Slabakova (2001)

Slabakova (2001) investigates L2 acquisition of English telicity by Bulgarian speakers. She discusses differences between English and Bulgarian telicity marking, as shown in (72)-(73).

(72) a. Clara ate cake. (ATELIC)

b. Clara ate a piece of cake. (TELIC)

(73) a. Bistra jad-e parče torta.

Bistra eat-PAST piece cake

‘Bistra ate at a piece of cake (but the piece is not finished).’ (ATELIC)

b. Bistra iz-jad-e parče torta.

Bistra PV-eat-PAST piece cake

‘Bistra ate a piece of cake (and there is nothing left).’ (TELIC)

(Slabakova 2001:4)

In English, the interpretation of situation aspect depends on verbal properties and specified quantity of the internal argument ‘SQA’ (Verkuyl 1999:36), such as cake vs.

a piece of cake in (72). Bulgarian telicity meaning, in contrast, is exclusively

determined by the absence/presence of the preverb (PV) on verbs, like iz- in (73b).

The lack of the overt affix in (73a) accounts for the atelic nature of the example, regardless of the properties of the direct object.

Slabakova ascribes the distinction between English and Bulgarian realization of telicity to a telicity marking parameter: an overt versus covert [+telic] morpheme. In English, the [+telic] morpheme is null, as represented in (74). It is situated in the head of AspectP (AspP) and is licensed by specified quantity of the argument [+SQA]. In the diagram, the DP direct object moves to [Spec, AspP] to receive accusative Case from the higher raised V⁰. The derived DP direct object c-commands the null morpheme Ø in Asp⁰, and thus its cardinality [±SQA] determines the telicity property of the null morpheme, on which the telicity interpretation of the entire VP crucially hinges.

(74) Null telicity marking in English VP

DPsubj V'

V AspP

CAUSE

Spec Asp' [±SQA]

Asp VP Ø[±telic]

DPobj V' V

[α telic] (Slabakova 2001:69)

However, the Bulgarian [+telic] morpheme is overt, as illustrated in (75).

(75) Overt telicity marking in Bulgarian VP

DPsubj V'

V PerfP

CAUSE

Perf AspP Preverb [±telic]

[+telic] Spec Asp' Asp VP

DPobj V' V

[α telic] (Slabakova 2001:86)

The overt morpheme is situated in the head of PerfectP (PerfP) and is realized by a

lexically selected preverb on verbs. Since Bulgarian telicity is marked by an overt affix in Perf⁰ that is structurally higher than the derived object position, the overt telic morpheme overrides the object with respect to the interpretation of situation aspect.

Four hypotheses regarding L2 acquisition of English telicity marking are advanced: (a) L2 learners will start out with the L1 value of the telicity marking parameter (L1 transfer); (b) they will show development towards the L2 value in their interlanguage stages, i.e. resetting is possible at later stages (full access to UG);¹⁸ (c) three purportedly telicity-associated constructions, viz. verb-particles, resultatives, and double objects, will co-occur with the acquisition of English telicity marking in their interlanguage grammar of individual learners; and (d) the three telicity-related constructions will also cluster together in the interlanguage grammar (Slabakova 2001:147-148).

To test her hypotheses, Slabakova conducted an experiment involving 130 Bulgarian adult learners of English and 32 English native speakers (16 North American and 16 British English speakers). The Bulgarian participants were divided into advanced, high-intermediate, and low-intermediate groups based on the results of a cloze test devised by Chen (1996).

There were four tasks in the experiment: an aspect task, a translation task, a stories task, and a grammaticality judgment task. The first three tasks focused on the English telicity interpretation, and the last one examined the L2 acquisition of the three telicity-related constructions. In the aspect task, the participants were asked to make acceptability judgment on 24 trial two-clause sentences plus 16 fillers. The trial sentences were classified into the following four conditions.

18 The full access proposal faithfully extends Chomsky’s UG framework to second language development. In this position, UG is taken to be fully accessible in second language acquisition in the sense that principles of UG will constrain L2 acquisition and that parameters inactivated in L1 can be reset in L2 acquisition ultimately (e.g. Schwartz and Sprouse 1994, 1996).

(76) a. Characteristic (atelic) + atelic

Sharon worked in a bakery and made cakes.

b. Characteristic (atelic) + telic

#Antonia worked in a bakery and made a cake.

(77) a. Atelic + atelic

Mr. Smith sold cars and now he sells motorcycles.

b. Telic + unfinished

#Mike drew a circle on a sheet of paper but the circle is only half-finished.

(Slabakova 2001:153)

In the translation task, the participants were asked to translate the verbal form of the English test clauses into Bulgarian, since Bulgarian marks telicity only with the preverb on verbal forms. As for the stories task, the participants were asked to read a short story and select one sentence from two that described the story better. The task included six stories depicting telic events, six establishing atelic context, and six fillers.

As far grammaticality judgment is concerned, the task consisted of 40 grammatical and 40 ungrammatical sentences, as shown below.

(78) a. The native men and women waited out the crisis. (VERB PARTICLE) b. *George loves out eggplant and basil pizza.

(79) a. Steven nailed all the top floor windows shut. (RESULTATIVE) b. *My friend Pamela feared the dinosaurs senseless.

(80) a. Simon gave Jenny a red scooter and a red hat. (DOUBLE OBJECT) b. *Sharon taught French the children in the neighborhood.

(Slabakova 2001:155)

The ungrammaticality of the sentences in (78b) and (79b) is due to the defective combination of a stative predicate with a telicity device, i.e. the verb particle up and the resultative senseless, respectively. The sentence (80b) is syntactically ill-formed as a result of flawed inversion between the two postverbal objects.

It was observed that lower proficiency learners were quite accurate in judging atelic sentences but were at chance in judging telic counterparts. Slabakova attributed the finding to the overt versus covert telicity marking contrasted between Bulgarian and English telic sentences, although the two languages are alike in the atelic sentences. Due to the L1-L2 similarity and difference, Bulgarian learners could perform better on English atelic sentences than telic ones. The results attested to the first hypothesis that L2 learners would transfer their L1 value of the telicity marking parameter at the early stage. Comparable to the performance of the native speakers, the higher proficiency participants were competent in differentiating between telic and atelic sentences and recognizing both types, which confirms the second hypothesis that resetting is possible at later stages. The findings appear to provide empirical evidence for the full transfer/full access hypothesis (Schwartz and Sprouse 1994, 1996).

As regards the three related constructions, it was found that all groups performed significantly less accurately than the controls and that individual scores on all three constructions were significantly correlated with each other at α=.05 when the influence of proficiency was factored out. The results obtained from the grammaticality judgment task showed that double objects might not cluster with the other two constructions. It was observed that both the advanced and high-intermediate groups were significantly more accurate at judging grammatical double objects (87%

and 84%, respectively) than transitive resultatives (74% and 60%) and verb-particles (71% and 66%) and that the low-intermediate learners performed mostly at chance,

with the exception of higher accuracy on grammatical double objects (71%). On the other hand, individual scores showed that each one of the three constructions was significantly well predicted by the other two constructions. Thus, Slabakova argued that individual scores lent mild support to the fourth hypothesis that the three constructions would cluster together in the interlanguage grammar.

Besides, Bulgarian learners’ skewed performance on telicity and the related constructions (i.e. performance of those who had acquired only either of them) posed a challenge for the third hypothesis that the three constructions would co-occur with the acquisition of English telicity marking in the interlanguage grammar, as shown in Table 2-2.

Table 2-2. Bulgarian learners’ acquisition of aspect and the related constructions

Related constructions Acquired Not acquired

Acquired 26

7

Not acquired

25

(Slabakova 2001:184)

Instead of viewing the seven learners who had acquired the related constructions but not telicity marking (5% of all participants) as a counter-example of the third hypothesis, Slabakova ascribed the skewed results to performance errors. For the 25 participants who had mastered telicity marking without acquiring the related constructions, Slabakova, inspired by Snyder (1996),¹⁹ claimed that the acquisition of telicity marking should be a necessary but not sufficient condition for the acquisition

19 Snyder (1996, 2001) conducted a crosslinguistic survey on the relationships between resultatives and productive N-N compounding, including American Sign Language, Austroasiatic (Khmer), Finno-Ugric (Hungarian), Germanic (English and German), Japanese-Korean (Japanese and Korean), Sino-Tibetan (Mandarin), Tai (Thai), Basque, Afroasiatic (Egyptian Arabic and Hebrew), Bantu (Lingala), Romance (French and Spanish), and Slavic (Russian and Serbo-Croation). It has been found that languages have either both resultatives and productive N-N compounding, or neither of them, with Basque as the only exception. Snyder argued that productive N-N compounding might be a member of the same cluster as resultative, verb-particle, and double object/dative constructions.

of the three constructions and that the acquisition of productive nominal compounding should be the other necessary condition. Nonetheless, an obligatory epiphenomenon of a parameter is a cluster of constructions, as Slabakova herself emphasizes. A question arises as to why only direct objects co-occur with telicity marking in the interlanguage if the three constructions belong to the same telicity-associated family.

Despite the innovation of the study, some limitations are noted as follows. In the aspect task, although the contrast shown in (76) is crystal clear when the pair are juxtaposed together, judgment of the trial sentences, without the contrast manifested, could be contingent on socio-pragmatic possibility²⁰ of them rather than good versus bad combination between telic and atelic clauses. Besides, the design of the contrast between (a-)telicity and (in-)completion may not be soundly conceived. Although the trial sentence in (77b) is well-designed, with semantic contradiction between the two conjoined clauses, the design of the sentence in (77a) appears to test the same property as that in (76) and is not parallel to that in (77b). Moreover, the aspect task as a whole focuses on the specification status of the internal NP argument only.

However, English telicity marking bears much on the lexical semantics of the main verb, but the indispensable property appears to be overlooked in the task.

Furthermore, in the grammaticality judgment task, the ability to judge the grammaticality of (78)-(80) does not necessary entail fledged acquisition of these constructions. Double object, verb-particle, and resultative constructions exhibit other essential characteristics, which are worthy of further investigation, together with the properties tested by Slabakova.

20 The judgment of the trial sentences is supposed to be based on good versus bad combination between two telic/atelic clauses in unmarked contexts. However, when the pair in (76) are not juxtaposed together, the contrast may not be clear and the subjects may judge the test sentences in terms of socio-pragmatic possibility—perhaps marked interpretations or contexts.

2.3.3 Summary

Snyder and Stromswold (1997) studied the child language acquisition of five English constructions, and Slabakova (2001) examined the L2 acquisition of English telicity marking and related constructions. These two studies have provided a light-shedding and empirically feasible framework for further investigation on telicity marking, controversially pertinent constructions, and parametric variations.

As seminal the two studies are, they are subject to some limitations. The L1 study adopted a quantitative analysis of the longitudinal data, and the results showed some correlation of the target constructions. However, the question remains unanswered as to whether or not some constructions emerge in some children’s speech around the same time as other constructions are recognized in other children’s, which might lead to some constructional correlation. Hence, it is worthwhile to probe into a qualitative analysis of individual children’s data.

In the L2 study, Slabakova employed a grammaticality judgment task to look into the L2 acquisition of the three telicity-related constructions in English. However, the comprehension task was restricted to one property of each of the constructions, which are characterized by other crucial properties. Moreover, the quantitative study analyzed group results as well as individual scores, and the individual data showed some correlation of the three constructions. Nonetheless, Slabakova’s study, adopting only one comprehension task, may be subject to a task effect. A production task cannot only circumvent such a nonlinguistic effect, but a qualitative analysis of production data may also lead to a generalization of error patterns and shed new light on the relationships of the constructions. Given the limitations of the previous research, the relationships of the telicity-associated constructions merit further empirical investigation.