The Count-Mass Distinction and English Articles

SECOND LANGUAGE Vol. 18 (2019, pp. 47–69) The Japan Second Language Association

http://www.j-sla.org/

- 47 -

The Count-Mass Distinction and English Articles 可算・不可算の区別と冠詞

Mutsumi Ogawa Nihon University 小川睦美 日本大学

Abstract This article explores the effect of the count-mass distinction in English article acquisition and argues for its importance in the research of second language (L2) article acquisition. While L2 article use has extensively been discussed in terms of semantic contexts (i.e., definiteness, specificity) and L1 effects, such factors as input and lexical information of nouns have not been given much attention in the literature. The article first overviews the previous research based on semantic contexts and points out the L2 phenomena which cannot be attributed solely to definiteness or specificity. Turning to the data of input frequency and cues, such phenomena seem to be plausibly explained by the frequency distribution of noun types and forms in the input. Next, the article argues that the count-mass distinction poses a more persistent problem than definiteness, by examining to what extent each feature involved in article choice (e.g., [±definite], [±count], [±plural]) contributes to L2 learners’ use of articles. After claiming the prolonged difficulty with the count-mass distinction, the article addresses acquisition issues of abstract nouns, for which L2 learners often have trouble choosing correct articles, especially indefinite ones (a/an or ø). An experiment was set up to investigate whether or not the lexical-semantic property of boundedness and/or the derivation type of verb-derived abstract nouns were potential factors to influence the countability of nouns. As a result, the two factors did not categorically affect the countability judgment of abstract nouns, rather indicating a possibility of L1 influence to be further investigated. At last, the article attempts to verify the claim that L2 learners, especially with low proficiency, determine the countability of nouns by intuition without considering contextual information. The comparison of L2 learners’ intuitive countability judgment of decontextulised nouns with the use of articles in context revealed that a majority of learners demonstrated no such correlation between countability intuition and article use. Although there were some who showed reliance on intuition, their L2 proficiency was not the discriminative factor to identify what kind of learners would strongly draw on intuition for countability judgment of nouns. In conclusion, this article emphasises that the effect of nominal properties, which are susceptible to L1 influence, can interact in the use of

Second Language 18

- 48 -

articles and require careful empirical scrutiny in order to grasp a whole picture of L2 article acquisition.

要旨

本論文では，英語の冠詞習得における可算・不可算の区別の影響について調査し，第二言語

（L2）冠詞習得研究の中でのその重要性を論じる．L2 での冠詞使用は，定性（definiteness）や特

定性（specificity）という文脈情報や，第一言語（L1）の影響の観点から広く議論されてきた．一

方で，インプットや名詞の語彙的情報などの要因はそれほど注目されてこなかった．ここでは，

まず先行研究を概観し，定性や特定性だけでは説明しきれないL2冠詞使用の現象を取り上げる．

そして，インプット頻度や手がかり（cue）のデータを提示し，インプットにおける名詞句の種

類と形式のパターンによって，それらの現象が説明できることを指摘する．次に，本論文では冠

詞に関連する素性（[±definite], [±count], [±plural]）が，それぞれの冠詞の選択にどの程度寄与する

のかを調査し，L2 学習者にとって定性より可算・不可算の区別の習得の方が困難であることを主

張する．その上で，L2 学習者にとって正しい冠詞，特に不定冠詞かゼロ冠詞かを選ぶのが難しい

とされる抽象名詞の習得に関する問題を見ていく．動詞由来の抽象名詞を取り上げ，境界

（boundedness）という語彙意味的性質もしくは名詞の派生タイプが可算性判断に影響を及ぼすの

かを検証する．実験の結果，その二つの要因は L2 学習者による可算性判断に関係がないことが

わかり，可能性として L1 からの影響が指摘された．最後に，習熟度の低い学習者は文脈情報に

頼ることなく，名詞の可算性を直感で判断するという主張について検証する．L2 学習者による文

脈なしの名詞単体に対する直感的可算性判断と，文脈を与えられた名詞に対する冠詞の使用を比

較した結果，学習者の多くが直感と冠詞の使用を結びつけているという傾向は見られなかった．

直感に頼る学習者も中にはいたが，どのような学習者が直感に頼る傾向にあるかということに，

L2 習熟度は関係がなかった．本論文の結論として，名詞特性は L1 からの影響を受けやすく，冠

詞の使用にも影響を及ぼすため，L2 冠詞習得の全体像を把握するためにも，名詞特性の影響は綿

密な実証研究が必要なことを強調したい．

1 Introduction It has commonly been reported that learners with article-less first languages (L1s) such as Japanese, Chinese, Korean, Polish, and Serbian have trouble using articles correctly. One of the most observed phenomena is the overuse of the definite article the in indefinite contexts where the indefinite article a/an or the zero article ø is required (Hawkins et al., 2006; Ionin, Ko, & Wexler, 2004; Snape, 2005; Tryzna, 2009). In particular, Ionin et al. (2004) explained the phenomenon under the Fluctuation Hypothesis (FH), identifying the context where the overuse of the was likely to occur. The hypothesis presupposed definiteness [±definite] and specificity [±specific] as universal semantic distinctions involved in the system of language in relation to the Article Choice Parameter (Ionin, Ko, & Wexler, 2003). In the informal definitions provided by Ionin et al. (2004, p. 5), [+definite] indicates that the speaker and hearer

Ogawa, M.

- 49 -

presuppose the existence of a unique individual in the set denoted by the noun phrase (NP), and [+specific] indicates that 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. The Article Choice Parameter requires either definiteness or specificity to be marked in a language with articles. The task for second language (L2) learners acquiring English, for instance, is to mark [±definite] with articles, instead of [±specific]. According to the FH, both definiteness and specificity are universal semantic notions and thus universally available to speakers of any language. In the course of L2 acquisition, therefore, learners tend to fluctuate between the possible settings (i.e., the definiteness or specificity setting), resulting in making errors in contexts where the values of the two features conflict, such as the [-definite, +specific] context. In fact, many studies documented overuse errors of the in such contexts. Interesting here is that fluctuation is often observed with learners with article-less L1s, but not with those who have articles in their L1s, such as Spanish, French, German and Greek (García Mayo, 2009; Hawkins et al., 2006; Snape, 2006). Such learners with article L1s establish the correct feature setting through the L1 and rarely show fluctuation errors. This suggests that L1 transfer overrides fluctuation.

Although the FH accounts well for the overuse of the definite article and positive L1 transfer in L2 article use, some studies draw attention to what is common for speakers of any L1 background. For example, learners are usually more accurate in using the definite article in definite contexts than in using the indefinite article in indefinite contexts (García Mayo, 2009; Hawkins et al., 2006; Ionin, Zubizarreta, & Maldonado, 2008). This directionality is presumably due to the fact that more features such as countability and number need to be specified for indefinite than definite nouns. Moreover, noun type differences are found even in definite contexts such that singular count nouns are more accurately supplied with the definite article than plural count or mass nouns are (French & Syrian Arabic for Sarko, 2009; Spanish for Snape, 2008). These observations cannot be explained well under hypotheses that only concern the effect of semantic contexts on L2 article use.

This paper will highlight some aspects of L2 article use which have not been discussed in depth in previous research of article acquisition: factors outside definiteness or specificity. If the L2 learner’s task is “to arrive at a linguistic system which accounts for the L2 input, allowing the learners to understand and speak the second language (White, 2003, p. 15)”, for example, it is reasonable to assume that learners are susceptible to the patterns they may encounter in their potential L2 input. The paper first introduces the studies which have focused on input frequency and cues relevant to articles. Then, it overviews the previous research reporting persistent difficulty with the count-mass distinction, which may be more problematic than definiteness. In its relation, articles with abstract nouns will closely be investigated particularly for the purpose to examine how the lexical-semantic feature of boundedness may influence countability judgment of abstract nouns. Acknowledging that countability judgment of nouns is crucial for correct article choice, a claim will finally be examined that L2 learners rely on intuition to decide whether a noun is countable or not.

Second Language 18

- 50 -

2 Input Frequency and Cues Previous research has reported the following tendencies: overuse of definite articles, directionality that the definite article is more correctly used than the indefinite article, and higher accuracy of the definite article with singular count nouns than with plural or mass nouns. If the directionality is assumed to be attributed to the compatibility of the definite article to occur with any type of noun, one would expect the effect to apply equally to any noun. However, L2 learners are unlikely to show the same degree of accuracy with singular count as with plural count or mass nouns in definite contexts, resulting in more erroneous uses of ø with plural count and mass nouns than those of a/an with singular count nouns (Sarko, 2009; Snape, 2008). This was even true for the learners with L1s which have articles (French, Syrian Arabic, Spanish). If L2 learners only misunderstood definite contexts as indefinite, then the suppliance of a/an to singular count nouns and that of ø to plural count and mass nouns would be expected to be the same. This noun type difference is thus not simply due to the misinterpretation of semantic contexts but involves other factors interacting.

Ogawa (2015) examined the effect of input as a potential factor and investigated the frequency distribution of different noun forms. L2 learners are possibly exposed to the following five types of NPs:

(1) a. the NP (definite singular count or definite mass) b. the NPs (definite plural count) c. a/an NP (indefinite singular count) d ø NPs (indefinite plural count) e. ø NP (indefinite mass).

If definite NPs (1a, 1b) are more frequent than indefinite NPs (1c, 1d, 1e) in the input, it may simply lead learners to prefer the to a/an or ø. Similarly, considering the noun type difference, input patterns may bring about the observed tendencies of L2 learners if singular count nouns occur more frequently with the (1a) than a/an (1c), plural nouns with ø (1d) than the (1b), and mass nouns with ø (1e) than the (1a).

A list of nouns created by Taler and Jarema (2007) which classified count and mass nouns based on concreteness and imageability was used for item selection. Twenty five count and 25 mass nouns with high frequency were selected, and most of them proved to be familiar to EFL learners in Japan, checked against a familiarity word list by Yokokawa (2006). The nouns were searched for input frequency in the possible grammatical forms of singular, plural, or mass. Although it was ideal to use L2 input corpora, no such corpora were available for Japanese learners at the time of data collection; therefore, the British National Corpus (BNC) was used to extract frequency counts. In effort to compensate for this weakness, an additional investigation was performed to compare the BNC data with potential L2 input for Japanese learners, using six English textbooks published in Japan. The comparison showed that frequency distributions of noun types and forms in the BNC data correlated closely with those

Ogawa, M.

- 51 -

found in the textbooks (r = .568, p < .05). In order to reflect input frequency particular to Japanese learners of English, furthermore, frequency counts taken from the BNC were converted for analysis to match the frequency ratio of noun types in Yokokawa’s word familiarity list1.

Table 1 shows the means of frequency counts. Looking at count nouns in different forms, definite plurals were least frequent (62.54) and definite singulars most frequent (282.42), and indefinite singulars and plurals did not significantly differ from each other (148.94 and 177.61). Mass nouns also showed a statistically significant difference between definite and indefinite forms (43.69 and 100.10), indicating that mass nouns appear more frequently in bare form (ø NP) than in definite form (the NP) in the input. The frequency orders are summarised in (2).

Table 1. Means of frequency counts (modified from Table 5 in Ogawa, 2015, p. 77)

Noun type Definite Indefinite the NP the NPs a/an NP ø NPs ø NP

Count 282.42 62.54 148.94 177.61 n/a Mass 43.69 n/a n/a n/a 100.10

(2) Count nouns: the NPs < a/an NP, ø NPs < the NP Mass nouns: the NP < ø NP As predicted, the input frequency patterns mirror the tendencies in L2 article use reported

in previous studies. Ogawa (2015) argues that given such frequency rates in the input, L2 learners may rely on the distributions of noun types and forms they are exposed to. This can consequently facilitate the use of the with singular count nouns and suppress it with plural and mass nouns, especially when contextual information is ambiguous or the semantics of articles is not fully acquired. L2 speakers learn to map meanings (i.e., features) to forms in the course of L2 acquisition, and they establish an interlanguage grammar with certain feature specifications both for contexts where articles are used and for articles themselves (Hawkins et al., 2006). For instance, a contextual feature of definiteness and lexical features of singular, plural and mass can be encoded in the definite article, while the same lexical information of nouns (singular, plural or mass) is also mapped to the indefinite or the zero article. When there is a competition between possible form-feature mappings, L2 learners need to determine which mapping dominates the others and may be influenced by input frequency under such decision making.

Similarly considering the effect of form-meaning mappings, Zhao and MacWhinney (2018) examined the acquisition of English articles under the Competition Model. The model views that language acquisition occurs through a competition between alternate form-function mappings (Bates & MacWhinney, 1989; MacWhinney, 2012). For example, functions compete

1 See Ogawa (2015) for detailed procedures.

Second Language 18

- 52 -

for mapping onto a form, such that the definite article the has usages like uniqueness or prior mention (i.e., functions) and the possible functions compete for the use of the. Likewise, forms compete for a function, given that all articles can express the generic as in the lion, a lion, and ø lions. There are a number of linguistic cues which signal mappings of functions to articles, and L2 learners are exposed to such cues in their input and have to establish correct form-function mappings for the use of articles. On this theoretical basis, Zhao and MacWhinney particularly investigated what kind of cues are available in the potential input for L2 learners and how frequent they are (i.e., cue availability), and how reliable they are to predict the use of specific articles (i.e., cue reliability). They identified 86 article cues from English grammar books and an ESL textbook, collected noun phrases from written English texts, and coded them according to the cues, as in (3). Then, the availability and reliability rates of the cues were calculated.

(3) Example sentence: The quality of a mother’s relationship with her toddler could affect that child’s

weight in adolescence. Cues:

a. The quality: singular countable with post-modifier → the b. a mother: singular countable → a/an c. relationship, weight: possessive → ø2 d. adolescence: noncountable → ø (Zhao & MacWhinney, 2018, p. 104)

Table 2. Article cues with high availability (extracted from Table 1 in Zhao & MacWhinney, 2018, p. 105)

Article Cue Example 1. plural → ø ø books 2. non-countable → ø ø water 3. singular countable with post-modifiers → the the man she is dating 4. singular countable → a/an a Shakespearean drama 5. plural with post-modifiers → the the letters I received today

As a result, they found that ten cues with the highest availabilities accounted for 76.3% of

all the coded tokens, and among them the top four cues, shown in Table 2, accounted for almost half of the total tokens. This indicates that frequent cues consist mostly of the occurrences of any given article usage. Moreover, the 86 article cues were all highly reliable with only a few exceptions. For instance, some cues scored the full value of reliability; that is, whenever the

2 Syntactically speaking, the NPs relationship and weight take a possessive in their determiner position; therefore, ø in this case refers to no article used, technically not equivalent to the zero article.

Ogawa, M.

- 53 -

cues apply, they do so correctly (e.g., plural → ø, non-countable → ø). On the other hand, two cues were found to have relatively low reliabilities, but highly available: singular countable with post-modifiers → the, plural with post-modifiers → the. What it means is that when singular count or plural nouns occur with post-modifiers, which is quite frequent, they do not always take the definite article; therefore, the presence of post-modifiers does not reliably signal the use of the. The Competition Model predicts that such unreliability should cause difficulty in language learning, while cues with high availability and reliability would be the ones that are learnt first. The results thus support that learners may strongly associate plural and mass nouns with the zero article ø.

In summary, the studies on the effect of input frequency and cues indicate that learners are exposed to various usages of articles with nouns and that the form-meaning mappings they encounter are so unbalanced that, for instance, the difficulty using the with plural or mass nouns cannot equally be compared to the difficulty (or relative ease) using the with singular count nouns. Although what requires the use of the definite article is indeed the contextual information free from noun types, it is a plausible argument that learners in the very process of L2 learning may draw on a variety of resources at hand including input frequency and cues until they come to fully understand the nature of definiteness. Nominal properties of the count-mass distinction and number marking can also be such resources for learners to rely on for article choice. 3 Persistent Difficulty with the Count-Mass Distinction over Definiteness The use of English articles involves three levels of decision making: first to decide whether the context is definite (the) or indefinite (a/an or ø), and when indefinite, next to decide whether the noun is uncountable (ø) or countable, further followed by number distinction for countable nouns, singular (a/an) or plural (ø). Many studies have documented that L2 learners have problems with the count-mass distinction in relation to articles (Hiki, 1990; Snape, 2008; Yoon, 1993). In L1 English, the nominal feature of [±count] is suggested to be encoded in lexical entries of nouns in the mental lexicon (Taler & Jarema, 2007). In contrast, the count-mass distinction is realised at the level of classifiers in L1 Japanese, and nouns are claimed to be stored with the feature [±individuated] in the lexicon (Iwasaki, 2002; Kobuchi-Philip, 2007; Ogawa, 2014). Given that nouns in the lexicon of L1 English and L1 Japanese are assigned with different features, Japanese learners of English need to restructure the L2 mental lexicon to encode [±count] as lexical information of nouns.

Assuming that L2 acquisition involves setting the configuration of feature specifications to that of a target language (Lardiere, 2009), Ogawa (2014) examined the extent to which features involved in article choice (e.g., [±definite], [±count], [±plural]) contribute to L2 learner’s use of each article. The study compared different proficiency groups and showed that the contribution of [+definite] to the choice of the increased largely as learners became proficient in the L2. It indicates that for the use of the definite article they become able to

Second Language 18

- 54 -

correctly rely on context irrespective of noun type. As for a/an and ø, however, [-definite] did not have as strong effect as [+definite] did for the choice of the, indicating that learners, especially at a low proficiency level, draw on the nominal property of whether a noun is pluralised or not for the use of a/an and ø, rather than the fact that the noun is indefinite. Based on the results, Ogawa has assumed that L2 learners associate each article with a certain feature or features and such associations with features vary in strength: a feature with smaller strength may easily be affected by stronger ones. For instance, low proficiency learners associate the with [+definite] as a dominant feature, but it is not as strong as [-plural] with a/an or [+plural] with ø. In this case, the is likely to be interfered with by a/an when nouns are definite singular, or with ø when definite plural. Based on the feature strength, Ogawa claims that the feature [+definite] gets more valued and starts to compete with the other choices like [±plural] as proficiency improves, and this is probably how the use of a/an and ø retreat from definite contexts over time.

For the use of ø, moreover, learners need to recognise that a noun is mass or plural as well as indefinite. Ogawa (2014) has noted that learners gain more reliance on indefiniteness of context for the choice of ø along with proficiency and this development was remarkably larger than the improvement on recognising nouns as being mass. It implies that learners are capable of identifying [-definite] of the context, but trouble remains with judging [+count] or [-count] of the noun. This persistent difficulty with the count-mass distinction over definiteness is assumed to be related to the fact that the two properties lie at different levels of semantic processing. The feature [±count] is at an intra-linguistic or lexical level where meaning is first mediated through L1 counterparts (Jiang, 2000), whereas the feature [±definite] is at an extra-linguistic or discourse level requiring incorporation of contextual information. Ogawa assumes that the former type of features is more susceptible to L1 influence, resulting in more persistent problems.

The claim that the count-mass distinction is more difficult than definiteness is also validated in teaching research. Akamatsu (2018) investigated the effect of explicit instruction on English articles, focusing on teaching noun countability and definiteness separately over four weeks. Learners took a 70-minute session every week, which included the explanation of noun countability and definiteness, a practice test, and self-correction and relearning of items in the practice test. The results showed a discrepancy in learning effects between distinct aspects of English article usage, particularly indicating that the improvement on definite article use was larger than that on the article use involving countability judgment. For instance, the accuracy rates of definite article usages were relatively high and retained at over 90% for three weeks after instruction. By contrast, the accuracy rates of noun countability after instruction were 75.6% for abstract nouns and 86.6% for material nouns. The difficulty was more serious for the countable usage of abstract or material nouns, whose accuracy was only 56% after instruction. It is thus evident that even with explicit instruction, making correct judgments on noun countability is a very puzzling task for L2 learners.

Ogawa, M.

- 55 -

4 Articles with Abstract Nouns The previous section has illustrated that article use involving countability judgment of nouns causes a prolonged difficulty which is not fully overcome by explicit instruction. In relation to such difficulty with the count-mass distinction, some studies pointed to a specific noun classification type that brings further problems to L2 learners, reporting that they make more errors in article use with abstract nouns than with concrete nouns and this seems to persist at an advanced level (Hiki, 1990; Hua & Lee, 2005; Ogawa, 2008). Differences between count and noncount nouns can be characterised by the notion of boundedness: whether the referent has a clear perceptual outline or not (Jackendoff, 1991; Langacker, 2008). Since the referent of abstract nouns does not exist in the form that can be physically perceived, their countability judgment may be more difficult than concrete nouns. Another account suggested was that L2 learners may consider countability as a fixed or static property of nouns which does not change in any context (Butler, 2002). As reference grammar books describe, abstract nouns are primarily noncount, but they can be count when denoting specific instances or types of abstract concepts (Ando, 2005; Yasui, 1996). For instance, the word education is used as both count and noncount in the following passage:

(4) There was a poor boy who had no access to education. One day, a wealthy man appeared and provided him with a good education. The boy completed primary education of good quality and received two types of education, private and public, throughout his life. In the end, he became very rich. He thought getting an education was very important and decided to sponsor educations for needy people.

However, such extension of abstract nouns is irregular and unpredictable, and it is probably where the learning difficulty lies (Huddleston & Pullum, 2002).

Taking the notion of boundedness into consideration, Brinton (1998) expanded the analogy between the countability of nouns and temporal boundedness of verbs. For example, verbs like arrive and die (i.e., achievement verbs) are considered temporally bounded as they are instantaneous and punctual, having an immediate endpoint of the event they denote as part of their lexical meaning. In this sense, they are similar to count nouns, which include a certain boundary as part of their lexical semantics that serves as a counting unit. In contrast, verbs such as live, know, and hate (i.e., state verbs) do not have any specific endpoint in their semantics, and verbs like drive, swim, and dance (i.e., activity verbs) are also considered unbounded for their atelic and durative properties. Such lexical aspectual features of verbs can be linked to the property of mass nouns having no determinable boundary. Lexical aspects of verbs are predicted to be inherited when nouns are derived from those verbs and have an effect on how they are perceived in terms of boundedness. In fact, Barner, Wagner, and Snedeker (2008) presented empirical evidence that English native speakers associate the aspectual properties of verbs with countability, implying that the process of derivation from a verb to a noun does not change the fundamental semantic property of boundedness.

Second Language 18

- 56 -

Lee Amuzie and Spinner (2012) examined whether the same association of temporal boundedness and countability is observed in the domain of L2. They looked at 50 Korean speaking learners of English at an intermediate level and tested four different types of abstract nouns: state (e.g., love, knowledge), continuous actions (e.g., discussion, announcement), non-continuous actions which are equivalent to the achievement type (e.g., kick, catch, cut), and bounded independent nouns which are not derived from verbs (e.g., story, article. job). Each category consisted of 12 tokens. They administered a forced choice elicitation task, where a short passage or dialogue was given with a blank followed by a target noun, and the participants chose either indefinite or zero article. Contexts were controlled to induce indefinite articles for continuous and non-continuous action nouns, and bounded independent nouns. However, they made zero articles as target for state nouns because it was rather difficult to set the contexts which would require count use of state nouns. They predicted that non-continuous action nouns were lexically bounded and therefore more likely to be judged as count than lexically unbounded state or continuous action nouns. As a result shown in Table 3, their prediction was not fully supported in that non-continuous action nouns were less likely to be correctly recognised as count when compared with continuous action nouns.

Table 3. Noun types and mean accuracies of article choice in Lee Amuzie and Spinner (2012)

Noun Type Boundedness Target Answer Mean Accuracy State unbounded ø 87.83% Continuous action unbounded a/an 70.83% Non-cont. action bounded a/an 54.18% Bounded Independent bounded a/an 88.5% Lee Amuzie and Spinner (2012) speculated possible reasons for the unexpected findings.

First, derivational morphology might have contributed to higher accuracy for continuous action nouns. Many of the non-continuous nouns were derived through a verb-to-noun conversion, whereas all of the continuous ones were derived through suffixation (e.g., -ment, -tion). The learners may have hesitated to assign two different grammatical classes (verb, noun) to a single form, avoiding using a/an with converted nouns. In contrast, nouns with nominal suffixation may have been easier to be categorised as noun and was supplied with an article with less hesitation. Another reason is L1 influence. Many of the continuous action nouns used in the study can take the optional plural marker -tul in Korean and may have been considered as countable by Korean speakers, resulting in relatively high suppliance of a/an. On the other hand, the translation equivalents of most of the non-continuous action nouns are usually expressed in the form of gerund and cannot take the plural marker in Korean. For this reason, it is likely that the learners considered most of the non-continuous action nouns as uncountable and were reluctant to use a/an with them.

Ogawa, M.

- 57 -

4.1 Experiment The current article addresses the remaining issues of Lee Amuzie and Spinner (2012) by examining L2 learners with a different L1, Japanese. The main research question is how L2 learners determine the countability of English abstract nouns; more specifically, whether or not the choice of articles, either indefinite or zero, is affected by (a) lexical aspectual properties, like state, action, or achievement, and/or (b) derivation type such as conversion or suffixation. The experiment also investigates how different proficiency levels may influence L2 countability judgment.

4.2 Participants Japanese learners of English, who were university students, participated in the experiment. They were divided into two proficiency levels based on TOEIC scores and the Oxford Quick Placement test (UCLES, 2001): lower-intermediate (LI; N = 21), and upper-intermediate/advanced (UI/A; N = 25). 4.3 Instrument Nouns used in the test are in Table 4. Each semantic type (state, action, achievement) included eight items, which were further divided by derivation type (conversion, suffixation). The suffix –tion was used, and all the items were selected from the word familiarity list for Japanese learners of English (Yokokawa, 2006, 2009). Sound and letter familiarities were controlled; therefore, there was no difference between the three semantic types (Sound: F (2, 21) = 0.102, p = .903; Letter: F (2, 21) = 0.022, p = .978). However, conversions were more familiar than suffixations (Sound: F (1, 22) = 10.007, p = .005; Letter: F (1, 22) = 34.586, p < .001). It is due to the fact that conversions can be both noun and verb as they are, and it contributes to higher familiarity rates than those of suffixation.

Table 4. Nouns used in the test by derivation and semantic types

Derivation Type State Action Achievement

Conversion

fear drive promise hope fight start

respect search stop worry move win

Derivation

confusion calculation registration imagination competition application

intention discussion decision satisfaction observation recognition

A forced-choice elicitation task was used in which context introduction was given in

Japanese, and an English passage followed with a blank to fill with either a/an or ø, as in (5). In this test, all the contexts were set to activate the count interpretation of nouns; therefore, target answers were all indefinite nouns. The test included 37 questions in total: 24 targets and 13 fillers.

Second Language 18

- 58 -

(5) とても綺麗な島を観光で訪れて言いました。

(When visiting a very beautiful island for sightseeing, someone said…) Everything looks so amazing here! Why don’t we take ( a / ø ) drive around the island?

4.4 Results The mean accuracies of the overall target items (max. = 24) were 50% for LI (mean = 12.0, SD = 3.91), and 64% for UI/A (mean = 15.3, SD = 3.95). This result itself proves the difficulty judging correct countability of abstract nouns.

Table 5 and Figure 1 show the mean scores of each category by the two proficiency groups. Repeated measures ANOVA revealed the main effects of semantic type (F (2, 88) = 12.18, p < .001, partial η2 = .217) and group (F (1, 44) = 8.12, p = .007, partial η2 = .156) and interactions between semantic type, derivation type, and group (F (2, 88) = 4.06, p = .021, partial η2 =.084).

Table 5. Mean scores of the target items by proficiency groups

Derivation Type

Semantic Type

Lower-Inter. (LI) Upper-Inter./ Adv. (UI/A)

Mean SD Mean SD

Conversion State 1.4 1.2 2.3 1.0

Action 2.1 1.2 3.0 1.2 Achievement 1.9 1.0 2.3 1.1

Suffixation State 1.6 1.2 2.1 1.4

Action 2.7 1.1 2.6 1.0 Achievement 2.3 0.9 3.1 1.0

Figure 1. Mean scores of the target items by proficiency groups

0.0

1.0

2.0

3.0

4.0

Conversion Suffixation

Lower-Inter. (LI)

State Action Achieve.

0.0

1.0

2.0

3.0

4.0

Conversion Suffixation

Upper-Inter./Adv. (UI/A)

State Action Achieve.

Ogawa, M.

- 59 -

Multiple comparisons revealed that in the LI group, state nouns with suffixation were significantly lower than action with suffixation (p = .003) and achievement with suffixation (p = .040), and the learners did not differentiate the three semantic types of converted nouns. In addition, suffixed action nouns were marginally better than converted action nouns (p = .054). For the UI/A group, action nouns with conversion were significantly higher than state with conversion (p = .043) and achievement with conversion (p = .032), and achievement nouns with suffixation were significantly better than state with suffixation (p = .001). When compared by derivation type, suffixed achievement nouns were significantly higher than converted achievement nouns (p = .001). 4.5 Discussion The results of Lee Amuzie and Spinner (2012) showed that continuous action nouns were judged as count more correctly than non-continuous action nouns. They speculated that relative easiness for a noun to be supplied with a/an may be related to its derivation type; that is, suffixed nouns are easy to be judged as noun, resulting in less hesitation of using a/an with them than with converted nouns. The present study did not find a strong effect of derivation type for countability judgment of abstract nouns. Therefore, the fact that suffixation -tion specifies the category of a word as noun does not seem to be a contributing factor for the choice between a/an and ø. Although the effect of derivation type was not statistically significant, the scores of the LI group may descriptively suggest a tendency in favor of the effect that is limited to low proficiency learners. In all the semantic types, for instance, the LI group scored higher on suffixation than on conversion: state (conversion 1.4, suffixation 1.6), action (conversion 2.1, suffixation 2.7) and achievement (conversion 1.9, suffixation 2.3). Suffixed nouns may thus be easier to be supplied with a/an especially by learners with low proficiency. While there may be such a tendency, it is not so strong as to interfere the article choice by high proficiency learners and does not seem to affect their judgment on countability of nouns in context.

The account based on boundedness predicted that bounded nouns (i.e., achievement) would be easier to be recognised as count than unbounded nouns (i.e., state, action). As the results of this experiment showed, the prediction was not supported. The phenomenon common to the both proficiency groups was that suffixed achievement nouns were judged as count more correctly than suffixed state nouns. Although it seems to partly corroborate the effect of temporal boundedness, the same was not observed with converted nouns at all. The LI group did not differentiate the three semantic types of converted nouns, and the UI/A group judged action nouns as count more correctly than state and achievement nouns. The reason why the predicted performance was not observed may be attributed either to the irrelevance of temporal boundedness in countability judgment or to L2 learners’ incomplete awareness to boundedness. The clarification requires a following investigation into whether or not L2 learners understand and distinguish lexical aspects of verbs per se.

Finally, looking at the descriptive differences between state, action, and achievement nouns, the learners in the LI group seemingly had more difficulty judging state nouns as count than action and achievement nouns. It is indicative that they may rather distinguish action type

Second Language 18

- 60 -

(i.e., action, achievement) from non-action type of nouns (i.e., state) and have difficulty perceiving non-action as count. As they develop in proficiency, their countability judgments appear to show more variability, which may presumably stem from some L1 interpretations of the nouns tested, irrespective of semantic boundedness and derivation type. Jiang (2000) claims that the meaning of an L2 word is mediated through the L1 counterpart and that a majority of L2 words fossilise with L1 lexical information, which makes it difficult to achieve successful establishment of L2 lexical information in the lexical entry. Therefore, the countability of L2 nouns may possibly be determined in reference to the interpretation of the L1 translation counterparts. Further investigation is required to verify the presence of such L1 influence. Although the variability found in the UI/A group cannot be explained well by the data at hand, it is at least demonstrated that boundedness based on lexical aspects is not likely to influence L2 countability, providing no evidence that achievement nouns were easier to be perceived as count than action nouns. Potentially to be examined in further research is that temporal boundedness may only serve as a determinant for learners with near-native proficiency or that there may be L1 influence from the interpretations of the equivalent nouns in Japanese.

There are some limitations in the current study which can be addressed in future research. First, the effect of discourse or linguistic context was not strictly controlled for regarding telicity. For example, the telicity of a verb drive can alter depending on the context in which the verb is used. Drive is primarily classified as action with having no endpoint of the event as part of its lexical meaning; therefore, it is considered as atelic (i.e., unbounded) in the current study. However, when it is embedded in the sentence like I drove to my friend’s house, the event of driving can certainly have an endpoint when the car arrives at the friend’s house, resulting in changing the telicity of the event to telic (i.e., bounded). Although the contexts used in the test were checked by native speakers to require the count use of the target items, additional manipulation might have been necessary not to alter the telicity encoded in their lexical semantics. Another is the absence of a control group. A group of native speakers could be included in the experiment to confirm its validity. Lastly, further investigation should be conducted by comparing both count and noncount uses of abstract nouns. It will surely offer deeper insight into L2 learners’ countability judgments of abstract nouns. 5 Intuition and Countability Judgment English article use involves countability judgment of nouns, and this paper has so far discussed the effects of nominal properties in L2 article choice and the persistent difficulty with the count-mass distinction. It has also been in question how L2 learners determine whether a noun is count or noncount. Butler (2002) assumes that noun countability may be a fixed property of a noun in L2 interlanguage grammar, noting that a number of learners deal with countability as if they had static lists of countable and uncountable nouns and could retrieve a word from either list regardless of context. Regrettably, however, there are not many studies which have directly

Ogawa, M.

- 61 -

examined the relationship between L2 countability judgment in context and learners’ perception of nouns.

Yoon (1993) is one such study which addressed what constituted the perception of noun countability by both L1 and L2 speakers of English and investigated how it would affect the use of indefinite and zero articles in particular contexts. She administered two tasks, an intuitive countability judgment (ICJ) task and an article cloze test, with English native speakers and intermediate and advanced learners of English whose L1 was Japanese. The two tasks were created from two essays written by English native speakers. For the ICJ task, Yoon (1993) made a list of nouns which were selected from the essays, and participants were asked to intuitively judge whether each noun in the list was countable or uncountable. The article cloze test was created by omitting articles in the two essays, and participants needed to supply articles. The results of the ICJ task showed that native speakers perceived 82% of the nouns as count, while L2 learners perceived 73% as count. She thus concluded that nouns are primarily countable rather than uncountable. However, there was a clear methodological problem. The 87 nouns selected for the ICJ task were not controlled for in terms of word familiarity to the learners or concreteness, so it is a premature conclusion and it may even be debatable if this result alone tells us something meaningful. More interesting in Yoon’s study is, however, its comparison to the result of the article test. Looking at the instances of indefinite articles, the study showed a correlation between L2 learners’ ICJ and article choices. However, native speakers did not demonstrate such a correlation. In other words, given a particular context, native speakers were quite accurate in choosing a/an even to the nouns they intuitively perceived as uncountable. On the other hand, L2 learners were less sensitive to contextual cues than natives, relying on their intuition to determine the countability of nouns.

A problem involved in Yoon (1993) is that she aggregated all the participants’ scores of the ICJ task, computed mean countability rates of each noun, and then compared them with mean suppliance rates of indefinite articles to the same nouns in the article test. In this way, it was not clear whether the correlation found was truly at the individual learner level to verify that some learners have a strong link between ICJ and article choice in context. Alternatively, there might be a possibility that the correlation reported in Yoon was due to some sort of qualitative features of the nouns tested irrespective of individual learner tendency. 5.1 Experiment This experiment examines whether a correlation between learner intuition of noun countability and the choice of articles in context is at the individual learner level or at the level of lexical semantics of nouns, specifically the semantic differences such as state, action, or achievement. 5.2 Participants The same participants as in 4.2 took part in the experiment: lower-intermediate (LI; N = 21), and upper-intermediate/advanced (UI/A; N = 25). An ICJ task was given after the article task in 4.3 in order to avoid priming certain countability before the article choice in context.

Second Language 18

- 62 -

5.3 Instrument Sixty nouns were selected based on Brinton (1998) and Ikehara et al. (1997): 30 concrete and 30 abstract nouns. The abstract nouns included the 24 nouns in Table 4 which were used in the article choice task, and they were the target nouns to be analysed in this section in relation to article choice. The nouns were presented in a list without context, and participants were told to judge whether they are countable or uncountable. When they were not sure about the meaning of nouns, they could choose an option of not sure. Some participants with knowledge that nouns can be both count and noncount depending on context were specifically advised to judge based on their first intuition of which a noun is more likely to be. 5.4 Results First, the responses to the 24 abstract nouns in the ICJ task were analysed to see whether L2 intuition was related to nominal properties such as derivation type (i.e., conversion, suffixation) or semantic type (state, action, achievement). The results are in Table 6. A chi-square test revealed that derivation type affected learners’ intuitive countability in all the semantic types (state χ2(1) = 6.018, p = .014, Cramer’s V = .128; action χ2(1) = 5.918, p = .015, Cramer’s V = .127; achievement χ2(1) = 10.516, p = .001, Cramer’s V = 169), but the effect of semantic type was only observed with suffixed nouns (conversion χ2(2) = .761, p = .683; suffixation χ2(2) = 45.193, p < .001, Cramer’s V = .286). Although the effects of derivation and semantic types appeared to be statistically significant, their effect sizes were rather small. A residual analysis then specified the following tendencies: (a) suffixed state nouns were more likely to be judged as noncount than converted state nouns, and (b) converted action and achievement nouns were judged as noncount more often than suffixed action and achievement nouns respectively.

Table 6. Percentages of intuitive countability by derivation and semantic types (raw count)

Derivation Type Semantic Type Intuitive Countability Count Noncount

Conversion State 33.2% (61) 66.8% (123)

Action 35.3% (65) 64.7% (119) Achievement 37.5% (69) 62.5% (115)

Suffixation State 21.7% (40) 78.3% (144)

Action 47.8% (88) 52.2% (96) Achievement 54.3% (100) 45.7% (84)

Next, the result of the ICJ task was compared to that of the article task in the previous

section in order to examine how many participants chose articles based on countability intuition. The number of matching answers were counted regardless of accuracy in article use; that is, the cases where a/an was chosen for the nouns intuitively judged as count and the cases where ø was given to the nouns intuitively perceived as noncount were counted per participant. There

Ogawa, M.

- 63 -

were 24 nouns used in the tasks, which is the maximum number of matching answers a participant may make. Figure 2 shows the number of participants who made matching answers. The mean number of matching answers was 14.39 (SD = 3.296).

Figure 2. Numbers of matching answers and participants

Three learners demonstrated a very strong link between intuitive countability and article

choice, showing 20, 22, and 24 matching answers out of 24 items. Statistically speaking, the probability of making more than 20 matching answers out of 24 is less than 1%. There were 14 participants who made 16 to 18 matching answers, and they can also be considered having a moderate link between intuition and article choice. Those who matched the answers 9 to 15 times out of 24 were actually at the chance level. The result thus showed that of 46 participants, 17 were inclined to choose articles that would match the countability of nouns that they intuitively perceived.

Lastly, the responses of the 17 participants with a link between intuition and article choice were scrutinised. Figure 3 summarises the number of matching answers of each participant in detail so as to verify whether the answers were count matching with a/an or noncount matching with ø. The former was the correct answers in the article test. Participants were considered to have a specific preference toward either count or noncount when they made either type of responses more than 80% of their matching answers. Participants 1, 2, and 3 were rather inclined to perceive abstract nouns as count and choose indefinite articles for them. In contrast, Participants 5, 6, 7, 8, and 9 preferred noncount by intuition and selected zero articles for the nouns in context. The rest of the participants displayed no such preference. As for the proficiency levels of the 17 participants, five participants were lower-intermediate (LI) and 12 upper-intermediate/advanced (UI/A). Those with count preference were all UI/A learners, whereas those with noncount preference consisted of three LI and two UI/A learners.

0

1

2

3

4

5

6

7

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

Num

ber

of p

artip

ants

Number of matching answers (Count = a/an, Noncount = ø)

Second Language 18

- 64 -

Figure 3. Participants who made matching answers and their proportion of matching types

5.5 Discussion The results of the ICJ task suggest that the effect of derivation type is not unidirectional as such, showing that neither conversion nor suffixation brings about a certain interpretation on countability. Also, the effect of semantic type is not so strong as to influence across any derivation type. L2 intuition on noun countability is thus assumed to be unaffected by nominal properties such as derivation and semantic types categorically. To be more specific, the results can be interpreted in terms of individual noun differences, such that the nouns confusion, imagination, intention, and satisfaction tend to be perceived as noncount more often than fear, hope, respect, and worry. Similarly, a set of nouns drive, fight, search, move, promise, start, stop, and win may have happened to be intuitively judged as noncount than another set of nouns calculation, competition, discussion, observation, registration, application, decision, and recognition. Although there are plenty of possible reasons for why different nouns are perceived differently in terms of countability, L1 effects might be the most plausible factor since lexical meaning of an L2 word is mediated through the L1 to some extent. Further research into countability intuition of the equivalent nouns in Japanese may shed light on the reason(s) for the findings of the ICJ task.

0 2 4 6 8 10 12 14 16 18 20 22 24

123456789

1011121314151617

Number of matching answers

Part

icip

ants

Count = a/anNoncount = ø

Ogawa, M.

- 65 -

The comparison of the responses in the ICJ task and the choices of indefinite or zero article in the article test revealed that 36% of the total number of participants showed a certain correlation between their countability intuition and article use. Butler (2002) claimed that learners especially with low proficiency may regard noun countability as a fixed notion; however, the results of the current study found that only five of 21 LI learners made a significant level of matching answers. In other words, the other 16 participants were, whether successfully or not, able to alter their countability intuition in the article test. Since the most participants who made the matching answers in both tasks belonged to the UI/A level, it may not probably be the proficiency level that leads learners to the belief that noun countability is a fixed property of nouns which does not change in any context.

Furthermore, almost half of the participants with a link between countability intuition and article choice, eight of 17 participants, showed a preference toward either count or noncount. In particular, the preference toward count was only evident with UI/A learners. Considering the effect of proficiency, this may suggest that only high proficiency learners have a flexible perception to see abstract nouns as countable. In fact, Ogawa (2017) found a correlation between concreteness and countability of nouns; that is, nouns judged as concrete are more likely to be perceived as countable, while nouns judged as abstract tend to be perceived as uncountable. This correlation is suggested to change over time as the proficiency improves, consequently showing more dispersion between concreteness and countability. In other words, proficient learners may judge nouns as abstract but perceive them as count. In the current tasks, the target nouns analysed were abstract nouns derived from verbs, and they are by nature inclined to be perceived as noncount. Higher proficiency in L2 may thus have contributed to the learners’ perception of abstract nouns as count. However, this is not definitive due to a small number of samples, and future research is necessary. 6 Conclusion The L2 acquisition of English articles has long been investigated from various perspectives, and the focus has dominantly been on the effect of semantic contexts such as definiteness and specificity. This paper introduced different approaches to L2 article acquisition to demonstrate that other factors related to nominal properties can interact in article use and require careful empirical scrutiny.

Section 2 attempted to explain why patterns of article use appeared differently with different types of noun: namely, the definite article is more correctly used with singular count than with plural count or mass nouns. The studies on input frequency and cues demonstrated that the frequency distribution of noun types and forms in the input is so unbalanced that it could possibly facilitate the use of the with singular count nouns and suppress it with plural or mass nouns. Showing that degrees of difficulty (or relative ease) using the definite article vary with different noun types, it is also suggested that the research needs to take into account not

Second Language 18

- 66 -

only the effect of semantic contexts but also a variety of resources that L2 learners may rely on, including input and noun type differences.

Looking closely into the scale of contribution of relevant features to article use in particular, section 3 argued that the count-mass distinction posed a more persistent problem than definiteness. It was assumed that L2 learners established the association between each article and a certain feature or features (e.g., the with [+definite], a/an with [-plural/-definite]) and that the strength of such associations differed and changed as L2 proficiency improved. This can justify the tendency that misused a/an or ø retreat from definite contexts over time. However, the contribution of [-definite] and [-count] for the use of the zero article ø showed different degrees of improvement, suggesting that the difficulty judging whether a noun is countable or not is more serious and prolonged than the difficulty judging whether a noun is indefinite or not. Moreover, the learning problem with the count-mass distinction does not seem to be fully overcome by explicit grammar instruction.

Section 4 focused on L2 countability judgment of abstract nouns. They have been reported as one of the most difficult noun types with which learners have trouble using articles correctly. Based on the account that boundedness is the notion to characterise the count-mass distinction, a study investigated whether or not the temporal boundedness encoded in nouns derived from verbs (state, action, achievement) influenced the countability judgment by L2 learners. The results showed no such tendency that L2 learners drew on the boundedness based on lexical aspects in the countability judgment of abstract nouns. In addition, the effect of derivation type (conversion, suffixation) was examined, and neither was it significant enough to determine the choice between a/an and ø. A possibility of L1 influence was suggested that the countability of L2 nouns may possibly be determined in reference to the interpretation of the L1 translation counterparts.

Lastly in section 5, L2 learners’ countability intuitions were examined to verify a claim that they would intuitively decide the countability of nouns without relying on contextual information. The comparison of intuitive countability judgment of decontextualised nouns with the use of either indefinite or zero article in context revealed that a majority of learners demonstrated no such correlation between countability intuition and article use. The proficiency level was not also the discriminative indicator to identify those who have a strong connection between intuition and article choice. The question of where countability intuition comes from needs to be addressed in future research, but at least the findings from the study indicate that neither derivation nor semantics related to temporal boundedness of abstract nouns is the influential factor.

The present paper has discussed the explanatory power of other factors rather than semantic contexts for L2 article use, especially focusing on the influence of the count-mass distinction. The accurate use of articles involves not only mastering the distinction between definite and indefinite but also correctly judging the countability of nouns, and the latter is maintained to be more difficult than the former. The reason is suggested that semantic processing of the count-mass distinction and definiteness lies in different levels. The count-mass distinction is more susceptible to L1 interference because the lexical interpretation of L2

Ogawa, M.

- 67 -

nouns is likely to be mediated through the L1 counterparts. In the case of L1 Japanese and L2 English, furthermore, both languages have the count-mass distinction, but it is realised at different levels, whereas definiteness is not grammaticalised in Japanese. Therefore, restructuring of properties that are similar in the L1 and L2 is arguably more difficult than acquiring a grammar which does not have a transferrable property in the L1. No research, to my knowledge, has in fact tackled the question whether a native-like level of countability judgment can even be attainable or not. Future research may well be directed to verify it. Acknowledgements I would like to thank the participants for their contribution to this research as well as the audience at the J-SLA Autumn Seminar in 2018 for comments and discussion. I also thank two anonymous reviewers for helpful comments that enabled me to clarify a number of issues and improve the article. This research was supported by JSPS KAKENHI Grant Number 15K16800. References Akamatsu, N. (2018). Does cognitive linguistic insights help Japanese learners understand the

English article system? SELT (Studies in English Language Teaching), 41, 1-20. Ando, S. (2005). Lectures on modern English grammar. Tokyo: Kaitakusha. Barner, D., Wagner, L., & Snedeker, J. (2008). Events and the ontology of individuals: Verbs

as a source of individuating mass and count nouns. Cognition, 106, 805-832. doi: 10.1016/j.cognition.2007.05.001

Bates, E., & MacWhinney, B. (1989). Functionalism and the competition model. In B. MacWhinney & E. Bates (Eds.), The crosslinguistic study of sentence processing (Vol. 3, pp. 73-112). Cambridge University Press.

Brinton, L. J. (1998). Aspectuality and countability: A cross-categorial analogy. English Language and Linguistics, 2, 37-63. doi: doi:10.1017/S136067430000068X

Butler, Y. G. (2002). Second language learners' theories on the use of English articles: An analysis of the metalinguistic knowledge used by Japanese students in acquiring the English article system. Studies in Second Language Acquisition, 24, 451-480. doi: 10.1017/S0272263102003042

García Mayo, M. del P. (2009). Article choice in L2 English by Spanish speakers. In M. del P. Garcia-Mayo & R. Hawkins (Eds.), Second language acquisition of articles: Empirical findings and theoretical implications (Vol. 49, pp. 13-35). Amsterdam: John Benjamins Publishing Company.

Hawkins, R., Al-Eid, S., Almahboob, I., Athanasopoulos, P., Chaengchenkit, R., Hu, J., Rezai, M., Jaensch, C., Jeon, Y., Jiang, A., Leung, I., Matsunaga, K., Ortega, M., Sarko, G., Snape,

Second Language 18

- 68 -

N. and Velasco-Zarate, K. (2006). Accounting for English article interpretation by L2 speakers. EUROSLA Yearbook, 6, 7-25. doi: 10.1075/eurosla.6.04haw

Hiki, M. (1990). The judgment of noun countability by Japanese college students: Where is the difficulty? JACET Bulletin, 21, 19-38.

Hua, D., & Lee, H. (2005). Chinese ESL learners’ understanding of the English count-mass distinction. Proceedings of the 7th Generative Approaches to Second Language Acquisition Conference (GASLA 2004), 138-149.

Huddleston, R., & Pullum, G.K. (2002). The Cambridge grammar of the English language. Cambridge University Press.

Ikehara, S., Miyazaki, M., Shirai, S., Yokoo, A., Nakaiwa, H., Ogura, K., Ooyama, Y. and Hayashi, Y. (1997). Goi-taikei: A Japanese lexicon (Vol. 1). Tokyo: Iwanami Shoten.

Ionin, T., Ko, H., & Wexler, K. (2003). Specificity as a grammatical notion: Evidence from L2-English article use. WCCFL 22 Proceedings, 245-258.

Ionin, T., Ko, H., & Wexler, K. (2004). Article semantics in L2 acquisition: The role of specificity. Language Acquisition, 12, 3-69. doi: 10.1207/s15327817la1201_2

Ionin, T., Zubizarreta, M. L., & Maldonado, S. B. (2008). Sources of linguistic knowledge in the second language acquisition of English articles. Lingua, 118, 554-576. doi: 10.1016/j.lingua.2006.11.012

Iwasaki, S. (2002). Japanese. Amsterdam: John Benjamins. Jackendoff, R. (1991). Parts and boundaries. Cognition, 41, 9-45. doi:

https://doi.org/10.1016/0010-0277(91)90031-X Jiang, N. (2000). Lexical representation and development in a second language. Applied

Linguistics, 21, 47-77. doi: 10.1093/applin/21.1.47 Kobuchi-Philip, M. (2007). Individual-denoting classifiers. Natural Language Semantics, 15,

95-130. doi: 10.1007/s11050-007-9011-1 Langacker, R. W. (2008). Cognitive Grammar: A basic introduction. Oxford University Press. Lardiere, D. (2009). Some thoughts on the contrastive analysis of features in second language

acquisition. Second Language Research, 25, 173-227. Lee Amuzie, G., & Spinner, P. (2012). Korean EFL learners’ indefinite article use with four

types of abstract nouns. Applied Linguistics, 34, 415-434. doi: 10.1093/applin/ams065 MacWhinney, B. (2012). The logic of the unified model. In S. Gass & A. Mackey (Eds.), The

Routledge handbook of second language acquisition (pp. 211-227). New York: Routledge/Taylor & Francis.

Ogawa, M. (2008). The acquisition of English articles by advanced EFL Japanese learners: Analysis based on noun types. Language and Information, 3, 133-151.

Ogawa, M. (2014). The role of the mass-count distinction in the acquisition of English articles by speakers of an article-less first language. Doctoral dissertation, University of Essex.

Ogawa, M. (2015). Input frequency of article-noun combinations as another factor for L2 article use. Ars Linguistica: Linguistic Studies of Shizuoka, 21, 67-88.

Ogawa, M.

- 69 -

Ogawa, M. (2017, August). Perceived countability of concrete and abstract nouns in L2 English. Paper presented at the 43th Annual Conference of the Japan Society of English Language Education (JASELE), Shimane, Japan.

Sarko, G. (2009). L2 English article production by Arabic and French speakers. In M. del P. García Mayo & R. Hawkins (Eds.), Second language acquisition of articles: Empirical findings and theoretical implications (Vol. 49, pp. 37-66). Amsterdam: John Benjamins Publishing Company.

Snape, N. (2005). The use of articles in L2 English by Japanese and Spanish learners. Essex Graduate Student Papers in Language and Linguistics, 7, 1-23.

Snape, N. (2006). The acquisition of the English determiner phrase by Japanese and Spanish learners of English. Doctoral dissertation, University of Essex.

Snape, N. (2008). Resetting the Nominal Mapping Parameter in L2 English: Definite article use and the count–mass distinction. Bilingualism: Language and Cognition, 11, 63-79. doi: 10.1017/s1366728907003215

Taler, V., & Jarema, G. (2007). Lexical access in younger and older adults: The case of the mass/count distinction. Canadian Journal of Experimental Psychology/Revue canadienne de psychologie expérimentale, 61, 21-34. doi: 10.1037/cjep2007003

Tryzna, M. (2009). Questioning the validity of the Article Choice Parameter and Fluctuation Hypothesis: Evidence from L2 English article use by L1 Polish and L1 Mandarin Chinese Speakers. In M. del P. García Mayo & R. Hawkins, (Eds.) Second language acquisition of articles: Empirical findings and theoretical implications (Vol. 49, pp. 67-86). Amsterdam: John Benjamins Publishing Company.

UCLES (2001). Oxford Quick Placement Test. Oxford University Press. White, L. (2003). Second language acquisition and Universal Grammar. Cambridge University

Press. Yasui, M. (1996). A better guide to English grammar. Tokyo: Kaitakusha. Yokokawa, H. (2006). Database for L2 research & pedagogy: English word familiarity for

Japanese EFL learners -Letters-. Tokyo: Kuroshio Publisher. Yokokawa, H. (2009). Database for L2 research & pedagogy: English word familiarity for

Japanese EFL learners -Sounds-. Tokyo: Kuroshio Publisher. Yoon, K. K. (1993). Challenging prototype descriptions: Perception of noun countability and

indefinite vs. zero article use. International Review of Applied Linguistics in Language Teaching (IRAL), 31, 269-289. doi: 10.1515/iral.1993.31.4.269

Zhao, H., & MacWhinney, B. (2018). The instructed learning of form–function mappings in the English article system. The Modern Language Journal, 102, 99-119. doi: doi:10.1111/modl.12449