5.7.1 Relative future tense uses
That lia can denote relative future tense has been overlooked in the literature on the Xining dialect, but Zhao (2015) mentions this use for the Gangou dialect. Relative future tense takes a contextually-determined time other than the utterance time as its reference point, and this can be denoted by lia, which in complex clauses can express the ‘future in the past’ (cf. Comrie, 1985: 74-5). In Klein’s (1994: 2) framework, aspect marking specifies the relation between TT and TSit in terms of precedence, simultaneity, or subsequence. Since lia lacks certain other properties associated with the prospective aspect cross-linguistically (see below), it is argued that the label 'relative future tense' is appropriate for 'future in the past' uses (i.e. TT < TSit < TU). A relative future use of lia is illustrated in (246):
zuotian no yaoshi Xining qu-le (*/lia) dehua yesterday 1SG if Xining go-PFV (*lia) COND
no node pengyou ha jian-ha lia
1SG 1SG.POSS friend OBL see-COMPL lia (Elicited) ‘If I had gone to Xining yesterday, I would have seen my friend.’
This is a counterfactual past conditional, which involves an event that did not happen, but the speaker hypothesizes about what the consequences might have been if the event had taken place. Future markers can be found in counterfactuals cross-linguistically, for example in Zulu (Halpert 2010), and also in Standard Mandarin where the modal hui can be used (Su 2008). In Xining Mandarin, lia marks the unrealized consequence clause, and cannot occur on the counterfactual past clause in place of the perfective marker le. This type of usage is
consistent with an analysis of lia as relative future tense because the event of the main clause (TSit) occurs after the event of the conditional clause (TT) and so TSit is future with regard to TT (in (246), seeing the friend follows going to Xining). In the next section, two possible alternative analyses of lia in terms of irrealis mood and prospective aspect will be examined. 5.7.2 Lia is not an irrealis marker or a prospective aspect marker
Given that irrealis marking is also used in conditional clauses and future contexts cross- lingustically, it is worth considering the possibility of treating lia as an irrealis marker.
However, an important argument against analyzing lia as irrealis is that it is only permissible on conditional clauses that have future time reference (relative or absolute future). Irrealis marking, on the other hand, would be expected to be possible in past conditional clauses as well, since it simply marks non-reality status rather than temporal reference. This is illustrated by two unrelated languages, Iatmul (Papuan) and Russian:
saanya kla-ikiya-j-a-n nyigi gu kla-ikiya-di money get-IRR-3PL-SR-NR bitter water get-IRR-3PL
‘If they had received money, they would have bought beer’ (Jendraschek 2014: 154) Esli by ja pribyl na vokzal, menja by posadili v tjur'mu
if I arrive(IRR) at station me put(IRR) in prison 'If I had shown up at the station, they would have thrown me in prison.'
(Chung and Timberlake 1985: 251)
By contrast, as shown in (246) above, lia cannot occur in past conditional clauses. This means that although lia shows some functional overlap with irrealis markers in that both mark future and conditional clauses, the core future semantics of lia constrain which irrealis contexts it can occur in.
An additional indication that lia is not an irrealis marker is that, as with Standard Mandarin hui (Su 2008), in condition clauses an indicative interpretation is available in addition to a counterfactual reading, which shows that the semantic contribution of lia is independent of reality status:
zuotian ni yaoshi Xining qu-le dehua, ni no de Yesterday 2SG if Xining go-PFV COND, 2SG 1SG POSS pengyou ha jian-ha lia
friend OBL see-COMPL lia
‘If you went to Xining yesterday, you will have seen my friend.’ (IND) ‘If you had gone to Xining yesterday, you would have seen my friend.’ (CF)
(Elicited) On the indicative reading, lia occurs in a realis rather than irrealis predicate, but on the counterfactual reading, it marks an irrealis clause. But the core futurity requirement is satisfied on both readings, since the lia marked clause temporally follows the topic time denoted by the conditional clause. Again, this indicates that lia is constrained to future contexts (here, relative future) rather than to irrealis contexts.
Finally, the possibility that lia could be a prospective aspect marker rather than a future tense marker also deserves consideration, since this also superficially resembles future marking. In the 'prospective aspect', a situation is viewed from the perspective of an earlier reference time (Comrie 1976: 64-7): a future event may be viewed from the perspective of the present, or a past event may be viewed from the perspective of an earlier point in the past. Prospective aspect is thus a ‘perspectival’ aspect (Dik 1997: 238-41), because it fulfils a temporal (tense) function of locating time spans relative to each other. But despite the tense-like function of prospective aspect, this category nevertheless differs from true tense, and lia shows the properties expected for tense rather than prospective aspect.
Firstly, lia is not associated with the truth conditions expected of the prospective aspect, but rather with those of the straightforward future tense. Comrie (1976: 64-5) notes that if an event expressed with prospective aspect does not occur, then the statement has not been falsified, whereas with the straightforward future, if the event does not occur, then the statement has been falsified. That is, the prospective aspect simply denotes 'the already present seeds of some future situation, which future situation might well be prevented from coming about by intervening factors' (Ibid. 65). Comrie illustrates this contrast as follows, where (250) is prospective aspect and (251) is future tense, and only the latter asserts that the event will certainly come about:
Bill is going to throw himself off the cliff. Bill will throw himself off the cliff.
This distinction exists because prospective aspect carries a caveat that the statement is made not absolutely, but ‘on the basis of what can be known at the (earlier) reference time’ (Dik 1997: 239). Lia does not possess these properties of prospective aspect. For example, if (252) is uttered, and then the event does not occur, (252) has simply been falsified.
Lao Bi mingtian Xining qu lia NAME tomorrow Xining go lia
‘Lao Bi will go to Xining tomorrow.’ (Elicited)
A second piece of evidence that lia denotes future and not prospective aspect is that simple clauses marked with lia are not compatible with past time reference, even in a suitable discourse context. Thus the simple clause in (254) cannot be used when, as here, TT is specified by the question as prior to TU.
Zuotian san dian tianqi amen le yesterday three o'clock weather how PFV
‘What was the weather like at 3 o’clock yesterday?’ (Elicited) #tianqi re lia
weather hot lia
(intended) ‘The weather was about to warm up.’ (Elicited)
By contrast, past time reference is acceptable for prospective aspect in simple clauses in languages with this category, even in out of the blue contexts, as illustrated for Turkish by (255) and for English in (256):
çalış-acak-Ø-tı
work-PROSP-COP-PAST
‘I was going to work.’ (Key and Schreiner 2014) Socrates was going to die. (Klein 1994: 116)
In summary, the properties of lia are those of a future tense marker, as assumed in the literature and in this chapter, rather than an irrealis mood marker or a prospective aspect marker. In the next section, the origin of lia is discussed.