Article Pedagogy: Encouraging Links between Linguistic Theory and Teaching Practice

Introduction

The English article system is a notorious source of difficulty for L2 learners. Whilst there is some variation depending on learners’ L1s, article omission and misuse (i.e. using the instead of a/an and vice versa) persist in the production of L2 learners of English even at advanced proficiency levels. English articles are difficult to acquire because multiple functions are represented by one morpheme (Master, 2002). That is, a single article might encode gender, number, definiteness, specificity or genericity for the noun it modifies. The persisting errors amongst L2 learners of English indicate that teaching English articles can be challenging.

Early studies in article pedagogy include Kaluza (1963), Whitman (1974), and Pica (1983). Kaluza (1963) proposed an approach for teaching Slavonic learners (no L1 articles) which centred on presenting English articles with nouns to make them meaningful. In another study, Whitman (1974) criticized describing a/an and the as ‘articles’ and suggested that a/an are associated with quantification whereas the is associated with determination. He provided six pedagogical steps for teaching the English articles: (1) quantity, (2) general plural, (3) non-count nouns, (4) determiners, (5) quantity and determiner, and (6) generic articles. Pica (1983) focussed on missing information from English teaching textbooks, and argued for including discourse-related rules in article pedagogy. She associated articles with communicative competence rather than linguistic competence and recommended using dialogues to present learners with examples of article use.

Building on this work, Master (1990; Master, 1994; Master, 1997; Master, 2002) introduced several proposals for improving article instruction for L2 learners of English. Master also developed a pedagogical framework (1997), which highlights suitable approaches to article instruction and recommends what should be taught at different proficiency levels. The main premise of his Binary System (1990) is a one-form/one-function approach and, as part of his framework (1997), he highlighted that this is appropriate for teaching articles to intermediate learners. In 2002, Master presented and tested Information Structure. This discourse-related method presents given information prior to new information; a practice called end-focus, so that learners are sure whether to use the (for given information) or a/an (for new information).

Following this work on article pedagogy, a body of theoretical research developed, building on a proposal from Ionin, Ko and Wexler (2004). To date, this theoretical work has had little influence on pedagogy and the two bodies of complementary research remain largely separate. The aim of the current article is to synthesize the insights from both fields in order to better understand the most effective method for teaching articles to L2 learners of English. We begin by explaining recent developments in theoretical work. This will then be contrasted with Master’s (1997) framework. The article ends with an examination of current teaching materials, where we make specific suggestions for how grammatical information can be made more linguistically accurate.

Recent Theoretical Work on Article Acquisition

A number of hypotheses have been presented in the last decade to explain why English articles are so problematic for L2 learners. One of the most influential proposals comes from Ionin, Ko and Wexler (2004), who provide a theoretically-grounded explanation for substitution errors in L2 article production. As stated above, a single article might encode gender, number, definiteness, specificity or genericity for the noun it modifies and, in formal linguistic approaches to article analysis, definiteness, specificity and genericity are discourse-related features of article meaning that contribute to article choice (Lyons, 1999).

Ionin et al., (2004) dealt mainly with definiteness and specificity. In English, article selection is based on definiteness and therefore specificity does not require a different article form as it is interpreted from context. The difference between the features is explained in (1) and (2).

The combination of definiteness and specificity leads to four contexts, as demonstrated in examples 3–6 (from Lyons, 1999: 167).

Based on these distinctions, Ionin et al., (2004) proposed the Article Choice Parameter (ACP), which shows that languages with two articles either encode articles on the basis of definiteness (such as English) or specificity (such as Samoan), but not both. The associated Fluctuation Hypothesis (FH) predicts that learners will have access to universal patterns of article choice, and will fluctuate between definiteness and specificity until the input guides them to set the appropriate parametric value. Fluctuation would therefore result in substitution errors that can be detected in contexts where definiteness and specificity do not match. Ionin et al., (2004) predicted that a will be overused in [+definite, −specific] contexts and the will be overused in [−definite; +specific] ones (examples 4 and 5, respectively). Following a further evaluation of Samoan articles by Tryzna (2009) and Ionin, Zubizaretta and Philippov (2009), the original prediction of the FH was revised to just overuse of the in [−definite; +specific] contexts. The two possible article groupings, based on the evaluation by Ionin et al., (2009), are presented in Figure 1.

OPEN IN VIEWER

Figure 1.

Article Grouping Cross-linguistically: Two-article Languages (from Ionin et al., 2009: 341).

Many studies have provided empirical support for the FH. Ionin et al., (2004) measured learners’ article choice using a forced choice elicitation task and a written production task. In the elicitation task, participants chose between the, a/an or Ø in dialogues, based on a context established for each one. Participants were intermediate or advanced level Russian and Korean speakers and neither language has a morphologically-realized article system. Analysis of the elicitation task revealed that both learner groups made significantly more article substitution errors in [+definite, −specific] and [−definite, +specific] contexts. In the production task, there were some cases of the overuse with indefinites (see 7), but few cases of a overuse with definites could be linked to specificity.

Results confirm the predictions of the original version of the FH, and it was concluded that errors in L2 English article choice reflect L2 learners’ access to the universal settings of definiteness and specificity.

In a further study, Ionin et al., (2009) compared L1 Russian adult and child learners. They found that children only made errors in [−definite, +specific] contexts whereas adult learners made errors in both contexts, possibly due to misuse of explicit strategies. The adult results replicate those of Ionin et al., (2004) but the child data represents the groupings shown in Figure 1 (above). Therefore, the child data supports the updated proposal for overuse of the in [−definite; +specific] contexts, and it was suggested that adults overextend these errors to an additional context. Based on these results, predictions for article choice are shown in Table 1.

OPEN IN VIEWER

Table 1.

Predictions for Article Choice in L2 English.

	[+definite]: target the	[−definite]: target a
+specific	correct use of the	overuse of the
−specific	correct use of the	correct use of a

Note: This is a revised version of the table from Ionin et al., 2004: 19.

Ionin et al.’s work (2004, Ionin et al., 2009) led many researchers to explore the L2 acquisition of English articles by learners from other language backgrounds, including those with an L1 article system. García-Mayo (2009) tested low-intermediate and advanced Spanish learners of English (whose L1 article system encodes definiteness) and found that they transfer their knowledge of Spanish articles when acquiring English and therefore do not fluctuate. Similarly, Sarko (2009) tested the FH on two populations whose L1 article systems encode definiteness (Syrian Arabic and French). The Syrian Arabic group differed significantly from the native control group in [−definite, +specific] contexts, whereas the French group did not. While this result suggests fluctuation patterns, Sarko argues for an L1 transfer effect. The Syrian Arabic learners mainly overused the when the noun was modified by a relative clause with an overt complementiser (see 8 for an example of target use), and no overuse of the was detected in absence of relative clause modifiers.

Trenkic (2008) tested the assumptions made by the ACP and the FH on L1 Mandarin/L2 English bilinguals. Even though she criticized Ionin et al.’s operationalization of specificity and presented an argument against the FH, ¹ she found a pattern in Mandarin learners’ production similar to Ionin et al.’s (2004) learners. Trenkic (2008) postulates that the substitution errors do not result from specificity; rather they occur because of the extra-linguistic factor of stated versus denied familiarity of the referent. Snape (2009) also explored article use amongst young adult Mandarin Chinese speaking or mainland Chinese learners of English at intermediate level. He tested three hypotheses, including the FH with a focus on the [−definite, +specific] context, and group results were consistent with the FH. Therefore, there is a substantial body of work which supports Ionin et al.’s proposal of systematic patterns of errors based on specificity effects.

A third semantic property of articles is genericity, which has previously been overlooked in L2 acquisition research (Ionin et al., 2011). Genericity indicates general reference, compared to definiteness and specificity where the referent is a particular individual(s) or object(s). Krifka et al., (1995) highlight the distinction between two varieties of generic structure. Noun-phrase Generics (NPG) refer to a well-established kind (examples 9a, and 9b). Sentence-level Generics (SLG) indicate generalizations based on the characteristics of individual objects (examples 10a, and 10b). NPGs are compatible with definite singulars and bare plurals. In contrast, SLGs are compatible with indefinite singulars and bare plurals (Krifka et al., 1995). Consequently, the type of generic structure determines article choice in a given context, and examples 9c and 10c demonstrate that article misuse will make a generic sentence ungrammatical.

From a pedagogical viewpoint, theoretical research such as Ionin et al., (2004; Ionin et al., 2009; Ionin et al., 2011) and others provides information about contexts that may be problematic for learners and could help teachers to target these. This body of work implies that teaching the following information (11, 12) could help L2 learners to overcome some of the causes of article misuse errors.

Some recent intervention studies (Snape and Yusa, 2013; Sabir, 2015; Snape et al., 2016; Lopez, forthcoming; Umeda et al., 2017) have applied this theoretical information to classroom instruction, as discussed next.

Recent Research in Article Pedagogy

The resurgence of interest in article acquisition in the last decade, arguably prompted by Ionin et al.’s (2004) seminal theoretical work, coincides with a corresponding resurgence of interest in article pedagogy. Several intervention studies have explored alternative methods of teaching articles in the hope of overcoming the well-attested difficulties experienced by English L2 learners.

Snape and Yusa’s (2013) pilot study provides one such example. They pre- and post-tested two groups (experimental and control) of Japanese learners (n=7 in each group) using a forced choice elicitation task (from Ionin et al., 2004), an acceptability judgement task (from Ionin et al., 2011) and a transcription task. The experimental group received two 70-minute instruction sessions on article semantics (one on definiteness/specificity and another on genericity) and a third 70-minute session on the perception of articles, over a period of three weeks. The control group received no instruction.

In their instruction, Snape and Yusa (2013) focussed on how definiteness and specificity function in English. Their instruction associated specificity with speaker knowledge, and whether they have a particular individual in mind. Citing Ionin et al., (2004), they expected learners to err in [+definite, −specific] contexts since Japanese has no articles (−ART) and errors in this context are not caused by parametric differences (Ionin et al., 2009). Snape and Yusa (2013) found no effect for instruction on definiteness, specificity, and genericity, though the participants did improve on their perception. They conclude that articles are very complex, compared to other properties of grammar, and the instruction period was too short. For genericity, Snape and Yusa (2013) added that learners may be basing article choice on noun countability rather than genericity, and suggested that instruction should focus on the difference between indefinite generics and definite generics.

A follow-up study conducted by Snape et al., (2016) applied a longer intervention period of nine weeks, with weekly 60-minute lessons on articles. Instruction was delivered in the L1 and L2 to assist learners to understand the subtleties of article semantics. Upper-intermediate and advanced level Japanese learners were tested using an acceptability judgement task. Snape et al., (2016) note that the instruction group improved significantly on three post-tests targeting genericity (at three weeks, nine weeks, and 12 weeks). The control group’s performance on this task did not change. For the same study, Umeda et al., (2017) include data from a delayed post-test conducted one year after instruction ended and found that, by this time, participants had returned to pre-test levels. Due to the short-term benefit, they conclude that theoretically-informed instruction on genericity is beneficial for high-level learners’ explicit knowledge when delivered over a sustained period, but that it does not improve their implicit knowledge of this form.

Two further studies (Lopez, forthcoming; Sabir, 2015) also explored instruction on article semantics. Lopez (forthcoming) delivered instruction on specificity to Chinese learners of English. Three groups of low-intermediate L1-Chinese learners of English (n=50) were tested before and after a teaching intervention using an untimed written elicitation task (Ionin et al., 2009) and a timed judgement task. The Specificity Instruction group was taught about definiteness and specificity using materials informed by Ionin et al.’s (2004) theoretical framework. The Standard Instruction group received instruction on definiteness using standard teaching materials and the No Instruction (control) group was not taught about articles. All groups showed similar levels of improvement on the elicitation task. Results for the timed judgement task differed between the groups. The No Instruction and Standard Instruction groups made significant improvements whilst the Specificity Instruction group did not improve. Lopez (forthcoming) concluded that article instruction is not beneficial for this group of learners. She stated that the complex construct of specificity is difficult to operationalize in teaching materials aimed at less proficient learners, and by simplifying the concept in order to make it more accessible, core elements of the definition were changed. This may have presented difficulties for the learners.

Sabir (2015) looked at both genericity and specificity in an intervention study including 67 Saudi (Hejazi) Arabic-speaking learners of English divided into four groups, and 23 native English speakers. The participants took three tasks (article elicitation, acceptability judgement and elicited written production) as a pre-test, an immediate post-test and a delayed post-test. Over three weeks, each group was provided with either implicit instruction (exposure to articles with no grammatical focus) or explicit instruction on definiteness, specificity and genericity, and to either translation or gap-fill activities that targeted article use in the contexts highlighted by Ionin et al., (2004; Ionin et al., 2011). Most groups fluctuated in [−definite, +specific] contexts in the pre-test, which suggests a specificity effect. Learners also distinguished between genericity types even though this distinction is not morphologically marked in Arabic, but showed evidence of L1 transfer in article generic use/interpretation (overuse of the). The study shows that explicit instruction and activity type did not have a clear effect on article accuracy, as there was no pattern of improvement in the immediate and delayed post-tests.

To summarize, these studies tested the effects of instruction on definiteness, specificity and/or genericity. Snape et al., (2016) and Umeda et al., (2017) found a positive short-term effect of instruction on genericity when instruction was delivered over a sustained period, but there was no long-term benefit.

Master’s Article Pedagogy

The studies discussed so far are not the first instances wherein theoretical research has inspired a review of article pedagogy. As stated above, Master (1990; Master, 1994; Master, 1997; Master, 2002) made several recommendations for article instruction based on acquisition research.

In his earliest work, Master (1990) proposes teaching English articles to intermediate-level learners as a binary division between classification (a and Ø) and identification (the). He states that these concepts should be introduced before rules of article usage. The main premise for the Binary System is a one-form/one-function approach to instruction. Master (1990) collapsed the features definite and specific into one, the outcome being that specificity as a feature was ignored. According to Master (1990), instruction on article usage involves familiarizing students with notions such as countability, first mention, subsequent mention, post-modification, proper nouns, and idiomatic phrases, often using multiple examples.

The implications of research for article pedagogy were further considered in Master (1997). By considering overuse of Ø in relation to L1 influence, the frequency of Ø in the input, and overuse of the which appears in learners’ production after they realize that Ø is not grammatically appropriate in all contexts of usage, Master (1997) concludes that speakers of an L1 with articles (+ART) overuse the at the early stages of acquisition and show less use of Ø than −ART speakers. Thus, +ART learners are considered one level ahead of their −ART counterparts, suggesting that the latter group needs more time to acquire the English articles.

Master’s empirical work (1994; Master, 2002) studies the effects of article instruction and provides support for the teachability of the English articles. Both studies apply a pre-test/post-test design, with his 1994 research examining the performance of learners on a forced choice elicitation task before and after systematic instruction on the English articles. His 2002 paper, in contrast, compared instruction on Information Structure with traditional article instruction. The relevant aspect of Information Structure to articles is giveness which focusses on presenting given (old) information prior to new information (end-focus). In both studies, participants were from mixed L1 groups, and the treatment group performed significantly better than the control group on the post-test. Furthermore, in both papers Master emphasizes the importance of appropriate use by English L2 learners at intermediate level and higher because article errors in writing could lead to comprehension difficulties for readers. The results of these two studies suggest that instruction on articles is beneficial to these groups of learners, despite the lack of control for whether the participants have an L1 article system which may be transferred to English.

The culmination of Master’s work on the English articles is best expressed at the end of his 1997 paper, where he presents a framework for article pedagogy. This framework provides recommendations about article instruction, with different advice for three proficiency levels (beginner, intermediate and advanced). For beginners, Master stipulates that focussing on rules of article usage is not worthwhile. His assumption is based on Little’s (1994) claims that beginners cannot fully understand linguistic rules since their L2 mental lexicon is not sufficiently developed to express complex meanings. However, Master recommends including a/an when vocabulary items are first presented to beginners (e.g. an apple versus rice) and further recommends the use of photographs or realia to indicate whether a noun is count or non-count. Additionally, he recommends that mass nouns should be presented to beginners later and that focus on the is best avoided except for the names of countries.

For intermediate learners, Master recommends the use of his Binary System (1990), as outlined above, and seeks evidence for the benefits of this system from his 1994 study. He suggests that a sufficient amount of time should be dedicated to each distinction, and that each distinction should be taught separately and be practised by exercises. Alternative approaches, also advocated by Master (1997) for intermediate level learners, include Information Structure (later tested in Master, 2002) and Processing Instruction (Van Patten and Cadierno, 1993). For advanced learners, Master (1997) declares that rules are no longer interesting to them and suggests that articles are better learned by −ART learners from context as part of lexical items. He also suggests that teachers should encourage high proficiency learners to keep a record of their errors and reflect on these by themselves.

What we conclude from reviewing Master’s (1997) pedagogical framework is that it was empirically informed and based on the results of his own research. In contrast, we also believe that Master’s insistence on a simplification of pedagogical grammars at the expense of linguistic accuracy (Master, 1990; Master, 1994) calls into question the benefit of applying all of these generalizations. Particularly, we disagree with his claim that specificity and definiteness should be compressed into identification. This very avoidance of contexts where there is a mismatch between definiteness and specificity means that learners are not given the opportunity to overcome a potential source of errors. Despite his argument against teaching specificity, Master is a vocal advocate of article instruction per se, especially as written accuracy is important to more advanced learners and this cannot be achieved without control of the English articles. To conclude, we will now examine published teaching materials in light of the insights provided by these bodies of research.

Standard Article Pedagogy

Here we will assess how closely Master’s (1997) framework matches article instruction in published texts, and consider whether some areas would benefit from more linguistically-informed explanations. A review of five general English coursebook series was carried out to show how articles may be taught to L2 learners of English. These titles were chosen because they are well-known and widely-used general English coursebooks which have been published since the proposal of Master’s framework. The books are: New English File (Oxenden and Latham-Koenig, 2006; Oxenden and Latham-Koenig, 2010; Oxenden et al., 2004); Cutting Edge (Cunningham and Moor, 2003; Cunningham and Moor, 2005; Cunningham and Moor, 2007); Language Leader (Cotton et al., 2008; Cotton et al., 2010; Lebeau and Rees, 2008); Global (Clandfield and Jeffries, 2011; Clandfield and Pickering, 2010; Clandfield and Pickering, 2011); and New Headway (Soars and Soars, 2000; Soars and Soars, 2003a; Soars and Soars, 2003b; Soars and Soars, 2006). For each series, the elementary, intermediate, and advanced level books were reviewed. From this review it is clear that most coursebooks teach articles at every level. All the materials focus only on definiteness and, as far as we are aware, no published language teaching materials provide rules on the specific/non-specific contrast or on the distinction between the genericity types. A possible reason why these details are missing is due to space restrictions in textbooks.

Articles are known to be complex for L2 learners but are introduced at an elementary level in all of the books reviewed. The rules at this level revolve around whether a noun is count or non-count, mass or concrete and whether it starts with a vowel or consonant (to account for the use of a/an). The Language Leader Elementary coursebook (Lebeau and Rees, 2008) presents the grammatical structure to learners first, and it is then practised. An extract of information from the elementary level book is shown in (13).

An important point to note in this example is the rule explaining that the definite article is used to talk about ‘a specific person or thing’ (Lebeau and Rees, 2008: 33). This is linguistically inaccurate and may lead learners to falsely overgeneralize that all specific contexts require the and, vice versa, that all definite contexts are obligatorily specific. This is just one example, but the same terminology is used in other teaching materials from other publishers. A small number of practice exercises follow these explanations in the Language Leader coursebooks, and each book also contains a ‘language reference and extra practice’ section.

The New English File books also introduce articles at elementary level (Oxenden et al., 2004), but only a/an. It is covered in three chapters, either alongside vocabulary items or with demonstratives or quantifiers. This appears to be motivated by vocabulary rather than grammar, which was recommended by Master (1997: 226) for beginner level learners. Cutting Edge (Cunningham and Moor, 2005) and New Headway (Soars and Soars, 2006) take the same approach of introducing the indefinite article to elementary level learners, as does Global (Clandfield and Pickering, 2010). Thus, all three of these books are really teaching vocabulary with little or no mention of the uses of a/an, as recommended by Master (1997). By the intermediate level book, Global (Clandfield and Pickering, 2011) has introduced both the and a/an although not the zero article, and it provides six different uses of the article (three for the, three for a/an). In contrast, the New Headway pre-intermediate book (Soars and Soars, 2000) focusses on finding examples of article use in a text then correcting some sentences using this information. The New Headway series takes the approach of loading learners with information about articles and only providing minimal opportunities to practise. In fact, this review suggests that it is a common occurrence.

The rules given in textbooks for genericity revolve around the idea of ‘general’ reference with no explicit mention of the types of genericity (NPG and SLG). As the first published work that dealt with teaching generics, Snape and Yusa (2013: 167) state, ‘no textbook mentions that there are two types of genericity: NP-level and sentence-level generic sentences’ even though this distinction is important for teaching the English articles. In pedagogic research, we saw that Master’s (1990) Binary System distinguished between generic and specific structures, but not between the two varieties of genericity. Moreover, Master consulted Whitman’s (1974) approach, which stipulated that genericity is better taught at later stages because it is not commonly found in the input, unlike referential structures. Because English language textbooks usually offer very little instruction about using articles in generic contexts and never mention the meaning distinction (Ionin et al., 2011), the distinction between NPG and SLG provides interesting implications for article pedagogy.

In conclusion, there are clear links between Master’s framework for article pedagogy (1997) and the instruction provided to elementary level learners in that the focus is on introducing a/an alongside lexical items. However, his recommendations for intermediate and advanced learners are not reflected in standard instruction materials. We found no evidence of instruction for intermediate level learners which applied the three approaches recommended by Master (1997). In addition, there are some instances of linguistically inaccurate terminology (such as specific being used to mean definite), and other key information about genericity that is missing. Although only five series of books were reviewed, these same issues appear to be widespread in most grammar instruction materials. The books we reviewed continue to teach articles to advanced learners and, at every level, appear to offer little opportunities for learners to explore the different uses. Our review shows that there is a divide between what theory suggests in terms of article acquisition and what is usually taught to English language learners.

Conclusion

In the two decades since Master’s (1997) framework was published there has been a proliferation of research into the acquisition of articles, and the aim of the current essay is to bridge the separate areas represented in theoretical and pedagogical publications. Research spearheaded by Ionin et al., (2004; Ionin et al., 2009; Ionin et al., 2011) into specificity and genericity has highlighted the gap between potential causes of errors and the knowledge provided to L2 learners of English (and their teachers). A review of textbooks highlights this divide, in the sense that specificity and genericity types are ignored. Additionally, the term specific is sometimes presented as synonymous with definite, which is linguistically inaccurate and may lead learners to false overgeneralizations.

When applying the specificity and genericity features to article pedagogy there were mixed results. Just one study found a short-term improvement following instruction on genericity (Snape et al., 2016; Umeda et al., 2017) with highly proficient learners who received a large quantity of instruction that, arguably, would not be replicable in real classrooms. Three other intervention studies (Snape and Yusa, 2013; Sabir, 2015; Lopez, forthcoming) found no such benefit for instruction on specificity. It appears that instruction on specificity does not improve learners’ article knowledge. In contrast, instruction on genericity was found to be effective in the short-term (Snape et al., 2016; Umeda et al., 2017) and more work on this area is warranted.

Despite this resurgence of interest in the acquisition and instruction of English articles, what is still lacking is some clear measure of the effectiveness of Master’s recommendations, and we suggest this as an interesting area for future research. What we conclude by reviewing work on article pedagogy in the last 20 years is that revival in the field is evident, and that a contribution to a successful well-acknowledged framework for teaching the English articles requires consideration of both theory and practice. We see the revival of interest in article pedagogy as promising, and hope that such work continues so that we can eventually uncover a reliable, empirically-tested framework for teaching the English articles that can be replicated in real instruction at all levels of proficiency.