Corpus Analysis of L2 English Article Usage Patterns & Pedagogical Implications

Abstract

There has been plenty of observational evidence displaying that L2 English learners whose L1 do not have article systems undergo persistent difficulties. It is known that functional categories of grammar are especially difficult for L2 learners, and the Korean language does not have a functional equivalent to English articles. The current study aims to investigate article usage patterns by L1 Korean L2 English learners by employing a large English interlanguage corpus data; ICNALE written corpus data. Particularly, the current study focuses on the analysis of a specific NP, “part-time jobs”. The current research claims that by narrowing down the scope of analysis to a particular NP, one can expect more reliable and meaningful interpretation from L2 learners’ authentic language production, which can ultimately lead to more effective language instruction and better outcomes for second language learners. The current study seeks to describe patterns of L2 English articles across different proficiency levels. According to the results, it was found that learners’ article usage patterns appeared to be more similar to that of the native speaker in the group with higher English proficiency. One of the interesting findings was the frequent use of bare nominals even in the most advanced learners. Based on the corpus analysis, the current study provides the pedagogical implication that teaching structural differences between learners’ L1 and L2 is very important and article instruction should be emphasized in L2 classrooms in EFL settings.

Keywords:

• corpus analysis
• English article
• L2 acquisition
• nominal acquisition

Introduction

In recent years, much research has been conducted on the acquisition of the English article system. A number of previous research indicate that English articles are regarded as one of the notorious features to acquire for second language learners (Master, 1987; Thomas, 1989; Robertson, 2000; Ionin, Ko & Wexler, Snape, 2013; Park, 2014, among many others). Despite the high frequency of L1 production in both written and spoken English (Celce-Murcia & Larsen-Freeman, 1999), the usage of English articles can be opaque to learners (as well as sometimes for native speakers). That is particularly so because English articles represent not only definiteness but also genericity, and the use of articles is also closely linked to the plurality of nouns. Therefore, learners are bound to receive very complex input, which even can be phonologically not very salient. Ekiert (2007) describes English article functions as “a complex set of absent distinctions which are, to some extent arbitrarily mapped onto surface forms” (Ekiert, 2007:1). This feature is known to be particularly difficult for L2 learners whose L1 does not have an equivalent function in their L1 (Cho & Slabakova, 2014; Díez-Bedmar & Papp, 2008; Ionin et al., 2011; Master, 1987; Snape et al., 2013; Snape, 2013). In other words, when some universal meanings such as definiteness, genericity, and plurality are expressed differently in L1 and L2, learners have to figure out how to express the universal meanings in a new language. Accordingly, L2 learners of English are bound to have difficulties when their L1 and L2 are distant.

Several studies have proven that English learners without an article or article-like system in their L1 have difficulties in assigning definite/generic features with the English article system (Cho & Slabakova, 2014; Díez-Bedmar & Papp, 2008; Ionin et al., 2011; Master, 1987; Snape et al., 2013; Snape, 2013). The majority of the article studies have been experimental studies such as production, grammaticality, or acceptability judgment tasks. On the other hand, corpus-based studies are relatively rarely found in terms of article acquisition. Some of the few studies are as follows.

Young (1996) investigated L1 Czech & Slovak learners’ article choices and the study revealed that in lower proficiency levels, learners showed a tendency to employ demonstratives where the definite article was expected. Crosthwaite (2013, 2014, 2016) conducted several corpus analyses and mainly compared L1 Korean and L2 Chinese learners’ noun patterns. The study suggests that there is a role of learners’ L1 in producing determiners. Crosthwaite argues that “bridging relations” are used when L1 English speakers infer article choices, particularly the definite article “the”. In a recent study by Crosthwaite (2019), L2 learners whose language does not have an article or article-like system will have difficulties in acquiring this subtle feature by presenting considerable differences between the L1 and L2 groups.

Leroux and Kendall (2018) explored the acquisition of English articles by L1 Chinese learners and they attempted to analyze article usage patterns using corpus data. The study revealed that article accuracy rates conform to the proficiency level of the participants, and they emphasized the importance of educators’ efforts to examine students’ article errors.

Given the relatively small size of the corpus analysis studies, more corpus-based research should be conducted to support and verify existing experimental studies. It is now fortunate that the number of learner corpus has been increasing recently and it includes the International Corpus of Learner English (ICLE - Granger et al., 2009), the Asian Corpus of English (ACE, 2014), and the International Corpus Network of Asian Learners of English (ICNALE, Ishikawa, 2011, 2013, 2019). Based on the learner corpus data, language use patterns should be investigated. Analyzing learners’ authentic production data can provide meaningful results which can help to facilitate second language acquisition. Meanwhile, as was shown by the abovementioned studies, most of the previous corpus-based studies have explored the role of L1 in the use of the article choices by comparing production patterns from different L1 backgrounds (Crosthwaite et al., 2016; Crosthwaite, 2013, 2014, 2019; Leroux & Kendall, 2018; Young, 1996). To my knowledge, so far, no corpus study has focused on analyzing the development of second language learners. Therefore, the current study attempts to investigate the learners’ pattern of article uses by different proficiency groups using one of the corpus data, ICNALE.

Using the ICNALE data, the current paper attempts to investigate L1 Korean learners’ article production patterns across four proficiency levels, focusing on only one NP that occurred most frequently in the data. The reasons for focusing on a particular NP can be accounted for as follows. Firstly, as it was given in the experiment question prompt, one can anticipate the chosen NP could occur often in the production data compared to other NPs, thus providing enough data for analysis. By focusing on a specific item of utterance, one might expect to have more controlled data, thus possibly yielding a more reliable result. What is more, despite the chronic difficulty in acquiring English articles, existing studies mainly focus on experiments such as acceptability judgment and grammaticality judgment tasks on using articles. I claim that a new approach is needed to identify the cause of the difficulties. In addition, among existing studies, studies on the use of one specific NP are rarely found. This may be because it is not easy and efficient to attain meaningful data on the use of one NP through experimental research. Therefore, I argue that in-depth analysis of large data of one specific NP using large corpus data will lead to interesting and meaningful results and they can be further compared to the results of previous studies.

The current paper is organized as follows. Section 1 introduces the background of the current research. Section 2 presents cross-linguistic differences between English and Korean in terms of denoting (in)definiteness and genericity. Section 3 provides data used in the current research and provides an interpretation of the data analysis. Section 4 suggests pedagogical implications and Section 5 concludes the study.

Linguistic background

The current section displays theoretical background regarding (in)definiteness and genericity. English marks the semantic notion of (in)definiteness via the article system. The English article “the” marks definiteness and the article “a” marks indefiniteness of the NPs. Fregean view claims that the definite article has presuppositions of both existence and uniqueness (Caplan, 2020). According to Heim (1991), the English article “the” has a definiteness meaning and it means a combination of the presuppositions of existence and the uniqueness of the referent. On the other hand, the English article “a” has indefinite meanings and lacks the presuppositions of existence and the uniqueness of the referent. In other words, “definiteness” is a semantic feature that refers to knowledge that is shared between the speaker and the hearer (see Heim, 1991 for more discussion). Therefore, “there” is used to refer to the previously mentioned NP as in (1).

1. I met a girl on the street. […] The girl lives next door.

The indefinite article “a” is used when the referent is introduced for the first time in (1). There is no presupposition that a specific and unique girl exists, thus it is not definite. On the other hand, when the same “girl” is mentioned again, “the” is used because the “girl” is mentioned before and the ‘girl’s uniqueness has been established. Therefore, the speaker and the hearer both can be aware of the existence of the “girl” in the discourse. However, “the” can be used when one intends to give reference to the object that is commonly known to the speaker and the hearer. When the uniqueness is satisfied by mutual world knowledge between the speaker and the hearer, “the” can be used and the following are examples of such cases (adapted from Brown, 1973:345).

• (2) a. I saw the moon tonight. It was really bright.

• b. Look at the ceiling in my office. It is made of glass!

• c. I fixed the engine yesterday. The car does not make noise anymore.

• d. I really like the first sentence of your essay.

In these sentences, “the” is used for the first mention. In (2a), for instance, the moon is unique for both the speaker and the hearer. In (2b), in a room that has one ceiling and “the ceiling” is unique and salient for both the speaker and the hearer. For (2c), “the engine” is also salient by our knowledge that normally cars have one engine. Likewise, ordinal expressions like the last something or the first something by nature meets the uniqueness presuppositions as in (2d). We have discussed the uses of English articles in terms of (in)definiteness and one other feature that English articles denote is genericity.

Many linguists and philosophers established important concepts of genericity, and it includes the distinction between generic NPs(a reference to a kind) and generic sentences(characterizing predications) (Krifka, 1987). Consider examples in (3).

• (3) a. The potato was first cultivated in South America.

• b. Potatoes were introduced into Ireland by the end of the 17^th century.

(Carlson & Pelletier, 1995, p.2) Underlined NPs in (3a) and (3b) make reference to the kind Potato (Solanum tubersum, a scientific name for the potato). The underlined NPs do not designate a specific potato or individual potato. “The potato”(definite singular) and “Potatoes”(bare plural) are called generic or kind-referring NPs. The example in (4) shows generic/characterizing sentences.

• (4) A potato contains vitamin C, amino acids, protein, and thiamine. (modified from Carlson & Pelletier, 1995, p.3)

In characterizing generic sentences in (4), genericity is a feature of the whole sentences or clauses. In other words, the whole sentences express the generalization and regularities of particular facts. The two generics of generic NPs and generic sentences can be combined in one sentence and occur together. Consider the examples in (5).

• (5) a. Potatoes are served whole or mashed as a cooked vegetable.

• b. The potato is highly digestible. (Carlson & Pelletier, 1995, p.3)

(5a) and (5b) have kind referring subjects (potatoes and the potato) and the sentences themselves also refer to regular facts of the kind mentioned in the subject position. In other words, when the predicates or the sentence itself has characterizing meanings, indefinite singular NPs as in (4), bare plural NPs as in (5a), and definite singular NPs as in (5b) can be used. On the other hand, there are some kinds of predicates that require only kind-referring NPs in their argument position.

• (6) a. The lion will become extinct soon.

• b. Lions will become extinct soon.

• c. *A lion will become extinct soon. (partly modified from Carlson & Pelletier, 1995, p.10)

Some predicates like, die out, be extinct, invent, and exterminate require kind-referring NPs in their argument positions. (6a) and (6b) show that “the lion” and “lions” can be used before “extinct” because they are kind-referring NPs. However, “a lion” in (6c) is unacceptable because “indefinite NPs” are not considered as kind-referring. Understanding the two different generics is very important for second language learners. One reason for the importance is that “definite NPs”, “bare plural NPs”, and “indefinite singular NPs” are often and casually considered as the main types of generic NPs (Quirk et al. 1985, 265; Lyons 1977; Carlson 1977). So far, how the English article system is employed to denote the notions of (in)definiteness and genericity has been discussed.

While the English article system is utilized in various contexts, Korean does not have a direct equivalent to those linguistic features. As is widely known, the Korean language does not have an article or article-like system at all. What is more, “(in)definiteness” is not overtly marked in Korean. To encode genericity, the topic particle “nun” is attached after bare singular nouns in Korean (I. Lee, 1991; H. Kim, 1991; Jun, 2001, among many others).

(7) Saja-nun kot myeolchong-hal geosita.

lion-TOP soon extinct-become FT DEC.¹

“Lions/The lion will become extinct.”

As exemplified in (7), when singular nouns are followed by the particle “nun”, the NP containing “nun” can be interpreted as kind-referring. While bare singular forms are used as unmarked generic NP forms in Korean, some argue that plural nouns can be also considered as generic NPs employed with the particle “nun” when the nouns are animate (C. Kim, 2005; Nemoto, 2005).

(8) a. Saram-tul-un tari-ka nea-gae-ta People-PLU-GEN leg-SUB four-CL-DEC² b. ? Uija-tul-un tari-ka nea-gae-ta. Chair-PLU-GEN leg-SUB four-CL-DEC ‘Chairs have four legs’

As shown in (8a), the “Saram(people)” can be pluralized and the (8a) sentence sounds acceptable, while “Uijatul(chairs)” in (8a) sounds odd because the inanimate noun is pluralized. While it is generally agreed that “bare singular” NPs are used to denote genericity in Korean, bare plural NP forms are also acceptable when the nouns are animate.

So far, the current section reviewed how semantic notions of (in)definiteness and genericity are realized in English and Korean, and it was found that while the English article is exploited in various contexts, Korean does not have a direct functional equivalent to encode such semantic notions. Therefore, it implies that acquiring the English article system would be not easy for Korean learners. Furthermore, as for the article uses, the decision-making process would be very complicated because learners should first notice 1) that they should use a linguistic device called “articles or determiners” in English, 2) then learners should decide whether the noun refers to a definite referent or indefinite referent, and also consider whether the NP refers to a generic referent.

Data & analysis

In recent years, the number of available large learner corpus has increased including the International Corpus of Learner English (ICLE-Granger et al., 2009 The Asian Corpus of English (ACE, 2014), and the International Corpus Network of Asian Learners of English (ICNALE, Ishikawa, 2019) The data for current research is sourced from the “written essay” of the International Corpus Network of Asian Learners of English (ICNALE). The current study employed L2 English data from L1 Korean speakers as well as L1 English data for a comparative purpose. ICNALE data is considered to be advantageous for the following reasons over other large corpora. Firstly, the ICNALE provides production data across the different proficiency groups. Therefore, one can investigate the developmental processes of L2 acquisition. The proficiency control of the ICNAEL data can be considered as reliable because various measures were employed, such as Vocabulary Size Test (VST), TOEIC, TOEFL, and IELTS test scores, to map the scores to the Common European Framework (CEFR, Council of Europe, 2001). ICNALE classified the learners into 4 bands: A2, B1_1 (B1 low), B1_2(B1 high), and B2+ based on the learners’ proficiency test scores or VST scores.

Secondly, the data were carefully collected to enhance reliability for any type of contrastive interlanguage analysis (CIA). Thus, prompts and tasks were rigidly controlled and time and the length of essays were also controlled. The two topics given to the learners include: (a) It is important for college students to have a part-time job and (b) Smoking should be completely banned at all the restaurants in the country. While two topics were presented, the current study chose to analyze article usages of the NP, part-time job from the topic (a). The noun phrase “part-time job” was particularly chosen because it is the noun phrase that was used most frequently in the data set. NPs in the topic (b) were also analyzed and “smoking” was used most frequently in the dataset. However, in order to control any potential variables that might give an effect on the choice of articles, the gerund form of “smoking” was not chosen in the current study. AntConc 3.5.8 was used to analyze 300 written essays which comprised of total word tokens of 69,594. Table 1 shows the Top 6 word rankings of the data.

Table 1. Word rankings of the dataset

Ranks	Frequency	Word
1	2459	to
2	2324	time
3	2094	a
4	2025	part
5	1880	the
6	1768	job

Among those word tokens, the most frequently used nouns in the top ranks include “time (2024), part (2025), and job (1768)” as underlined in Table 1. The current study examines how learners choose to express “part-time job” in their written test in terms of the article uses. AntConc 3.5.8. was employed to investigate the number of utterances of each NP type across the different proficiency levels. Five types of NP forms were identified concerning “part-time job”, and the 5 types of NP forms include bare plural, indefinite singular, bare singular, definite singular and definite plural forms.

Table 2 summarizes usage patterns of NPs containing “part time job” in different proficiency levels. At all proficiency levels, both the “indefinite singular” and the “bare singular” forms account for the largest portion of usage. In the lowest proficiency group, A2_0, indefinite singular and bare singular forms account for 36.6% and 46.3 %, respectively, and the proportion of both forms is 82.9%. In the B1_1 group, the indefinite singular and bare singular forms take up 40.6% and 39.4%, respectively, and 80% together. A similar proportion can be found in the B1_2 and the B2_2 groups. B1_2 shows 84.2% usage rates for the indefinite singular and bare singular, and B2_2 displays 78.2% usage rates for them. Around 80% of the proportion was comprised of indefinite singular and bare singular forms in all proficiency groups. One can also find that definite plural forms were barely employed by the learners. It is also noticeable that English native participants used “indefinite singular” forms most frequently by showing 78.58%. Let us consider Figure 1.

Figure 1. Usage rates of indefinite singular and bare singular NP forms.

The figure contains a graph of usage rates of indefinite singular and bare singular NPs by each participant group including A2_0, B1_1, B1_2, and L1_E. The graph shows that usage rates of the indefinite singular increase as the participants’ proficiency level increases.

Table 2. Usage rates of np forms containing the “Part-Time Job”

	A2_0 (n=75)	B1_1 (n=61)	B1_2 (n=88)	B2_2 (n=76)	L1_E (n=200)
BP (part time jobs)	12% (36)	10.6% (36)	9.3% (44)	17.1% (66)	19.42% (88)
indefinite singular (a part time job)	36.6% (108)	40.6% (138)	45.0% (213)	58.2% (224)	78.58% (356)
*bare singular (part time job)	46.3% (139)	39.4 % (134)	39.1% (185)	20.0% (77)	0.0% (0)
definite singular (the part time job)	5.7% (17)	8.8 % (30)	6.6% (31)	4.4% (17)	1.76% (8)
definite plural (the part time jobs)	0.0% (0)	0.6% (2)	0.0% (0)	0.3% (1)	0.2% (1)
Total	100% (300)	100% (340)	100% (473)	100% (385)	100% (453)

One of the interesting findings of the result is that the usage rates of the indefinite singular increase as the participants’ proficiency level increases as shown in Figure 1. The steady increase of indefinite singular forms by L2 learners appeared to be close to the performance of English native speakers who displayed 78.58% usage rates. One reasonable speculation for the high usage rates of “a” can be probably found in the question sentences given to the participants; It is important for college students to have a part-time job. In other words, one can speculate that L2 Korean learners could have repeated the sentences or the nominal phrases within the given prompt. What is more, one other possible interpretation of this phenomenon can be related to the verb-object construction and a topic, “a part time job”. Learners would have had different levels of understanding of this topic. Therefore, it is possible that some learners might have provided lengthy explanations on their understanding of “a part time job” rather than talking about their personal experiences with any part time jobs they have had in the past and it could have affected the learners’ article choices, resulting in the ample use of indefinite articles.

In order to examine the sentences containing “a part time job”, AntCon 3.5.8 was employed to analyze collocated words. It was found that “have a part time job” and “having a part time job” were used quite often, showing 248 and 99 tokens, respectively. In fact, in terms of percentage, 50% of the overall uses of “a part time job (683)” appeared with the verb have and the gerund having. Therefore, one cannot exclude the possibility of repeating the phrase “a part time job” from the given question sentence. Two native speakers participated in the analysis in order to investigate more about the uses of the remainder of “a part time job” except for 50% of the participants repeating it. It was found that in most cases learners correctly used “a” in indefinite contexts. For example, phrases like “do(ing) a part time job(52), get(ting) a part time job (39)”, find(ing) (25) a part time job, need a part time job’ were often found.

One other interesting finding was that learners used “a part time job” in generic sentences to give a definition to “part time job” or to provide characteristic features of “part time job” as shown in (9).

(9) a. A part time job is a job in which an employee does not work full hours.

• b. A part time job gives a social nature to part timer.

• c. A part time job often has the advantage of more flexible hours.

• d. A part time job can enable college students to become an adult in some aspects including a financial thing.

Also, it is very interesting that overuses of “a” were rarely found in the data instead of using “the” in the definite context but instead bare uses of nouns were distinct as in (10).

(10) I agree that college students have a part time job. Because if college students have part time job … … . I need a part time job.

The sentence in (10) is from one of the participants, and it clearly shows fluctuation between the uses of indefinite singular and bare singular forms. Previously, it was briefly mentioned that the bare singular NP form (part time job) was found quite frequently in the dataset. As was displayed in Figure 1, usage rates of bare singular forms decrease as the participants’ proficiency-level increases, and this is not surprising because as learners are bound to learn that the bare singular NP form itself is ungrammatical in English. Nevertheless, an unexpected result was that even in the highest proficiency group, usage rates of the ungrammatical form “bare singular” reached 20%. The usage rates of the bare singular NPs are higher than those of the bare plural NPs. Consider Figure 2 for usage rates of bare plural and singular NP forms.

Figure 2. Usage rates of bare plural and bare singular NP forms.

The figure contains a chart of usage rates of bare plural and bare singular NP forms by each participant group, including A2_0, B1_1, B1_2, B2_2 and L1_E. Interesting findings are bare plural forms are used less than bare singular forms throughout the different proficiency levels.

Figure 2 compares usage rates of bare plural and bare singular NP forms. It is very interesting that usage rates of bare plural forms appear with relatively very low frequency. Bare plural forms are regarded as the most unmarked and default forms of NPs in English and were also claimed to be acquired earlier than other NP forms including definite singular and indefinite singular (Carlson & Pelletier, 1995, Ionin, Ko & Wexler, Snape, 2013; Park, 2014) According to the previous studies, acquisition and development of DP(Determiner Phrase) occurs incrementally and learners start from the bare NP forms and its projection DP (Hawkins, 2001). In a longitudinal study, Huebner (1985) found that learners start with bare NPs, and they overuse the definite article “the”, and use of the article “a” appears afterward. Similar results were found in Parrish (1987) and Anderson (1978). The literature generally consented that the acquisition of the English article system happens incrementally and the order of acquisition is as follows: Bare NP → definite article “the” → indefinite article “a”.

What is more, contrary to the results of previous studies, in the current paper, it can be seen that the usage rates of article “the” were very low. One possible reason for the low usage rates of “the” can be inferred by examining the individual essays. In other words, the phrase “part time job” can be regarded as hypernym and the participants used the “part time job” for the first mention. However, when referring to more specific jobs they have had, the participants showed a tendency to refer the name of the specific job (e.g. the job in the café, the convenience store job, the delivery work, among many others). In other words, “the part time job” should be used in the definite context, but specific words (hyponym) replaced “the part time job” in the given data, thus resulting in the relatively low occurrence of “the part time job”.

So far, article usage or preference patterns of L2 English learners have been analyzed. Although there is a limitation of the current study in that it is limited to only one NP, it appears that more accurate and interesting results were obtained by controlling various variables which might have affected learners’ word choices. For instance, most of all, despite the use of bare form being the earliest phenomenon in the acquisition of the article system, it is very surprising and interesting that bare singular forms, which are ungrammatical expressions in English, were found with quite high usage rates even in the most advanced learner group. Therefore, it is worth discussing why the omission error appears even at the end of the acquisition stage. In the next section, the omission error phenomenon will be discussed concerning article acquisition in the L2 classroom.

Discussion & pedagogical implications

In recent years, there has been an increased interest in investigating and clarifying the varying difficulty of acquiring different language components such as modules, interfaces, operations, and constructions. Researchers such as Lardiere (2008), Slabakova (2006, 2019), and White (2003) have examined these differences. Lardiere (2008), for instance, argued that morphological competence should be treated differently than syntactic competence, leading her to propose the Feature Assembly Hypothesis. This hypothesis suggests that learning a second language involves reconfiguring the formal features of one’s native language and those from Universal Grammar into new configurations in the second language. Language acquisition depends on the assembly and re-assembly of these formal features, which is usually not a straightforward mapping process. Lardiere (2008) also argues that one of the biggest obstacles for L2 learners is selecting and assembling the correct combination of features for their L2 language. Furthermore, feature reassembly is especially challenging in cases where the target features are present in L1 but are structured differently in L2. In such situations, L2 learners must dissociate the features that are manifested in their L1 and reassemble them to match L2 language.

White (2003) explores whether knowledge of inflectional morphology drives learning of syntax or if the reverse is true. She labels these two viewpoints as “morphology-before-syntax” and “syntax-before-morphology.” Slabakova (2006, 2019) suggests the Bottleneck Hypothesis and seeks to explain why some aspects of language are more difficult to learn in a second language than others. By comparing different linguistic modules and their learnability, the hypothesis suggests that functional morphology, which is usually found in functional categories, is the primary obstacle in L2 acquisition. To verify this hypothesis, researchers have tested the learnability of various linguistic features related to functional morphology. For example, studies have examined the acquisition of tense, aspect, and agreement morphology, as well as the use of pronouns, prepositions, and determiners (Guijarro-Fuentes et al., 2017; Jensen et al., 2020; Slabakova, 2015, 2019). The results of these studies have generally supported the Bottleneck Hypothesis, suggesting that functional morphology is indeed a challenging aspect of second language acquisition.

Hong (2019) examines English article omission error that appears in the early grammar of child L2 learners in terms of “Underspecification Analysis”. He examined the article omission errors of L1 English child learners and those of L2 Korean child learners who were in the early stages of acquisition. It was claimed that the omission errors of L1 child acquisition were caused by the underspecified linguistic features. On the other hand, for L2 child acquisition, the omission errors were claimed to be caused by the “Null Determiner Parameter” setting in their L1, in which articles are not required. In other words, unlike L1 child language acquisition, L2 acquisition involves the influence of learners’ L1, thus the acquisition processes between the two are different. Therefore, for L2 learners to be able to produce English articles correctly, they should reset the parameter based on the target language input they receive.

Previous studies which attempt to account for the linguistic features with persistent difficulties commonly presuppose that the learning difficulties stem from differences between L1 and L2. In the same vein, one possible explanation for the persistent uses of bare singular NP forms by Korean adult learners can be explained by nominal features of L1. In Korean, bare singular nouns can be correctly used to represent both singular and plural nouns. Likewise, plurality can be represented by two noun forms including nouns with a plural marker “–tul” and nouns without a plural marker, which can be equivalent to bare singular nouns in English. Therefore, it can be suggested that the bare singular form was frequently employed in L2 production since the bare singular form is used in place of plural nouns in the learners’ L1.

Although further research through comparative research with other native speakers without articles in L1 should be conducted to support the impact of L1, the difficulties that the L1 and L2 discrepancy pose to L2 learners seem to be certain.

In order to understand the bare singular form of nouns that are prevalent in L1 Korean L2 English learners, it may also be meaningful to examine the nominal acquisition of native English language learners. One might wonder whether bare singular NP forms are used by L1 speakers at all at any stage of the acquisition process. Abu-Akel and Bailey (2000) developed a corpus by collecting children’s utterances from 17 native English-speaking children who are 18 months old at 3-month intervals for 2 years. Children utter only bare nominal phrases while omitting articles in the early stages of acquisition, and the correct use of indefinite and definite articles in the determiner phrase was 19% for the 18-month-old children. After 2 years, at 3 years and 6 months, up to 81% of the articles were used correctly. According to previous L1 acquisition studies (Brown, 1973; Warden, 1981), in general, children start uttering “words” when they are 12 months old and start making “phrase structures” through the use of grammar between 18 and 30 months of age. At the early stage of language development, word constructions are made based on content words rather than functional words, and the use of functional words gradually begins to increase. By around the age of three, child learners of English as their L1 can supply correct article forms at 90% (Brown, 1973), and their article choices become indistinguishable from adult English native speakers by the age of five (Warden, 1981). This seems to suggest that, in terms of L1 English acquisition, structural functions of articles are acquired first and meanings are acquired later.

The importance of understanding structural functions before meaning is also found in the L2 acquisition study. Trenkic (2009) investigated the reason why some second language learners present persistent difficulties with article omissions. It was argued that learners whose L1 do not have article systems have persistent difficulties because they focused mainly on article meanings (definite or indefinite), thus analyzing articles as a nominal modifier like demonstratives or adjectives which has the closest meanings to the “a” and “the” in their L1 rather than as grammatical elements. Trenkic (2009) thus suggested that syntactic misanalysis leads to article omission patterns and this pattern can be detected even with learners with high proficiency in English whose article production seems to be native-like at first glance (Trenkic, 2009).

As demonstrated in the foregoing discussion, acknowledging the differences between a learner’s L1 and L2 must be preceded for successful language acquisition. Notably, an essential initial step in this process would be the recognition of syntactical and structural differences between the L2 learners’ L1 and L2, especially with linguistic features of functional categories such as articles. However, these days, achieving communicative skills has been emphasized in language classrooms and such communicative skills encourage learners to employ context, world knowledge, and pragmatic strategies to understand the utterance and thus focus on meanings (Clahsen & Felser, 2006). Therefore, in such circumstances, acknowledging structural differences between L1 and L2 grammar, especially functional categories, can be easily neglected.

What is more, in fact, some L2 acquisition studies argued that it is questionable whether the comprehension of the L2 message and figuring out the L2 syntax are directly related (Cook, 1996; Gass & Selinker, 2001). In the meantime, some researchers believe that paying attention to and focusing on a grammatical form can be beneficial for effective L2 learning (Hong, 2019; K. Lee, 2019; Lopez, 2015) In fact, Lopez (2015) investigated the effect of delivering instructions to adult L2 learners without articles in their L1 in the acquisition of English articles. The empirical data of the study suggested that learners who had been given instructions on the specificity distinction improved in the use of articles. Therefore, communicative competence teaching methods that exclude focus on form may not be the most effective approach for achieving the ultimate objective of second language acquisition, which is to develop a mental grammar of the learners’ L2 language. The results of the current study support a such conclusion and encourage an increased emphasis on teaching and practicing grammar in the L2 teaching classroom.

However, in Korea, grammar instruction especially on English articles is rarely found in elementary, middle, and high school textbooks (K. Lee, 2019). K. Lee (2019) claims that in the current educational situation where grammatical elements are not explicitly taught at all, it may be difficult for students to naturally acquire functional grammar through continuous exposure to the linguistic feature. Krashen and Terrell (1983, p59) pointed out that “while most of the classroom time should be spent on activities that foster acquisition, learning exercises are important in certain cases”. Hong (2019) argues that one meaningful way of assisting the parameter resetting in L2 is to increase the opportunity to write and speak phrases that contain functional words. It was emphasized that teaching the meanings of articles, for example, “the” as definite and “a” as definite, should be accompanied by practicing determiner phrases. That is because words in functional categories are more efficiently acquired by activities of speaking and writing the phrases.

Consequently, the present study posits two key points regarding the second language classroom. Firstly, focusing on form should not be disregarded in the second language classroom, and secondly, apprehending the structural differences between the first and second language ought to serve as the foundation for instructing linguistic features, notably those that encompass relatively intricate features like articles.

Conclusion

The current research analyzed the large corpus of data collected from Korean L2 English learners in relation to the use of English articles. It investigated article usage patterns used with a particular nominal phrase of “part time job”. The result showed that learners’ article usage pattern becomes more native speaker-like as their proficiency level becomes higher. Particularly, in the current analysis, both learner and native speaker groups preferred using indefinite NP forms, and definite singular forms are rarely used. One of the most interesting findings was the high usage rates of the bare singular form across all proficiency levels, including the most advanced groups. Drawing upon the findings from previous research on second language acquisition and the present study, this paper posits that a comprehensive grasp of the distinct forms and structures of both the first language and the second language is crucial for effective language acquisition, particularly in areas where learners encounter persistent difficulties. Thus, the present study presents a pedagogical recommendation that grammar instruction should not be disregarded when acquiring functional categories such as article systems. The study also contends that in order to prevent the fossilization of learners’ grammar, some form of conscious-raising instruction should be provided alongside naturalistic input.

Disclosure statement

No potential conflict of interest was reported by the author.

Additional information

Notes on contributors

Sunyoung Park

Sunyoung Park is a visiting professor of General English Department at Sejong University. She earned her MA and Ph.D. in Language Acquisition at University of Sheffield, U.K., and her research areas are SLA and English Language Teaching(ELT). She has published a journal article titled “Investigation on TOEIC score trends in Korea and its pedagogical implications “with Cogent Education in 2020.

Notes

1. GEN: Generic Marker, FT: Future Tense, DEC: Declarative Sentence.

2. PLU: Plural, GEN: Generic Marker, SUB: Subject, CL: Classifier, DEC: Declarative.