gulf littoral dialects in general, however, he observes a distinction between ba-/bi-expressing intention (baru:ḥ 'I'm going to go') and yabbi+verb for volition (Ɂabbi aru:ḥ 'I'd like to go'), and he concludes that ba-/bihas lost the volitional sense of the verb(s) it derives from. 26According to Al-Tajir, b(i)-, which he also derives from yabi, is employed in Baḥārna to indicate a future action. 27Feghali, finally, simply states that the future in Gulf Arabic can be (but not necessarily is) expressed by the prefix b(a)-added to a p-stem of the verb.The use of the b-prefix expresses an assumption about a future event. 28he Najdi dialect of Western Saudi Arabia falls outside of the scope of this study but is, nevertheless, interesting for comparison.There, Ingham reports, both the full form of the verb baġa/yabi, 'to want', and 21 Qafisheh 1977, 59, 224-229. 22 Brocket 1985, 21f.Khabura is a town on the coastal line between Sohar and Muscat in the Batina region. 23Brustad 2000, 242f.The scarcity of her data, however, prevents any further conclusions.Johnstone 1967, 143, 152. 25 Al-Maʿtūq 1986, 195. 26 Holes 2004, 247 n.29. 27 Al-Tajir 1982, 80, 110. 28 Feghali 2004, 62. its reduced form b-are used to express plain future, future of intent and imminent future as well 'being on the point of'. 29Ingham further reports that the basic meaning of the b-prefix in the Najdi dialect is 'wish/intend'. 30he status of the b-prefix in the Gulf Arabic dialects thus seems to be rather unclear and in need of a more consistent and thorough study.This treatise is an attempt towards that end.

Database
The present study is based on a database of twenty-three hours of recorded speech from Bahrain, Kuwait, Oman, Qatar and the UAE. 31 The 80 informants are village or town people representing over 20 different cities and villages. 32Most of them are university educated although around 20% have little or no schooling.The majority of my informants (70%) are women.The recordings are largely of narrative or expository character but also include a fair amount of dialogue.
From this database, comprising at the moment about 4,000 analysed predicates, I have initially selected all predicates that have future timereference i.e. that are liable to encode future tense as well as future in the past.Among the 882 instances thus collected are both main and subordinate clauses; final, conditional and temporal clauses as well as complement clauses.In addition to these, I have included in this study all instances in my database of the b-prefix and rāḥ33 together with the occasional rāyiḥ and ḥa-34 that was found in the database as well as all instances of the verb yabġi/yabi, 'to want.'Only two instances of rāyiḥ, 16 instances of ḥa35 and one of the combination rāḥ ḥa-were found.

PRE-PAPER JAIS INTERNET
These 19 instances have, for the purposes of this treatise, been counted together with rāḥ.Lists of verbs in the same form such as, 'He will peel, chop and cook the onions,' were counted as one single instance in the corpus and the number of predicates with future time-reference was thus reduced to 851.
According to these criteria, a total of 1,012 instances have been collected.The number of predicates with future time-reference is, as said, 851 whereof 385 are preceded by b-or rāḥ and 4 are preceded by the verb yabġi/yabi, 'to want' whereas 462 are unmarked.The other part of the database viz.predicates with time-references other than future that are preceded by b-or rāḥ amount to 138 instances.As for the verb yabġi/yabi, 'to want', 23 instances of non-future use of them were found.
Although the whole database is more evenly distributed, the part that was selected according to the above criteria turns out to be dominated by material from Oman and the UAE and is divided on the countries as follows: The use of b-and rāḥ in clauses with future time-reference 45% of the 851 instances of future time-reference in the data base are marked with b-or rāḥ.This could indicate, in accordance with Johnstoneʼs observations, that they are possible, but not necessary, markers of future tense.A short survey of the grammatical contexts (types of clauses) with future time-reference in which these markers are used vs. not used reveals the pattern shown in the tables 2 and 3 (overleaf).We find that b-/rāḥ are relatively sparingly used in complement clauses and purpose clauses.Purpose clauses are dependent contexts and would typically be encoded as subjunctive/jussive in languages that mark grammatical mood.On the other hand, complement clauses can be both and the most frequent use of b-/rāḥ with future timereference is in ordinary main clauses. 36They are rather extensively used of an elderly, Qatari male. 36Cf.Givón 1995, 125f; Khan 1988, 40; Noonan 1985, 91f; Persson 1999, in result clauses of conditions but there is also a fair amount of apodoses without any particle/prefix.Grammatical structure of the sentence does, in other words, not seem to play an important role in the use of these markers, nor does any straightforward indicative/subjunctive mood distinction appear to be at hand.The use of b-and rāḥ in non-future contexts The 161 clauses with verbs preceded by any of the prefixes/particles that have other than future time-reference are distributed as can be seen in Table 4 (overleaf): 200f and 2002, 109-122, 134-7.See also Palmer 2001, 111-131.Generally speaking, complements of assertions are commonly marked for the indicative whereas complements to non-assertions are marked as dependent (subjunctive).Cf.Palmer 2001, 3f, 111. 37 Whereof 12 s-stems of the verb, 1 AP and 6 p-stems of the verb. 38Whereof 2 s-stems of the verb, 1 AP and 17 p-stems of the verb.The first conclusion to be made is that rāḥ is hardly ever used in nonfuture contexts with the exception of apodoses.This supports its status as a future marker.The second important observation to make is the very extensive use of b-, and to some extent rāḥ, in conditional clauses.Of 138 occurrences of b-or rāḥ in non-future contexts, 88 instances (64%) are in conditional clauses.Second to this comes the use of b-in ordinary main clauses with 43 occurrences (31%).We will start a closer look at non-future usages of rāḥ and b-by examining these conditional uses.

Conditional clauses
The extensive use of our presumed future markers in conditional clauses is noteworthy.Clive Holes writes about conditionals that: an s-stem verb is used to express the condition even though the action to which it refers is in the future, relative to the time of utterance.[…] These observations apply to MSA, and also to the Bedouin-type dialects, although they are becoming less true of the urban dialects, in which the s-stem/p-stem distinction has evolved further towards a true tense system, and p-stem verbs (= 'non-past') are often used in 'open' conditionals. 39at the future marker is used in apodoses with future time-reference need not be surprising.To the contrary, Bybee, Perkins and Pagliuca found that future markers are common in apodoses where predictions are made.They conclude that 'apodoses are prime environments for future grams'. 40However, in the present Gulf Arabic database the b-prefix is used not only in apodoses with future time-reference but also in apodoses with past or present time-reference: Only conditional clauses introduced by either of these particles were included in the study so far.In order to try and throw light upon what might trigger the use of rāḥ and bi-in conditional clauses I will now expand the study to include all conditional clauses found in the original data base.These will then be examined in search for any consistency in, and possible motivation for, the use of rāḥ and b-.
The total number of protases (conditional main clauses) in the whole data base at its present stage is 121.Of these, 29 have future timereference.The number of apodoses (result clauses to conditions) is 170 of which 68 have future time-reference.These distribute as follows over p-stem verbs preceded by b-prefixes or rāḥ, s-stem verbs and other types of predicates: Gulf Arabic it was, however, recorded already by Reinhardt 1894, 287-290. 41 Teenage schoolgirl, al-Ain, UAE. 42Middle-aged well-educated female, al-Ain, UAE. 43Young, well-educated male, Muscat, Oman. 44Elderly, upper-class male, Kuwait.The use of the b-prefix in apodoses is particularly striking: 50% of all the result clauses in the corpus contain verbs prefixed with this particle.Of these one-third (35%) have future time-reference.The share of apodoses that are futures introduced by b-comprises less than a fifth (18%) of the total number of apodoses.It is also noteworthy that almost a quarter of the protases have p-stem predicates with b-prefix and that b-prefixes do not dominate among protases with future time-references.The data is too small to make any far reaching conclusions.It is, however, clear that the b-prefix is used in conditional clauses to a much greater extent than as an indicator of future time-reference.Indeed, time-reference, at first glance, does not seem to play a role here at all, especially if we consider the significant use of the b-prefix in apodoses with past timereference.In languages in general, conditional clauses are neutral in terms of time-reference. 45Still, such a frequent use of an assumed future marker in conditionals calls for some explanation.I will discuss possible motivations for the use of the b-prefix in conditional contexts below.Before that we will, however, take a look at other contexts than the conditional clause where b-and/or rāḥ are used with timereferences other than future.

PRE-PAPER JAIS INTERNET
Other uses of b-and/or rāḥ with non-future time-reference Table 4 showed the sentence types in which b-and rāḥ are used with time-references other than future.The distribution, without the conditionals, is shown again for reference in Table 7.There is only one instance of rāḥ with other time-reference than future, but 49 examples of the use of b-in non-future context outside of the conditional use of these particles.In other words, the corpus contains a considerable number of ordinary main clauses with time-references other than future where the b-prefix is used.Hence, although the b-prefix is extensively used in clauses with future time-reference, its quality as a future marker seems less obvious.We saw in Table 4 that rāḥ only occurs 12 times in the data base with non-future time-reference.Rāḥ and its equivalent ḥa-as well as rāyiḥ thus basically appear when the time-reference is future.Hence, these particles can, with a rather high degree of certainty, be considered markers of future tense.The particle ḥa-was mainly noted in the speech of one young mother from Abu Dhabi.Reportedly, the particle ḥa-is used in some villages around Nizwa. 46 Rāḥ occurs in the speech of several informants from Oman and a number of young Kuwaiti informants as well as one Bahraini informant together with eight instances from Fujairah and one from Abu Dhabi in the UAE.Hence, the database so far indicates that rāḥ is more common in the Omani cultural and geographical sphere and possibly also in Kuwait than in other areas of the Gulf.Though I have no statistics for this yet, my impression is that these particles are used mainly by younger, educated speakers. 4746 Hamad Al Bu Sa'id, personal communication. 47Comments heard in Nizwa that it is 'Muscat dialect' leads one to think that the use of rāḥ is marked by prestige and/or fashion.We must not forget, though, that ḥa-is not a recent future marker in the Omani dialect.Cf.Reinhardt 1894, 149f, 287-290.It may be assumed that the use of rāḥ/ḥa-has been reinforced by the fact that virtually all schoolteachers in Oman from the 1970s until recently have been of either Levantine or Egyptian origin.As for Kuwait, where labour force immigration started at a considerably earlier stage, we find that

PRE-PAPER JAIS INTERNET
I will return to the use of b-in conditional clauses and main clauses with some proposals to an explanation shortly.Let us, however, first consider the suggested modal values of this prefix.

The use of b-and rāḥ as modal markers
The b-prefix is mentioned in the literature both as a temporal and as a modal marker or a combination of both.As a modal marker it is said to imply intent or volition.Most references suggest that the Gulf Arabic bstems from the verb(s) baġa (p-stem: yabġi) and yabi meaning 'to want'. 48The verb(s) baġa/yabi in their full forms are used sparingly in the present database.Only 26 instances are found in the corpus and they are all expressions of volition. 49 grammaticalization process, by which a word originally meaning wish/want/intend develops into a future tense marker, is not at all already Johnstone 1967, xxviii reports a strong Egyptian and Levantine influence on the local dialect.Brocket 1985, 21;Cohen, M 1924, 63ff;Holes 2004, 247, n.29;Al-Tajir 1982, 80. Ingham 1994, 94, 120 (in accordance with Nöldeke 1904, 67) considers yaba to be a reduced form of yabġi.Such a common origin can explain that 'both' verbs are mentioned as the source of the b-prefix. 49Examples are such as: kānat tabi tilabbis-na aḥdaṯ šē.be3sg.f.s-stem want3sg.f.p-stem dress3sg.f.p-stem-us more.modernthing She wanted/used to want to dress us in the latest thing (middle-aged female, Manama, Bahrain).Sometimes they seem to be used more as a particle than as a full verb, i.e. to be on their way to be grammaticalized as a modal marker.Cf.Ingham 1994,  120.Thus, in the following example from Kuwait where the verb yaba is not conjugated for number, ḫalas!yirūḥūn ḥadāyiq u yirūḥūn yilʿabūn mallū!yiba yrūḥūn li-l-barr.Enough!go3pl.m.p-stem parks and go3pl.m.p-stem play3pl.m.p-stem, get.bored3pl.m.s-stem!want3sg.m.p-stem go3pl.m.p-stem to-the-desert Enough!They go to parks and they go and play.They're bored.They want to go to the desert!(Young mother, Kuwait).Tim Peverill (personal communication) reports the use of baġa in the form of the active participle in Nizwa.As in the p-stem variety yaba the /ġ/ is lost in the feminine AP giving the forms bāyiġ/bāyeh, 'wanting,' for masculine/feminine respectively.It is easy to see how these forms could be further diminished into a b(a)-prefix.Curiously enough, however, my data from Nizwa contains only limited instances of the b-prefix; the particle rāḥ being much more frequent.
uncommon in the languages of the world. 50English is an obvious example of this and an example from Arabic dialects is provided by Maltese in the form of the future particle se. 51Future markers across languages are in general more or less optional; that is to say, most languages have a variety of ways in which they express the future.Moreover, future is seldom if ever a truly temporal category.More often than not, markers of future tense come with various modal meanings. 52ometimes the converse is also true and it is important here to note that intention often comes with a tint of futurity.Bybee, Perkins and Pagliuca go as far as to assume that all futures go through a stage of functioning to express intention. 53They propose that: intention is the crucial bridge to prediction and that the change from intention to prediction occurs via the inferences that hearers make on the basis of the speaker's utterance.The more common sources for futures are those that yield the intention inference most easily -desire, strong obligation, and movement toward a goal. 54nce, if the Gulf Arabic b-prefix is both a future marker and a marker of intentive mood, and especially if it has developed from a verb meaning to wish/want/intend, it could be assumed that the function as a marker of future tense is a secondary function to the modal one.I have therefore surveyed the various modal contexts in which the b-prefix occurs in search for predicates and contexts expressing an intention to act.Intention was given a broad definition so that I would not exclude anything that could be analysed as intentive.I found 213 predicates situated in a context that expressed some kind of conscious intention to act.As may be expected, almost all express intention towards a future action, but a handful (20), mainly conditional clauses, express intention that is not directed to any particular time: 5. Al-ḥēne yōm b-aḥuṭṭ šāy hina dāḫil b-abannid al-bāb 50 Bybee et al. 1994, 254;Heine and Kuteva. 2005, 104.See also Bybee et al. 1994, 279f and Palmer 2001, 105. 51 Ebert 2000, 763. 52 Cf.. Fleischman 1982, 24 who concludes that 'The future tense of all languages […] is always partly temporal and partly modal.Bybee et al. 1994, 254. 54 Ibid., 279f.In addition to these, I count 64 instances of future conscious intention that is unmarked.All this goes to say is that b-at least is not a necessary marker of intention, just as it is not a necessary marker of future; neither of which should surprise us--indeed only the opposite--viz.a necessary marker for either of these would be surprising. 57However, more significant in this respect is the observation that b-is also used together with verbs with non-personal subjects, as well as with verbs expressing state, ability and so on, i.e.where intention can not be at hand: 58  Cohen, D. 1984, 281, Fleischman 1982, 2. 58 According to Brustad 2000, 243, the Kuwaiti b-prefix can function as a marker of intention also outside the scope of the human will.However, neither the context of the example she provides for this nor her translation suggests an intentive rather than a future reading. 59Elderly female, Fujairha, UAE.Note the use of the verb baġa/yabi for volition here.

PRE-PAPER JAIS INTERNET
The number of inhabitants in Oman will increase/increases every day. 609. fi-l-mustaqbal al-iḫwān ma b-yaʿrifu baʿḍ in the-future the-brothers not b-know3pl.m.p-stem each other In the future brothers won't know each other. 61 therefore seems as if the context is more decisive than the presence or absence of b-to signal intention.At least, if b-has an intentive meaning, this connotation is not strong enough to bar the particle from use in contexts where intension is excluded.This leads to the question what previous scholars might have had in mind when they described b-as being intentive.We could be watching a diachronic development, my data being considerably more recent than most of the other studies made on these dialects.I find it likewise possible, though, that older works that state that b-signals 'future with a hint of intention' and similar reports have overlooked the fact that future in itself often comes with 'a hint of intention' and therefore ascribed this modality to the b-prefix.It is interesting to remember in this context that Holes, whose observations are more recent, also ascribe intentionality to the b-prefix: there is a contrast between baru:h 'I'm going to go' (intention) and ?abbi aru:h 'I'd like to go. 62e distinction Holes draws is one between intention and volition, not between future and intention.Furthermore, the example he gives for intention (baru:ḥ 'I'm going to go') suggests that he does not, in this context, make a distinction between conscious intent, on one hand, and the general sense of intent often connected with future, on the other.
Contrary to what is often assumed, the b-prefix in the surveyed data does thus not seem to be used as a marker of the intentive mood.This observation is pertinent since the connection between this marker and the intentive mood has been so strongly asserted for these and related dialects.
Instead, we have here a prefix which may and may not be used in contexts with future reference, which may and may not be used in contexts expressing intent, and which may and may not be used in conditional clauses.In addition, it is used in a number of ordinary main 60 Middle-aged, male teacher, Nizwa, Oman. 61Young, educated female, Buraimi, Oman. 62Holes 2004, 247 n.29.

PRE-PAPER JAIS INTERNET
clauses that fall outside these criteria. 63I will return to these shortly.We will, however, first revisit the most dominant use of the b-prefix outside of the future context, viz. the result clauses of conditions where rāḥ is also used to a notable extent.
The use of b-in main clauses and its relation to the use of b-in future and conditional contexts.Suggestions for a unified interpretation Verbs used to express the future tense are used also for the conditional in other languages in the world, though the verb form used in each context is often not the same.Thus we can, for example, talk about the conditional mood in English.Past forms of future markers are common in languages. 64Hence, the appearance of future markers in this context is not surprising.Strange as it may seem at first sight, we can also accommodate the use of rāḥ and the b-prefix in conditionals within a theory that interprets them as future markers, viz., if we look at the clauses from the speaker's perspective.The speaker may, when asked about how life would have been if he/she had lived a long time ago, make a temporal journey in his/her mind perhaps along the lines of: 'if we imagine how it would have been had I lived thirty years ago, what will we then find?In such an imaginary world into which I am now going to put myself, I will be living in a tent and I will marry at young age….'In other words, the speaker may think of creating an imaginary world and then consider what will be the consequences as he builds up the scenery.In such a context the use of future is logical.This is similar to that which Tedeschi suggests when expanding on his theory of branching futures.'Our intuitions about the analysis of counterfactual sentences', he says, 'indicate that we are considering truth relations with respect to a possible, albeit unrealized, future of a past time'. 65urthermore, 'we evaluate counterfactual conditional sentences as if we returned to the past and looked at possible futures with respect to that past'. 66This kind of reasoning works for both protases and apodoses: 63 It may be noted in this context that a selection of the database was also tested for the collocation of rāḥ and b-with stative vs. dynamic and telic vs. atelic verbs.No correspondence between these factors and the use of the markers were found.Cf.Bybee 1985, 194f.Cf.Tedeschi 1981, 255f. 66 Ibid.Brustad 2000, 202 makes a similar statement about the role of the speaker's perspective, not about time-reference or tense but about aspects, 'As a means of representation, aspect is controlled by the speaker, who chooses Following the theory of branching futures, the use of b-in 10 would come out of an (unconscious) 'reasoning' on the lines of 'lets go to a time where it is true that 'they are going to walk according to old times', then the conclusion is…'.Similarly, the use of b-in both protasis and apodosis of 11 can be explained by an imagined reasoning saying 'let's go to a point today where it is true that 'they will fight', then it is also true that 'tomorrow they will chat again''. 69o these explanations regarding the use of future in conditionals, we may add that futures also often have a conditional trait in the sense that their fulfilment often depends on certain conditions. 70Ingham even states that, 'In the Arabic system it seems that all future events, being uncertain, are treated as conditionals'. 71Thus, it is not only the case that there may be good reasons for the use of future tense in (non-future) conditionals, but it is also the case that futures share traits with conditionals.This leads us to a consideration of the modal aspects of futures and conditionals.The above discussion about time-reference and time perspective in conditional clauses may be helpful towards explaining the use of the b-prefix in conditional clauses.Another feature in the data remains, however, unexplained viz. the use of b-in ordinary main clauses.Of the 126 collected instances of b-prefix in non-future aspectual representation according to his or her perspective of an action or event'.Fleischman 1982, 20f comments on the role of the speaker's perspective saying that '…all too often in attempts to reconcile time and tense the focus is exclusively on sequence of events in real time, while the crucial role of speaker's perspective is neglected'. 67Elderly saleswoman, Sur, Oman. 68Young, educated female, Buraimi, Oman. 69For a discussion on Tedeschi's theory with applications, see also Dahl 1997, 101-107. 70 Fleischman 1982, 20. 71 Ingham 1994, 133.cases are statements about something that cannot be proved.The b-prefix is used for assumptions about a world that has not been experienced (futures, conditionals) or generalisations about past tendencies (habitual pasts).It is worth noting that in most of the habitual pasts where the b-is used the speaker has not himself experienced the past he is relating.He is relating what his parents' generation has told him.The b-is thus used for the unlived, the non-evidential.Mithun's definition of the realis/irrealis distinction may be helpful here: The realis portrays situations as actualized, as having occurred or actually occurring, knowable through direct perception.The irrealis portrays situations as purely within the realm of thought, knowable only through imagination. 76 The b-prefix in my data fits well into this definition.That a language has a marker for irrealis modality does not simply imply that everything that is not marked as irrealis is by definition realis, i.e. that all unmarked forms are notionally realis.Rather, languages that make a realis/irrealis distinction vary in the scope of these two categories. 77Reportedly, in some Papauan languages future is the only category marked as irrealis; 78 in others irrealis marking covers the imperative as well; in others still counterfactual is included and then there is one example where future, imperative, counterfactual and habitual past are all marked as irrealis. 79almer describes a marker in the Native American language Kiowa that has been said to denote future, and that definitely seems to indicate future, but that is also used in many contexts that may be described as irrealis such as stating a future that did not occur, the conditional or an obligation.Kiowa has, however, other markers for the imperative and the negative imperative as well as for 'hearsay'.The future-cum-irrealis marker in Kiowa does thus not belong to an organized system of realis/irrealis or even realis/irrealis/unmarked. 80On a hierarchy ranging from the most prototypical realis to the most prototypical irrealis, future seems to hit the end on the irrealis side in most languages.This does not mean, though, that a language that has irrealis marking for the future 76 Mithun 1999, 173 quoted in Palmer 2001, 1, See also Mithun 1995, 368. 77 See Palmer 2001 for an extensive discussion.Cf. also Mithun 1995, 386. 78 In which case the marking could be one of tense just as well as one of mood. 79

PRE-PAPER JAIS INTERNET
cannot leave futures unmarked or even mark future tenses as realis. 81I mentioned above that counterfactuals can be marked as irrealis.Palmer also reports several languages in which conditionals are marked as irrealis. 82he corollory of the above suggests that Gulf Arabic b-prefix, which is used to quite some extent for futures, conditionals and habitual past, has much in common with markers of irrealis in some other languages.That not all categories which are notionally irrealis receive this marking is not relevant since languages differ in what can receive irrealis marking.Also, irrealis being a modal distinction, the speaker may choose to mark this or not according to pragmatic and other speechsituation oriented choices.
This conclusion about the irrealis character of the b-prefix does not disqualify the traits shared between future and conditionals as discussed above.The reasons for using b-in conditionals may very well be a combination of its character as an irrealis marker as suggested here and of a speaker-oriented perspective frame putting the past conditional event in a future time frame.The one can have developed out of the other. 83If the b-has developed from the verb(s) baġa/yabi, 'to want', then the overall meaning of irrealis may be the one that has remained and made the particle suitable for all three contexts.However, the association of this particle with the future seems to have been and still be strong and this may have led to its use in conditionals as described above, with the irrealis connotation reinforcing its suitability for use in this context.We may recall Tedeshi's words that: 'we are considering truth relations with respect to a possible, albeit unrealized, future of a past time.' 84he remaining twenty, non-future main clauses with b-prefix that have so far been recorded do not display any obvious shared traits.It may be noted, however, that seven of the predicates encode a mild form of obligation corresponding to English should.This also falls within the realm of irrealis.A larger sample and further study on how this kind of obligations are generally encoded in Gulf Arabic is needed before conclusions on the use of the b-prefix in these few remaining cases can be drawn.

Summary and concluding comments
This study was undertaken to examine the use of the b-prefix in order to establish whether or not this is a tense-and/or a marker of the intentive mood as has been assumed.Other possible uses of the b-prefix were explored in order to find out what this prefix is used for if it is not a marker of future tense or intentive mood.Finally, I looked for other possible future markers than the b-prefix.
The research survey established the use of one rather obvious future marker, the particle rāḥ, at least in some parts of the dialectal area.This particle has received little attention in the literature on the Gulf Arabic dialects following Johnston's descriptions from 1967.On the other hand, the verb(s) baġa/yabi, that have been mentioned as indicators of future time in the nearby Najdi dialect, turned in the present database out to be used only for the volitive, viz. to express a wish and not to signal future time-reference.
However, the most important result of the study is that the b-prefix cannot be said to be a pure marker of future tense.Similar to so-called future markers in many other languages futurity is one--but not the only or necessarily the most important one--of its connotations.Although the b-prefix does co-occur with the future, it is much more consistently used in conditional clauses.Furthermore, it has a significant use in main clauses expressing the habitual past.The b-prefix does also not seem in any way to be an obvious marker of the intentive mood.The most important argument here is that it can occur in contexts where intention is impossible.
Instead, the main characteristic of the b-prefix is here suggested to be as a marker of irrealis/a marker of some notionally irrealis categories.This does not imply that Gulf Arabic has a binary system for marking the realis/irrealis distinction, only that it has a marker that can be used to mark some events as 'purely within the realm of thought, knowable only through imagination.'All three areas of language where the b-prefix is more or less systematically used -future time-reference, conditionals and past habituals -can be characterised as notionally irrealis.Thus, where the labels 'future' and 'intentive' fail, the notion of irrealis seems to account well for all the uses of b-as recorded in the rather large database that has been surveyed for this study.Continued work on tense, aspect and mood in Gulf Arabic, viz. the kind of research carried out within the larger project of which this treatise is a part, will hopefully throw more light on this and related questions such as the use of the bprefix in main clauses that do not express the habitual past.The data suggests that a broader study of how other notionally irrealis categories

PRE-PAPER JAIS INTERNET
are encoded in Gulf Arabic may be fruitful.Further research is also needed on other types of texts as well as on a broader range of speakers in terms of age and social environment to see if the use of b-as described here is language general or limited to certain types of discourse and/or classes of speakers.

Table 1 .
Distribution of the defined portion of corpus over countries

Table 2 .
Types of clauses with future time-reference in which b-, rāḥ and yabġi/yabi occur

Table 3 .
Types of clauses with future time-reference without any of the surveyed

Table 4 .
Types of non-future clauses where b-, rāḥ or baġa/yabi occurThe non-future contexts with b-or rāḥ comprise 27 sentences with present time-reference, 55 with past time-reference and 56 with no specific time-reference.The latter are mainly conditional clauses where relative tense is not stated.The 23 instances of baġa/yabi are divided on 16 sentences with present time-reference, 5 with past time-reference and 2 with no specific time-reference.

Table 5 .
Distribution of verb forms and markers in protases (no instances of rāḥ or similar)

Table 6 .
Distribution of verb forms and markers in apodoses

Table 7 .
Types of non-future, non-conditional clauses where b-, rāḥ and baġa/yabi occur 26 of the predicates were futures introduced by rāḥ, one by ḥaand one was preceded by rāyiḥ.Four are examples of the use of the verbs baġa/yabi to express volition or intention: 6. Lo wāḥid hindi hināk b-isawwi ʿalēna yikūn kalb yabi yiḏbaḥna ma naʿrif huwe šu yiqūl If one Indian there b-do3sg.m.p-stem on.usbe3sg.m.p-stem dog want3sg.m.p-stem sacrifice3sg.m.p-stem us not know1pl.p-stem he what say3sg.m.p-stemIf an Indian out there would do something against us, if he'd be a bastard, want to cheat us, we wouldn't know what he's saying. 56 Roberts 1994 159referring to a table inRoberts 1994, 31.8Palmer 2001, 163f.Palmer refers to Watkins 1984, 170-2where the marker is glossed as a future marker, and Mithun 1999, 173 who treats the same marker as irrealis.