1 Specificity Affects Determiner Choice Even When Definiteness Transfers
Asya Achimova and Viviane DĆ©prez
1 Introduction
Specificity and definiteness, Ionin et al. (2004) argued, are two features that parametrically determine article lexicalization choice in the languages of the world. To account for the characteristic specificity effect that describes errors that second language (L2) learners have been observed to make in their determiner uses, Ionin et al. (2004) proposed that second language learners with determiner-less first language (L1) access the Article Choice Parameter available from Universal Grammar (UG) and initially fluctuate between its definiteness versus specificity settings. After an initial fluctuation phase, L2 learners fix the parameter, settling on the appropriate value, so that specificity driven misuse disappears. Second language research on the acquisition of determiners has subsequently been concerned with the question of whether the article parameter remains accessible to L2 learners.
In this chapter, we argue that specificity effects on determiner acquisition, unlike definiteness effects, are not due to fluctuating access of a parametrized universal semantic feature system, as they appear to occur even in cases of predicted transfer, when the L1 and L2 both feature the same parameter setting. Here, we look at the acquisition of L2 French by native speakers of English. Both languages have articles that mark definite versus indefinite noun phrases (NPs). We explore the possibility of full transfer of the article system from the L1 to the L2. However, as we show, the acquisition data also reveal an effect of specificity in some contexts. The rest of the paper is structured as follows. In Sections 1.1 and 1.2, we review the Article Choice Parameter hypothesis and related second language acquisition studies. Section 2 introduces our experimental techniques, followed by results in Section 3 and a discussion of the possible source of the observed specificity effect in Section 4. In Section 5, we conclude that specificity should be viewed as a pragmatic notion rather than a grammaticalized parameter.
1.1 The Article Choice Parameter
Acquisition of determiners is a well-known challenge for speakers of languages that lack overt articles. Such L2 learners have been observed to overuse the definite article in [+specific, ādefinite] contexts (Huebner, 1983; Master, 1987; Parrish, 1987; Thomas, 1989), and overuse the indefinite article in [āspecific, +definite] contexts (Ionin et al., 2004; Leung, 2001). Ionin et al. (2004) developed an account that traces these two types of errors to the same source: learners initially fluctuate between the two settings of the Article Choice Parameter and therefore sometimes use the definite determiner to mark [+definite] and sometimes [+specific] NPs. In the developing L2 grammars, the indefinite determiner marks either [ādefinite] or [āspecific] NPs. This view of early stages of language acquisition became known as the Fluctuation Hypothesis.
Ionin et al. (2004) conceptualize definiteness and specificity as semantic features that they informally define as follows (Ionin et al., 2004: 5) (for formal definitions, see Heim, 1982):
ā¢ If a determiner phrase (DP) of the form [D NP] is [+definite], the speaker and the hearer presuppose the existence of a unique individual in the set denoted by the NP.
ā¢ If a DP is the form [D NP] is [+specific], the speaker intends to refer to a unique individual in the set denoted by the NP, and considers this individual to possess some noteworthy property.
In English, the definite determiner the marks definite NPs, independent of whether these NPs are specific (1a) or nonspecific (1b) (examples from Ionin et al., 2004: 8). As the following examples show, the definite article the is used to encode uniqueness whether or not the referent is known to the speaker with a noteworthy property.
(1) a. Iād like to talk to the winner of todayās race ā she is my best friend!
b. Iād like to talk to the winner of todayās race whoever that is; Iām writing a story about this race for the newspaper.
Unlike English, languages such as Samoan, according to Ionin et al. (2004), have determiner systems based on specificity. Thus, the Samoan article le marks specific singular DPs, while the article se appears with nonspecific singular DPs (Mosel & Hovdhaugen, 1992). Plural DPs are not marked.
Nevertheless, the validity of this classification is questioned by Tryzna (2009). She conducted an experiment to find out whether the article se in Samoan could appear in definite nonspecific contexts. If Samoan lexicalizes articles based on their specificity values, se is predicted to be able to occur in nonspecific definite and indefinite contexts, and le in specific definite and indefinite contexts. Tryzna (2009) created a data-elicitation questionnaire. The test sentences in Samoan contained DPs in four contexts: specific definite, specific indefinite, nonspecific definite, and nonspecific indefinite ones. The Article Choice Parameter (Ionin et al., 2004) predicts that se should appear in definite nonspecific contexts. However, as Tryzna (2009) shows, the nonspecific definite context requires the article le, as in (2). The article se only appears in nonspecific indefinite contexts.
(2) Aāfai āete manaāo āe tautala i*se/le malo faāatali seāi uma le tautuuna.
āIf you want to talk to the winner, stay until the race is over.ā
(Tryzna, 2009: 72)
The author proposed the following system of articles in Samoan (see Table 1.1) (Tryzna, 2009: 71).
Additional evidence that questions the status of specificity as a grammaticalized feature come from typological studies. Contrary to early accounts of the article systems in French-based Creoles, recent studies (e.g. DĆ©prez, 2011) show that definite articles encode familiarity ā that is, presupposed unique existence by speaker and hearer ā rather than specificity, and that indefinite determiners can be used in both specific and nonspecific contexts (see also DĆ©prez, 2013, 2016).
Ionin et al. (2009) addressed this cross-linguistic evidence by modifying their original Fluctuation Hypothesis. They argued that the article system in Samoan is more complex than initially assumed. Samoan uses the article se to mark NPs in nonspecific indefinite contexts, while le appears with specific indefinites and all definite NPs (specific and nonspecific). Drawing from typological work on other topics, Ionin et al. (2009) showed that a similar division is relevant in a number of domains. Thus, specific indefinite and definite NPs require the same marker in Spanish, for example, the dative preposition a ātoā (Aissen, 2003; Leonetti, 2004; Torrego, 1998) receive accusative case marking in Turkish (EnƧ, 1991; Kelepir, 2001), and trigger the appearance of an initial vowel in nominals in Luganda (Ferrari-Bridgers, 2004). Ionin et al. (2009) note that in their 2008 experiments (Ionin et al., 2008), L1 Russian speakers acquiring L2 English made more specificity-related errors with indefinites than with definites. The authors attribute the larger number of errors with indefinites to the natural language pattern described above: specificity distinguishes between different classes of indefinites. Here the specific indefinites pattern with definites and appear with the definite article, and nonspecific indefinites are marked with a separate article (the indefinite one).
Table 1.1 Specificity and definiteness interaction in Samoan
Context type | An example of a test sentence (target DP in bold) | The corresponding Samoan DP |
1. Nonspecific indefinite | Iām looking for a hat to go with my new coat. | se polou |
2. Specific indefinite | Iām looking for a hat. I must have left it here yesterday. | le polou |
3. Specific definite | I want to talk to the winner of the race. She is a good friend of mine. | le malo |
4. Nonspecific definite | If you want to talk to the winner, wait until the end of the race. | le malo |
In sum, the updated version of the Fluctuation Hypothesis (Ionin et al., 2008) makes a different set of predictions. If L2 learners have access to the semantic features in UG, they should make errors with specific indefinites, namely manifest overuse of the definite the in [-definite, +specific] contexts ā as this is an option available in natural languages, such as Samoan. At the same time, errors with nonspecific definites should not occur, since, according to Ionin et al. (2009), no known language makes an article distinction between specific and nonspecific definite NPs. In fact, Ionin et al. (2009) discovered that children made more specificity-related errors with indefinites, while adults made such errors both with indefinites and definites. The authors argue that the adultsā data may show the effect of explicit strategies that adult learners apply in choosing an article.
Further studies on the acquisition of articles show that there are other factors that may influence article choice in L2 learners of languages, such as English. Ionin et al. (2012) examine how speakers transfer the semantics of demonstratives into their L2. For example, native speakers of Korean learning L2 English show different preferences in choosing between the definite article the and a demonstrative that, possibly reflecting the influence of their L1 that only has demonstratives and lacks overt articles.
1.2 Second language acquisition studies
According to Ionin et al. (2004), speakers of article-less languages should fluctuate between definiteness and specificity-based article systems at the initial stages of L2 acquisition; however, not all studies have found evidence in support of the fluctuation hypothesis. Tryzna (2009) examined the acquisition of English determiners in L1 speakers of Mandarin and Polish. Both languages lack determiners, so the Fluctuation Hypothesis predicts that speakers should alternate between using the definite article the to mark definite or specific NPs. In a forced choice elicitation task, participants had to complete the dialogues where determiners were missing. The results appeared more complex than predicted by the Fluctuation Hypothesis alone. Chinese L1 speakers either adopted the target ādefinitenessā setting or showed a fluctuation pattern. Polish speakers, on the other hand, showed a greater variety of strategies. While 21% of the advanced learners followed the fluctuation pattern, 26% showed optional use of the both with specific indefinites and definites. In the intermediate group, none of the speakers adopted the āspecificityā setting, with 11% following the fluctuation pattern and 53% using the optionally with all NPs. Tryzna (2009) concluded that the expected overuse of the in specific indefinite contexts was by far not the only pattern of errors.
Tryzna (2009) also made a proposal about the nature of specificity as a semantic feature. Since L2 learners do not overuse the with nonspecific indefinite NPs, they must have access to the specificity value; otherwise, we would expect no difference in the overuse of the with specific versus nonspecific DPs. Consequently, she views specificity as a universal semantic feature available to L2 learners regardless of their L1. The latter conclusion, however, seems to make incorrect predictions. If both specificity and definiteness are universal semantic features available to the learner, why is it that only some learners fluctuate between these settings in developing an L2 determiner system?
In another study, Jaensch (2009) looked at L3 acquisition of German by L1 speakers of Japanese and L2 learners of English. She found that Japanese speakers did not fluctuate between the settings of the Article Choice Parameter. Speakers did not successfully transfer their L2 article system into L3, and frequently omitted the articles, especially in the oral production task. Thus, this study provides only partial evidence in favor of the Fluctuation Hypothesis. As the updated Fluctuation Hypothesis (Ionin et al., 2009) predicts, learners in Jaensch (2009) behaved in a more target-like manner in definite contexts than in indefinite ones. Specificity had an effect on the article misuse, but only in definite contexts: when the NP was nonspecific there was a higher rate of the indefinite article used.
The studies reviewed above focus on the acquisition of articles by speakers of article-less languages. We would now like to turn to situations where a speaker of an L1 with articles acquires an L2 with an article system. Such studies can shed light on the problem of transfer: do semantic features transfer to L2? Sarko (2009) examined L2 English acquisition by native speakers of French and Syrian Arabic. Both languages have articles, yet French is different from English as French requires overt articles for both singular and plural definites and indefinites. The article system of Syrian Arabic is also different from English, as Arabic lacks an overt article for indefinites singular and plural. In a forced choice elicitation task, Arabic speakers picked an incorrect article the in indefinite specific contexts (count singular nouns: 31% in the intermediate group, 23% in the advanced group), while the French speakers did not (5% and 3% respectively). Sarko (2009) attributes these errors to L1 transfer and interprets the findings as evidence in favor of the Full Transfer Full Access Hypothesis (Schwartz & Sprouse, 1994). According to this hypothesis, L2 learners have full access to the possible parameters of any grammatical feature in a language. At the same ti...