The Acquisition of English Articles by Second Language Learners

Despite the numerous studies on English article acquisition, the acquisition sequence, differences, and difficulties by English-as-a-second language learners (ELLs) remain to be further explored. With a detailed literature review, this study investigated the acquisition sequence, differences, and difficulties of English articles by ELLs with diverse first language backgrounds. Eighteen college ELLs participated in the study. Data were collected through cloze exercises. The results, through the analysis of its database of 6,178 article used, indicated that “zero article” was the last to be acquired while article “a” is the first to be acquired; positive correspondence existed between the growth of the acquisition of article “a/an,” “the,” and “zero” between “+Art” (with article system) group and “−Art” (without article system) group. The results also revealed that (a) the most difficult article for participants in both groups to acquire was “zero article,” (b) “the” was the most difficult article for participants in “+Art” group, and (c) “a” was the easiest one acquired by both groups. These findings, in fact, challenge the commonsense belief that “+Art” ELLs acquired English articles faster than “−Art” ELLs and that when two languages were similar, positive transfer would occur.


Studies on Article Acquisition by L1 Learners
Despite the abundant studies on second language (L2) acquisition (Geranpayeh, 2000), the literature has yet revealed limited studies on article acquisition by ELLs. Before the exploration of English article acquisition by ELLs, it is essential to review the major studies that relate to the acquisition of English articles by L1 learners.
Outstanding contributors to the study of article acquisition by L1 learners are Zehler and Brewer (1982), Whitman (1974), Bickerton (1981Bickerton ( , 1984, Maratsos (1974Maratsos ( , 2009, Brown (1973), Warden (1976), Karmiloff-Smith (1979), and Chesterman (1991). Zehler and Brewer (1982) examined the sequence and principles in article system used by L1 learners. They found an early acquisition sequence of zero article use, a use only, essentially correct a and the patterns of use, and overextended the use.
Based on the assumption that English article structure was a sequence of quantification and determination rather than a choice between specified and unspecified, Whitman (1974) recommended pedagogical sequence.
According to Bickerton (1981), the most comprehensive examination of the acquisition of English articles was that of Maratsos (1974Maratsos ( , 2009) who confirmed Brown's (1973) naturalistic observation that the article system was mastered at a very early stage by L1 learners. Bickerton (1981) also found that some of Maratsos's (1974Maratsos's ( , 2009 findings were challenged by Warden (1976) and Karmiloff-Smith (1979) who questioned the earliness with which the definite-nondefinite distinction was acquired.
In his language bioprogram hypothesis, Bickerton (1984) indicated that there was a SNSD (specific-nonspecific distinction) by English-speaking children. When examining child's early language, he found an outstanding higher percentage of articles demonstrated in specific-reference NP, whereas zero articles would have continuously existed in the non-specific environment longer than anywhere else. Hence, he claimed the innateness of the SNSD. That is, human beings have an innate ability to make distinctions of specific-and non-specific-reference NP. The SNSD would be quite impossible to learn by means of linguistic data. It is innate. Bickerton (1981Bickerton ( , 1984 also stated that only non-specific reference was marked by zero article, but the persistence of the notion in advanced-proficiency of non-native English speakers and the evidence from the pidgin studies suggested that unambiguous specific reference was also likely connected to the zero article (as cited in Master, 1997).
Although Brown (1973) and Maratsos (1974Maratsos ( , 2009) believed that the article system was mastered at a very early stage by L1 learners, Bickerton's (1981Bickerton's ( , 1984 theory of SNSD, as aforementioned, emphasized the innate perspective of article acquisition. However, Chesterman (1991) challenged the theory of SNSD, claiming that it was not helpful to link article use too directly to noun class, and hence, to the distinction between count and non-count.

Bioprogram and Article Acquisition by ELLs
Worthwhile to mention are Ionin (2003) and Ionin, Ko, and Wexler (2004), who applied Bickerton's bioprogram theory to English articles acquisition by ELLs. Ionin (2003) found that through UG (universal grammar), ELLs could assess semantic distinctions beneath article choice and there was a specificity distinction that discriminated the definiteness distinction. Ionin et al. (2004) investigated the role of specificity in article semantics and asserted that articles could interpret the feature +definite or the feature +specific. They found that ELLs could approach universal semantic features +definite and +specific and there was direct access to universal semantic features and +specific features.

Studies on Sequence, Differences, and Difficulties of Article Acquisition by ELLs
Unlike the acquisition of English article by L1 learners, the English article system is one of the most difficult aspects of English grammar for ELLs and one of the last to be fully acquired (Master, 1987). Following are some outstanding studies on the article acquisition sequence, differences, and difficulties by ELLs of various L1 backgrounds. Ekiert (2004) examined the L2 development sequence of article acquisition by Polish-speaking ELLs and found evidence supporting the hypothesis that the sequence of L2 article acquisition mainly followed the L1 natural order of article acquisition. Lu (2001) investigated acquisition sequence by Chinese-speaking ELLs and found an order of the = a > ∅, and the > a > ∅, among all groups.
Through the study on the acquisition of the English definite article by Chinese-and Malay-speaking ELLs, Wong and Quek (2007) found that the acquisition sequence of the four non-genetic uses of the followed a natural order of situation > structural > textual > cultural, regardless of their L1 backgrounds, and the L2 rate of accuracy on article usage improved outstandingly as proficiency level increased. Master (1990Master ( , 1997 suggested that the English article system could be learned through the separation between classification (a and zero) and identification (the). He claimed that speakers of −Art needed more time to acquire the article system than +Art speakers did, but they would eventually acquire it. zero-for-the errors would continue to occur until advanced interlanguage for −Art speakers. He found that +Art group produced larger number of correct answers than the −Art group. The English articles acquisition order was null > the > a in the (−Art) group and the > null > a in the (+Art) group.
Through the examination of the acquisition sequence of English article functions, Park (1996) found that French and German students (+Art) earned higher scores than the Japanese and Korean students (−Art). In addition, the acquisition order of English article functions was similar no matter whether or not L1 had an article system. Snape, García-Mayo, and Gurel (2013) studied L2 acquisition of English generic NPs by Spanish-speaking ELLs (+Art), Turkish-speaking ELLs whose L1 had an indefinite article but no definite article, and Japanese-speaking ELLs (−Art). They found that L2 article preference was mainly influenced by their L1.
Ionin, Montrul, Kim, and Philippov (2011) investigated whether Russian-and Korean-speaking ELLs (−Art) could discriminate different kinds of English genericity. Their study showed that although their L2 showed zero article for plural nouns and indefinite article for singular generics, it did not demonstrate the acquisition of definite article for singular generics.
Ionin and Montrul (2010) examined L2 acquisition of the expression of plural NPs. The results of the study supported the hypothesis that Spanish-speaking ELLs (+Art) over-accepted the generic expression of English definite plurals far more than Korean-speaking ELLs whose L1 had no article system (−Art). Ionin, Zubizarreta, and Philippov (2009) investigated English article use by ELLs of adults and children, whose L1 was Russian (−Art). They found that both groups of learners demonstrated sensitivity to definiteness and specificity.
Ionin, Zubizarreta, and Maldonado (2008) explored three sources of knowledge in the acquisition of English articles by L2 learners: L1-transfer, L2-input, and Universal Grammar. They found that all three sources were related to the English article acquisition.
Through the examination on L1 transfer in article acquisition, Sharma (2005) found that the new article system produced by L2 learners showed no resemblance with their L1 article system. Definite article the did not disappear completely in their speech when their L1 had no article, although L1 transfer was obvious when it contained the specific indefinite article a. Zegarac (2004) investigated the relevance theory (Sperber & Wilson, 1995) for the acquisition of English article "the" by ELLs of −Art. He found that L2 learning was influenced not only by L1 transfer but also through the process of designing and testing hypotheses. Romaine (2003) found that ELLs of +Art made faster progress than those without one. Mayo and Pilar (2008) found that (a) the "four nongeneric use" of articles demonstrated different difficulty levels for the ELLs, (b) ELLs' "underuse of obligatory the" decreased outstandingly from elementary to low-intermediate level until they moved up to advanced levels (pp. 550-565), (c) the participants' performance in the overuse of the was strongly influenced by their L1 and improved significantly with proficiency level, and (d) the difficulty level for an ELL setting was substantial. Geranpayeh (2000) examined the difficulties of English article acquisition by Persian-speaking ELLs and found that these ELLs had problems identifying the English definite marker when it was in the subject position. Chen (2000) indicated that English article could be one of the most difficult grammatical components for Taiwanese ELLs because L1 (Chinese) did not have article system. He also claimed that errors made by these ELLs were due to L1 interference.

Purpose of the Study
Despite the outstanding aforementioned studies on English article acquisition, the acquisition sequence, differences, and difficulties by ELLs remain to be further explored. Based on the definition of articles and literature review of article acquisition mentioned above, this study examined the sequence, differences, and difficulties of article acquisition demonstrated by ELLs with various L1 backgrounds.

Participants
The participants of the study (Table 1) are 18 college students learning English as a second language (ESL) at a language institute in the United States, with age ranging from 20-(from18) to 30+. The participants were divided into three levels according to the level of ESL class they were attending.
Level A participants were students placed in advanced classes according to college placement testing (with an equivalent Test of English as a Foreign Language [TOFEL] score of approximately 430-480). Among them, three were male and three were female. They had various L1 backgrounds-Korean, Chinese, Spanish, French, and Hebrew and had been in the United States for at least 2 years.
Level B participants were those who were placed in the intermediate classes based on college placement test (with an equivalent TOFEL score of 300-430). Among them, two were male and four were female. Their L1s were Polish, French, Spanish, Russian, and Urdu. They had been in the United States for at least 9 months.
Level C participants were students attending the beginners' classes according to college placement test (with an equivalent TOFEL score below 300). Among them, two were female and four were male. Their L1s were Spanish, French, Urdu, and Bangla, and they had been in the United States for at least 3 months.

Data Collection
Data were collected by means of cloze exercises (sample, Appendix A). Participants were asked to complete conversation, dialogues, sentences, and paragraphs with article a or an, the, or zero. The participants were required to complete the cloze in 45 to 60 min.

Data Analysis
After the cloze exercises of the 18 participants were completed, the answers were corrected according to the answer keys provided by the resource. The scores were categorized into three groups: Level A (Advanced), Level B (Intermediate), and Level C (Beginning) under the forms of article a or an, the, and zero, and the percentage and raw scores were documented (sample table of data analysis, Appendix B).

Reliability of the Data
To insure the reliability of the data, three major considerations were taken. First, the same length of time and condition were provided for each group of participants. Second, Kuder-Richardson reliability coefficient (K-R 21) formula (Hatch & Lazaraton, 1991;Richards et al., 1992) was applied to estimate internal consistency. Third, the cloze exercises were corrected according to the answer keys provided by the resources so that subjective judgment could be avoided.

Results of the Studies
The results of the study (Appendix Table C1 The results of the study (Appendix Table D1) reveal that in the group of Level A (Advanced), the participants whose L1s contain article systems demonstrated an accuracy of 83.9% (193 out of 230) for article a or an. They demonstrated an accuracy of 81.4% (285 out of 350) for article the. In the group of Level B (Intermediate), the participants whose L1s contain article systems demonstrated an accuracy of 91.6% (220 out of 240) for article a or an. They demonstrated an accuracy of 86.7% (260 out of 300) for article the. For zero article, the participants showed an accuracy of 50% (185 out of 370). The participants whose L1s do not contain article systems demonstrated an accuracy of 62.5% (150 out of 240) for article a or an. They demonstrated an accuracy of 63.3% (190 out of 300) for article the. For zero article, the participants showed an accuracy of 1.4% (5 out of 370).
In the group of Level C (Beginning), the participants whose L1s contain article systems demonstrated an accuracy

Discussion, Conclusion, and Implications
Through the study (Appendix Table D1), the following was found and discussed: First of all, the scores in Appendix Table D1 indicate that the most significant difference demonstrated in the acquisition of article a or an, article the, and zero article at Level B between ELLs whose L1 had article systems and those whose L1 had no article systems.
Next, significant difference was also demonstrated in zero article acquisition at Levels B and C between ELLs whose L1 had article systems and ELLs whose L1 had no article systems.
Finally, the results shown in Appendix Table D1 indicate that it was consistent with what Master (1987) found: that (+Art) group had produced larger number of correct answers than (−Art) group. However, different finding demonstrates that the acquisition sequence of the English articles was a > the > zero for Levels A and C and the > a > zero for Level B participants in (−Art) group and a > the > zero for Levels A and B participants and a > zero > the for Level C participants in (+Art) group. In other words, the most difficult article for the ELLs in both (−Art) and (+Art) groups to acquire was zero article except for Level C participants in (+Art) group, who tended to have the most difficulty in learning article the, whereas a was the easiest article for the ELLs in both groups except Level B participants in (−Art) group.
In regard to the sequence, differences, and difficulties of the article acquisition by the participants, the following can be concluded based on the results of the study (Table C1, Appendix C): First, zero article was the last one to be acquired and was the most difficult one for L2 learners of all levels; indefinite article a was the first one and the easiest one for L2 learners of all levels to acquire.
Second, a steady but insignificant growth (4.8 points) has demonstrated for the acquisition of indefinite article a from Level C participants to Level B participants and growth (2.2 points) from Level B participants to Level A participants.
Third, for definite article the, Level B participants demonstrated a growth of 9.8 points from Level C participants, but Level A participants showed a decrease of 1.8 points from Level B participants to Level A participants.
Fourth, for zero article, there was a decrease by six points from Level C participants to Level B participants but a significant move up by 14.5 points from Level B participants to Level A participants.
Fifth, the results demonstrated that positive correspondence existed between the growth of the acquisition of indefinite article a and the proficiency level of L2 learners, but no regular correspondence had been demonstrated between the acquisition of definite article the and zero article.
For researchers, educators, and learners in the field, the following implications are suggested: First of all, as aforementioned, the study revealed that the most difficult article for the ELLs in both (−Art and +Art) groups to acquire was zero article, except for Level C participants in (+Art) group who tended to have the most difficulty in learning article the. In fact, this finding, which is uncommonsense (Mayher, 1990;Sun, 2000) in nature, challenges the commonsense belief that ELLs whose L1 contains article systems (+Art) acquired English articles faster than ELLs whose L1 did not contain article systems (−Art; Master, 1990Master, , 1997Park, 1996;Romaine, 2003) and that when two languages were similar, positive transfer would occur (Ekiert, 2004;Master, 1987;Park, 1996;Romaine, 2003;Snape et al., 2013); when they were different, negative transfer would happen (Chen, 2000).
In addition, as shown above, a or an was the easiest article for the ELLs in both groups, except Level B participants in (−Art) group. What are the underlining reasons for these interlanguage phenomena?
Moreover, as aforementioned, Bickerton (1981Bickerton ( , 1984 claimed the innateness of the SNSD. Does the acquisition of the English articles by English language learners associate with SNSD? If the answer is positive, what is the underlining reason for the differences demonstrated in the interlanguage produced by ELLs from various L1 backgrounds with or without article systems? Finally, although Bickerton (1981Bickerton ( , 1984 emphasized the innate aspect of article acquisition, Chesterman (1991) held that it was not helpful to connect article use too directly to noun class, and hence, to the distinction between count and non-count. Will further study focusing on the article distribution on count and non-count nouns by ELLs determine whether or not the distinction ever exists in the interlanguage by ELLs? If the answer is positive, how and why does it occur?

Limitation of the Study
Because the number, age, and L1 backgrounds of the participants in the study were limited, it is necessary to increase the scale of the participants in the study with a larger number, more various age groups, and divers L1 backgrounds before we can claim the universality of the findings demonstrated in the study. b. Dialogue (Azar, 1999, p.  c. Sentence (Azar, 1999, p. 117): 1. We need to get _______ new phone. 2. Alex, would you please answer ______ phone?

Sample of Cloze Exercises
d. Paragraph (Miller & Cohen, 1998, p. 146) Appendix B ________ Religious Society of Friends, commonly referred to as ________ Quakers, was founded in 1652 after George Fox received _____ vision from God on Pendle Hill in Northwest England. ______ vision helped Fox to realize that ______ spiritual presence of God was the basis for _______ Quakers doctrine of the inner light. ____ Quakers believe that _____ spirit of God enters ______ consciousness of both men and women equally and that it is evidenced in human beings most honorable behavior.

Sample Table of Data Analysis
(The participants were required to complete the cloze in 45-60 min.) Appendix C Table C1. Accuracy Rate of the Acquisition of English Articles.

Participants
Article a/an % (score) SD Article the % (score) SD Zero article % (score) SD

Acknowledgments
Beyond words is my appreciation to the professors of New York University, John S. Mayher, Harold Vine, and Frank L. Tang, for their insights and inspirations, and to the participants, for their contributions.

Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.