Ways of expressing a future meaning: evidence from French child language Temptypac Workshop, March 11-12, 2010
Oana Lungu Universit de Nantes / LLING EA-3827
1. Introduction We present results of a parallel production and comprehension study investigating the construal of semantic future in complement clauses in French child language. We show that children produce / accept a future, an imparfait and a conditional (future in the past) in a future-shifted context (the embedded event occurs after the matrix event, but before the utterance-time) We argue that these findings bring evidence for the presence of zero-tenses (cf. Kratzer, 1998) in L1 French. A zero-tense in child French surfaces as either a present or a past below an explicit or an implicit modal operator. The data validate the Multiple Grammar approach advocated by Roeper (1999) and Yang (2000) according to which children go through a multiple grammar stage before they settle on a single grammar.
2. Background. Future in Complements of Speech Act Verbs
Mary will be busy.
1. John : 2. John said that a. Mary would be busy. b. .... Mary will be busy.
EV-T1 EV-T2 UT EV-T1 UT EV-T2 3a. [|][///////////]|> b. [|]|[////////////]>
SAY BUSY SAY BUSY Context 1: Future-shifted before UT Context 2: Future-shifted after UT
4. John said that Mary was busy. infelicitous in both Context 1 & Context 2 5. Temporal subsequence in embedded contexts gives rise to 2 possible configurations:
i. Ev-T2 is subsequent to Ev-T1 but prior to UT, as in (3a) ii. Ev-T2 is subsequent to both Ev-T1 and UT, as in (3b)
English-like languages use 2 morphologically distinct items to express semantic future: would : used to report a situation where Mary is busy at a time following Johns saying but anterior to UT ((5i)/(3a)), as well as a situation where Marys being busy obtains at a time following Johns saying & UT ((5ii)/(3b)) will : used only to report that Mary is busy at a time following both Johns saying time and the UT ((3b)) 3. Analysis of embedded future Abusch (1988) : will woll + present tense
would woll + past tense
Woll shifts the evaluation-time to the future. Kratzer (1998): the inventory of English tenses contains indexical (past and present) tenses and a zero-tense (-tense) Tenses are pronominal items (Partee, 1973, Heim, 1994, a.o). Just like pronouns they act as variables, which can be either free or bound. A -tense is a bound variable which surfaces as either a past or a present in certain embedded contexts (when c-commanded by a verb with the same temporal feature). 6. a. John PASTisay that Mary WOLL-i be busy. [(2a)]
b. John PASTisay that Mary WOLL-PRESj be busy. [(2b)]
(6a) contains an embedded -tense woll locates Ev-T2 in the future wrt Ev-T1 yields a dependent construal.
(6b) contains an embedded PRES woll locates EV-T2 in the future wrt UT yields an independent / indexical construal. 7. Constraint on the use of woll-PRES: PRES cannot be the realization of a -tense. Woll-PRES can only be used in contexts following UT. The indexicality of woll-PRES in a language like English follows from the assumption that present is always indexical (be it in embedded or simple/independent contexts). Languages that allow an embedded present to express temporal overlap with the Ev-T1 (i.e., so-called non-SOT languages such as Japanese, Korean) would not impose such a constraint on the use of an embedded woll-PRES.
4. Semantic future in child grammar Experimental study testing the construal of semantic future in complement clauses in
French child language (Elicited) production and comprehension (Truth Value Judgment Task, Crain and
Thornton, 1998) tasks with 14 kindergarten children between 3;8 and 5;4 (mean: 4;7) Stories acted out with props and toys in front of the child. The child is engaged in a game
where s/he is asked to interact with a puppet, Cronos, a penguin from another planet
Future shifted construals tested under 2 experimental conditions : i. Condition 1: Ev-2 occurs before UT ii. Condition 2: Ev-2 does not occur before UT
Predictions: Assuming French is like English, 2 options to express a future meaning: a future in the present equivalent to will FUTPRES a future in the past corresponding to would FUTPAST
Children who have acquired the Constraint in ((7)) should not produce / allow a future in the present under Condition 1
They can produce / accept future in the past under both conditions
The acquisition of tense in embedded contexts
Hollebrandse (2000) : the acquisition of Sequence of Tense by Dutch and American children < 5 year olds allow (4) in Context 1 ((3a)) as a felicitous report of Johns saying in (1)
Puzzle in Hollebrandses experiment : High rate of acceptance of the future-shifted (FS) reading of (4) in comprehension : ( 80%) Very low rate in production (8%) The production task was carried out with Dutch children only. Hollebrandse cannot explain the unexpected production findings. Hollebrandses explanation: children interpret the embedded past as expressing anteriority wrt UT rather than to Ev-T1 they allow all readings that are in the past wrt UT, including the non-adult FS reading in (3a). 8. Complementation Hypothesis:
The acceptance of the FS reading of (IMP) past is due to the lack of complementation (temporal dependence between the matrix and the embedded tenses)
(They would therefore interpret a sentence like John said that Mary was busy as John said something and Mary was busy.) 9. Two important stages in child tense grammar:
(i) Stage 1: children lack complementation (i.e., evaluation time for EV-2 is UT) independent construals
(ii) Stage 2: children have acquired complementation (i.e., the evaluation time for Ev-2 is Ev-T1) dependent construals
Predictions according to Hollebrandse: If children produce / accept the imparfait under Condition 1, they have not acquired
subordination These children should not accept /produce a future in the present under Condition 1 These children should not accept /produce an imparfait under Condition 2
Production task Sample stories 10. Condition 1 (Ev-2 occurs before UT) :
Peter and Tommy: Oh, look mom, a chimpanzee! We will give the chimpanzee a banana ! [the boys give the chimpanzee a banana].
Experimenter: What did the boys say about the chimpanzee? Cronos (the puppet): The boys said that
Oh, I forgot. Can you help me?
11. Condition 2 (Ev-2 does not occur before UT) : There are two little girls Lucy and Suzy, two big girls, Anne and Mary, and the twins Tommy and Peter. They are playing together. Tommy and Peter: We will jump over the trunk! [ The boys do not jump over the trunk.]
Cronos (the puppet): Peter and Tommy said that
Table 1 : % of responses in the production task
Condition 1 (Ev-2 occurs
before UT) Condition 2 (Ev-2 does not occur
Target Children Target Children
FUTPRES NO 21% YES 42%
PAST IMP NO 41% NO 17%
FUT PAST YES 26% YES 35%
Codes: FUTPRES = simple future; periphrastic future (aller in the present tense + infinitive); PASTIMP = imparfait; FUTPAST = future in the past (i.e., present conditional, aller in the imparfait + infinitive)
Future in the present (FUTPRES): Les garons ont dit quils donneront / vont donner une bananes au chimpanz. The boys said they will give a banana to the chimpanzee.
Children volunteer a FUTPRES under both Condition 1 (21%) & Condition 2 (42%) The results for Condition 2 are expected since the embedded event (the giving of the banana) does not occur by the end of the story FUTPRES is felicitous The results for Condition 1 (Ev-2 occurs before UT) are surprising and represent an apparent violation of the Constraint in (7). We will come back to this result later. Imparfait (PASTIMP):
Les garons ont dit quils ramassaient la neige. The boys said that they were collecting the snow.
Recall: Dutch children produce an imperfective past 8% of the time under Condition 1 (Ev-2 occurs before UT) (These) French children, unlike the Dutch children, produce a past (IMP) at a more significant rate (41%) under Condition 1 and they also produce it under Condition 2 (Ev-2 does not occur before UT) Dutch children do not produce an imperfective past under Condition 2 We will discuss these results and their implications shortly. Future in the past (FUTPAST):
Tommy et Pierre ont dit quils sauteraient/ allaient sauter par-dessus le coffre. Tommy and Peter said that they would jump over the trunk.
Finally, the production of FUTPAST is expected since FUTPAST is felicitous in both contexts. Comprehension Task Same experimental conditions Cronos, the puppet, utters the test-sentence with FUTPRES, PAST IMP, FUTPAST The child is asked to judge whether the puppets sentence is correct or not
Table 2: % of YES answers in the Comprehension Task
Condition 1 (Ev-2 occurs
before UT) Condition 2 (Ev-2 does not occur
Target Children Target Children
FUTPRES NO 88% YES 96%
PAST IMP NO 92% NO 91%
FUT PAST YES 94% YES 98%
The overall results in the comprehension task show an extremely high rate of acceptance ranging between 88% and 98%. We do not think that children did not understand the task since the control items we used showed they are capable of giving NO answers, but we admit there may be some other factors that may have induced a bias for a positive answer (i.e., The Principle of Charity) 5. Discussion 5.1. When future means future in the past FUTPRES was produced 21% of the time under Condition 1, where Ev-2 occurs before UT FUTPRES can only be construed as a dependent tense (expressing subsequence with respect to Ev-T1). Children have subordination. If French is like English and children have acquired subordination, why do they produce a FUTPRES in Condition 1 ?1 Because they have not acquired the constraint in (7) repeated below 7. Constraint on the use of woll-PRES: PRES cannot be the realization of a -tense. So, they do not know that, in French, FUTPRES / woll-PRES is only used when Ev-T1 follows the UT. 12. a. Pierre et Tommy ont dit quils donneront/vont donner une banane au chimpanz.
Peter and Tommy said that they give-FUT a banana to the chimpanzee. b. [IP1 PASTi [IP2 WOLL-i ].
(12b) is the representation of the target sentence in (12a) for Condition 1 ((10)).
Children allow the PRES component of the embedded future to be a -tense. Thus they assign an interpretation according to which Ev-T2 is in the future with respect to Ev-T1. Demirdache and Lungu (2008) : French children (aged 5-7) construe present and past as -tenses in both complement and relative clauses. BUT, notice that children also produced a FUTPAST in the same context (26%)! So, they do know that their language has this other option to express a future in the past meaning.
1 It would be interesting to know how English-speaking children behave in this situation. Unfortunately, we do not have the relevant data, except for one child, Amina (aged 4;5), who has never produced a simple future in this context. (Otherwise, Amina has a perfect tense grammar.)
In these childrens grammar woll- surfaces as either FUTPRES or FUTPAST (in the same context they pronounce a -tense as a PAST and as a PRES) We dont know of any other adult grammar having both these options. Children entertain more than one grammar plausible on the Multiple Grammar hypothesis (Roeper, 1999; Yang, 2000) according to which children consider more than one grammar before settling on a single grammar. Deviations from the target grammar are explained by the presence of default non-target grammars that have not yet been eliminated during language acquisition. Alternative explanation: children treat FUTPRES as an anaphoric tense Evidence: adult French Simple Future: English vs. French 13. a. Le roi mourra en 1715. a. The king would die in 1715.
b. #The king will die in 1715. b. #Le roi mourrait en 1715. The French future in (13a) unlike its English counterpart expresses subsequence with respect to a salient past time. In simple / unembedded contexts, French future patterns with would, not with will. French FUTPRES can be an anaphoric / dependent tense. English will is strictly indexical. 5.2. When imparfait means future Small detour Recall: the FS reading of past in complements of attitude verbs is unavailable This is a well-known fact which has received a lot of attention in the tense literature. FS readings of past are excluded by The Upper Limit Constraint (ULC): The local evaluation time is an upper limit for the reference of tenses (Abusch, 1997).
The ULC is taken to be universal. All languages are supposed to obey this constraint, and therefore not to allow the FS reading.
Kusumoto (1999) : apparent violations of ULC ( 2 cases where the FS reading of a past is possible) : Schedules events ((14a)) & future oriented verbs ((14b)): 14. a. The announcer said that the Red Sox played tomorrow.
b. John correctly predicted that the Red Sox won.
(14a) is used when the announcer actually said : Red Sox play tomorrow. (14b) with the future oriented verb predict is possible only when the prediction proves to be correct. Otherwise there is a preference for would :
15. John wrongly predicted that Red Sox *won/ would win. Examples such as (14a/b) are not really evidence against the ULC. In (14a) the embedded past may simply be the result of morphological agreement with the higher past (SOT environment) (14b) does not involve subordination.
(14b) can be reformulated as John correctly predicted the fact that Red Sox won. If (14b) involved subordination we wouldnt have expected a difference between (14b) & (15)
But, Italian imperfetto
16. Gianni ha detto (laltro ieri) che mangiava una banana (ieri / domani). G. said (the day before yesterday) that he eat-IMP a banana (yesterday/tomorrow). The imperfetto is used to express temporal subsequence with respect to both EV-T1 and UT2. Back to French children French children produce an IMP 41% of the time under Condition 1 (Ev-2 subsequent to Ev-1 but occurs before UT).
17. a. Pierre et Tommy ont dit quils donnaient une banane au chimpanzee. Peter and Tommy said that they give-IMP a banana to the chimpanzee.
- expected on Hollebrandses complementation hypothesis in (10) repeated below : 10. Complementation hypothesis:
The acceptance of the FS reading of an imperfective past is due to the lack of complementation (temporal dependence between the matrix and embedded tenses)
Conclusion following Hollebrandse: these children lack complementation. they have independent tenses in their grammar But, some of these same children also produce a future in the present under Condition 1 (21%) where the embedded tense can only yield a dependent construal ! not predicted by Hollebrandse Some of the children also produce an IMP under Condition 2 (17%) not predicted by Hollebrandse either. 18. Condition 2 :
The boys are in the garden. Oh, look it snowing, its snowing! Daddy looks desperate! The boys: Dont worry dad, we will shovel the snow ! [The embedded event does not take place].
a. Justine (4;9) : [Ils] ont dit quils ramassaient la neige, mais en fait ils ont chang davis. They [the boys] said that they shovel-IMP the snow but they changed
another child, Julie, not only did she produce an IMP for the story in (18) but also rejected a future and proposed an IMP, in the same condition, in the comprehension task : b. Julie (4; 3) : Non, elles [les filles] ont dit quelles volaient les chapeaux de pirates. No, they [the girls] said they steal-IMP the pirate hats. IMP in Condition 1 & 2 cannot be due to the lack of subordination French children have an Italian grammar
Again, children analyze the IMP as an inherently - tense below an implicit modal (WOLL)
2 The same seems to be true for Spanish imperfecto, at least in certain dialects of Spanish
19. a. The boys said that they shovel-IMP the snow. [(18a)] a. [IP1 PASTi [IP2 (WOLL)- i ].
b. They [the girls] said they steal-IMP the pirate hats. [(18b)] b. [IP1 PASTi [IP2 (WOLL)-i ].
The presence of the implicit modal operator is imputed to the properties of the French imparfait
French imparfait tense and modal 20. a. Un simple coup de tlphone et je venais tout de suite!
A simple phone call and I come-IMP immediately A simple phone call and I would immediately come.
b. Linstant daprs, le train draillait.
The next moment the train derail-IMP The next moment the train would have derailed. c. Linstant daprs, le train aurait draill. [Le Goffic, 1986] The next moment the train derail-PAST COND The next moment the train would have derailed.
Unlike the English imperfective past, French IMP is able to commute with conditional in certain contexts ((20a), (20b)). -however, even in modal contexts, the IMP retains the factual3 force of a past tense. Compare (20b) with its paraphrase in (20c), with a conditional. 6. Conclusions The present study argued for a unified account of the future and the imparfait as complex items containing a zero-past / present and a modal component. These findings corroborate previous results (see Demirdache & Lungu 2008) that revealed that 5-7 year old French children have and sometimes enforce zero-tense construals of the present and the imparfait in complement and relative clauses. The data are compatible with a Multiple Grammar approach to language acquisition (Roeper, 1999, Yang, 2000) according to which language acquisition is language competition between target and non-target grammars that have not yet been eliminated during the process of language learning.
3  limparfait fictif retient bel et bien une assertion sur le mode du certain, au mme titre que limparfait du pass effectif.  Do limpression de force, de sincrit, de vcu, de ce pass fictif: on rcrit le pass, on nonce ce qui, vridiquement, pouvait (devait, allait) tre  (bolds mine) [Le Goffic, 1986: 66]
References: Abusch, D. (1988). Sequence of Tense, Intensionality and Scope, WCCFL7, Stanford University Abusch, D. (1997). Sequence of Tense and Temporal De Re. In Linguistics and Philosophy, 20.1, Kluwer Academic Publishers: Netherlands Crain, S. and Thornton, R. (1998). Investigations in Universal Grammar: A Guide to Experiments in the Acquisition of Syntax and Semantics. The MIT Press: Cambridge, MA. Demirdache, H and Lungu, O. (2008). Sequence Of Tense in (French) Child Language. In J. van Craenenbroeck and J. Rooryck (eds), Linguistic Variation Yearbook 8, John Benjamins : Amsterdam/ Philadelphia Goffic (Le), P. (1986). Points de vue sur limparfait. Centre de publication de lUniversit de Caen Heim, I. (1994). Comments on Abuschs Theory of Tense. In Ellipsis, Tense and Questions, H. Kamp (ed), 143-170. DYANA, University of Amsterdam. Hollebrandse, B. (2000). The Acquisition of Tense. PhD Dissertation, University of Massachusetts at Amherst Kratzer, A. (1998). More Structural Analogies between Pronouns and Tenses. Proceedings of SALT VIII. CLC Publications, Cornell University, Ithaca, N.Y. Kusumoto, K. (1999). Tense in Embedded Contexts. Ph.D dissertation, Umass Amherst, Massachusetts. Partee, B. (1973). Some Structural Analogies Between Tenses and Pronouns in English. In Journal of Philosophy 70: 18. Roeper, T. (1999). Universal Bilingualism. In Bilingualism 2.3. Yang, C. (2000). Knowledge and Learning in Natural Language. PhD Dissertation, MIT.