Chapter 4: Event decomposition and causativized stative verbs
4.5. Future and progressive causativized stative constructions
In this section, I will apply the theory elaborated during the present chapter to future and progressive contexts. According to the stipulation, I adopted this thesis: accusative Case is licensed by the closest c-commanding v°/V° linked to T°.
The following sentence shows that the proposal made earlier, and therefore the stipulation, both predictably apply to future contexts as well. Consider sentence (92):
(92) le-kteb heḏa beš yḥabbeb-ek fī-l-muṭālʕa the-book this fut make_love.imp-you fi-the-literature
‘This book will make you love the literature.’
The sentence in (92) shows that causativized stative sentences in the future present us with mismatching Cases on the direct objects; notably the Case configuration found in (92) is the same as the one found in perfective sentences of the same kind, whereby the experiencer object is assigned accusative Case but the theme object requires the insertion of aspectual fi.
Case mismatches of this kind arise in perfective sentences because the perfective operator can bind the event variable associated with the causative subevent but not the lowest variables which occur in the scope of OPGen. I want to argue in this section that the same reasoning applies to example (92), with the difference that perfective aspect is not realized by the main verb but is associated with the future morpheme.
Tunisian expresses future reference by means of a preverbal particle. This particle has various dialectal realizations (see Mion 2017 for a detailed description of the expression of future in Tunisian); I will take the particle beš as the standard for ease of explanation. Tunisian beš possibly derives from the word mēši, which is the active participle of the verb ‘to go’. While
active participles in Tunisian present person and number agreement, beš is invariable, suggesting that it has undergone a process of grammaticalization. The use of a participle for the expression of future time is quite typical of modern Arabic varieties which have systematically replaced Classical Arabic future forms with periphrastic constructions requiring an imperfective lexical verb and a future marking element derived from a verb of motion (cf. Benmamoun 2000; Aoun, Benmamoun, and Choueiri 2009).
I propose that Tunisian beš occurs in sentences that express future time and perfective aspect.
According to Demirdache and Uribe-Etxebarria (2007), perfective future derives sequential readings with temporal adjunct clauses; while future time associated with a neutral viewpoint aspect in the sense of Smith (2013) allows the overlapping reading. As example (93) illustrates, Tunisian sentences with beš followed by a lexical verb in the imperfective definitely do not allow an overlapping reading:
(93) waqtelli beš nūṣṣel le-d-dār semi beš yerqod when will arrive.imp to-the-house Semi will sleep.imp
‘When I will arrive home, Semi will sleep’
# ‘When I will arrive home, Semi will be sleeping’ Sequential reading only Overlapping reading and future reference is instead expressed by a different type of construction in which beš is followed by the imperfective auxiliary, a possible adverb marking anteriority and, finally, the lexical predicate in the participial form, as in (94):
(94) waqtelli beš nūṣṣel le-d-dār semi beš yekūn (dežā) rāqed when will arrive.imp to-the-house Semi will be.imp already sleep.prtcpl
‘When I will arrive home, Semi will already be asleep’
# ‘When I will arrive home, Semi will sleep’ Overlapping reading only
The contrast illustrated by sentences (93) and (94) shows that the overlapping reading (in which the event described by the main predicate begins before the arrival time) is only possible in the presence of a periphrastic construction that involves an auxiliary and a participial predicate. The simple future construction, i.e. (93), in which the future marker directly precedes the imperfective lexical verb, provides a sequential interpretation. According to Demirdache and Uribe-Etxebarria’s (o.c.) diagnostic, the sequential interpretation is associated with perfective aspect.
Hence, I conclude that simple future sentences express perfective aspect and future time reference6.
With these notions in mind, we can now proceed to see how Case licensing operates in sentences like (92), meaning simple future sentences whose predicate is a causativized stative verb. The above diagnostic shows that Tunisian future is perfective and, consequently, I propose that sentences of this kind are associated to a perfective operator and future time.
6 Future sentences, i.e. constructions of the type beš + imperfective verb, marginally allow a prospective reading. I will disregard this issue and only deal with the cases in which beš appears to encode future
Both the perfective feature and future tense bind the event variable of the causative subevent.
Thus, sentence (92) presents one event variable valued by perfective aspect and future. On the other hand, the two variables associated with the lowest verbal projections are inherently bound by OPGen. The analysis from this point on is the same as past perfective contexts, with the only relevant difference being that the moment of speech and reference time have their order inverted.
See the representation in (95):
Just like in past perfective contexts, the experiencer object of a future sentence whose predicate is a causativized stative predicate is Case licensed by a verbal head whose event variable is valued by the same operator that defines reference time, namely beš. The theme object is also fi marked in this case, since OPGen acts as an intervener and prevents the variable in its binding domain by acquiring all the features that verbal heads need in order to avoid being defective. This results in the following syntactic representation, i.e. (92), which corresponds to sentence (96):
Incidentally, the analysis of future sentences presented above confirms the hypothesis that imperfective verbal forms in Tunisian are not endowed with imperfective aspect, but are default verb forms which are temporally and aspectually underspecified, since they can occur in contexts requiring operators of opposite types: imperfective verbs, in fact, are compatible with perfective
aspect in future contexts and with the absence of perfective aspect in progressive and generic sentences.
Finally, contexts in Tunisian require the presence of qāʕed, which is optionally deleted at PF. As discussed in chapter 2 section 5.4, progressive aspect in Tunisian is only compatible with morphologically imperfective verbs. The following example (97) illustrates that the direct objects of progressive sentences whose predicate is a causativized stative verb are uniformly fi marked:
(97) ?? tkarreh fi-yya fī-l-kosksi make_hate.imp fi-me fi-the-couscous
‘She is making me hate couscous.’
(Adapted from Brahim 2007, 99: (15)b)
The comparison between (97) and (98) illustrates that there is no difference in the Case licensing of progressive sentences whose underlying predicate is stative or dynamic, since both predicates in the two sentences are fi marked.
(98) semi (qāʕed) yḍawwak fī-yya fī-šoklata Semi prog make_taste.imp fi-me.dat fi-chocolate
‘Semi is making me taste chocolate.’
According to the theory presented in the present chapter, fi marked objects indirectly result from the presence of a defective v° and verbal heads are defective if they do not occur in a complete T-chain. Examples (97) and (98) show that the causative head fails to license accusative Case in progressive contexts in the lower v° projection as well as perfective causativized verbs.
However, while we know that OPgen is the reason for the defectiveness of the heads in its domain, it is somewhat surprising to notice the defectiveness of the head associated with the causative subevent that is in the domain of the sentence aspectual operator, at least according to the assumption adopted so far. Assuming that defectiveness is always the result of intervention, the presence of fi marked objects in progressive contexts must involve intervention too.
This result is unexpected when we look at perfective sentences, since they only show the presence of one operator, i.e. qāʕed, in their overt structure. In order to solve this puzzle, therefore, I propose that progressive aspect entails a higher degree of syntactic complexity and that the progressive operator is not itself the element that puts in relation T° to (e).
As Harwood (2015) suggests, there is something that makes progressive aspect unlike any other aspectual distinction. Arche (2014) proposes that the peculiarity lies in its ability to put two viewpoints in relation to aspect, which is also the point beyond the locative approach discussed in chapter 2. According to Bybee and al.’s (1994) theory, in effect, the progressive aspect is inherently associated to a locative relation which, following Fontanals and Simon (1999), is syntactically encoded as a predicative relation whereby a subject is located within an event.
I will refine this idea in the next chapter; however, in light of the direction taken by the authors of the above works, I will assume that progressive entails two operators: qāʕed linked to T° and a lower operator that does not have an overt realization which bounds the variable in the event structures and does not form a chain with T°.
Since the event variables present in the vP are all valued by the lower operator, intervention is inherent in progressive contexts. This means that every verbal head associated with a dynamic predicate or with a causativized stative predicate is unavoidably defective, since intervention occurs above the vP domain.
This means that sentences like (97) underlie the syntactic structure in (99):
(99) TP AspP
fī-yyaobl v° VP
fī-l-muṭālʕa obl V°
Since the two verbal heads responsible for the licensing of Case on the objects are associated with operators that are different from one another and different from the one that defines reference time, i.e. OPx and OPgen, Case on the objects of a causativized transitive stative predicate in a progressive sentence is only licensed via aspectual fi insertion.
The system I elaborated above aims to account for the intuition that DPs are involved in the subevent projected by the head which licenses its Case in a special way. We can think of this relation in the following terms: an argument DP1 in the specifier of a subevent projecting head (e1) is inherently involved in that subevent, so that the subevent both starts and ends with it.
(100) VP DP1 V’
This means that DP1 in (100) is involved in the event described by V° for the exact duration of its run time. We can take the example of mono-argumental predicates: the only argument of these
verbs occurs in the specifier of the only subevent present in the structure according to the principle that there is a one-to-one correspondence between the number of arguments present in a structure and the number of its subevents. This argument is understood as participating in the subevent that projects it for the entire subevent run time: a dying being participates in the dying event for the whole duration of the event. According to the same logic, a runner must also run in order that the running takes place, that the sinking boat must sink etc.; in these events, whenever the participant drops out, the event ceases to take place. There is no possible way to disentangle the duration of the subevent and the involvement of the projected argument.
When we consider a dyadic predicate, there are two arguments which are involved in the same subevent, so that in an eating event, the agent and the theme are involved together throughout the run time of the event. I think of accusative Case as expressing the relationship between the run time of a subevent and the involvement of a participant that is not base generated in the specifier of the corresponding projection. This means that in a structure like (101), v°
licenses accusative Case on DP2 and links this participant to its run time.
(101) vP DP1 v’
v°(e1) VP DP2 V’
There are languages like Italian or English where accusative Case is licensed across the board to all objects. We could say that in these languages, events are always perceived as being internally uniform with respect to time: all the events take place at the same point in time and for the same amount of time.
Other languages, like Tunisian and Icelandic, can conceptualize a temporal distinction in the length of the subevents and make it superficially visible in syntax via their Case system. In these languages, it is possible to describe a complex event in which some participants can conclude its involvement before the others do, without having to say that the event ceased because that argument dropped out.
Svenonius (2013) already mentioned that examples with ballistic motion verbs illustrate this linguistic property in a clear way:
(102) a. skjóta fuglinn ‘shoot the bird’ (acc) b. skjóta kúlunni ‘shoot the bullet’ (dat) c. skutla hvalinn ‘harpoon the whale’ (acc) d. skutla skutlinum ‘throw the harpoon’ (dat) (Icelandic, Svenonius 2013, 4: (7)a-d)
The agent of shooting or throwing does not provide any contribution to the movement of the projectile after the initial release of power that induced it. The projectile continues the movement along its trajectory alone (cf. (102)a). This is also true when a bird is shot by a bullet, but in this case the relevant subevent does not introduce the projectile as its argument, but it introduces the