Thinking differently about grammar and metalinguistic understanding in writing

In the light of ongoing international debate about the purpose of explicit teaching of grammar, this paper considers the relationship between metalinguistic understanding and development as a writer. Drawing on a cumulative series of studies over a period of ten years, adopting a functionally-oriented approach to grammar, the paper argues that purposeful grammar teaching occurs within the teaching of writing, not divorced from it; and that this teaching develops students’ metalinguistic understanding of how written texts are crafted and shaped. In this way, grammar is positioned as a resource for learning about writing and one which can support students in becoming increasingly autonomous and agentic decision-makers in writing. We show through practical examples how the pedagogy works in practice, and through classroom interaction data we highlight how metalinguistic talk (metatalk), which enables and encourages the verbalisation of choice. The data also shows, however, that teachers’ skill in managing metatalk about metalinguistic choices in writing is critical in framing students’ capacity to think metalinguistically about their writing and to be autonomous writerly decision-makers.


INTRODUCTION
The question of grammar and its role or value in the teaching of writing is one which has been beset by controversy, uncertainty and vacillation between its inclusion in, or its expunction from, the curriculum. This argument has been well-documented, particularly in Anglophone contexts: see for example, the discussion of the 'grammar debate' in Kolln and Hancock (2005); Locke (2010); and Myhill and Watson (2014). More recently, Boivin et al. (2018) and Fontich and Garcia-Folgado (2018) have drawn attention to broadly similar debates in countries whose first language is not English. At the heart of the issue has been the concern that merely teaching grammar makes no difference to students' capabilities as language users, particularly their capacity to write well. Successive studies and meta-analyses have argued that teaching grammar has no beneficial impact on writing competence (Braddock, Lloyd-Jones and Schoer, 1963;Hillocks, 1984;Hillocks and Smith, 1991;Andrews et al., 2006;Graham and Perin, 2007). Indeed, Graham and Perin maintained that there was evidence of a negative effect for 'the explicit and systematic teaching of the parts of speech and structure of sentences ' (2007, p. 21), echoing Braddock et al.'s earlier claim that teaching grammar 'has a negligible or, because it usually displaces some instruction and practice in actual composition, even a harmful effect on the improvement of writing' (Braddock et al, 1963, p. 37).
However, it is not the purpose of this paper to focus on whether grammar should be taught or not. It is important, nonetheless, to note that the body of empirical research against grammar teaching is not strong and the same studies are often cited in each successive meta-analysis. Significantly, Braddock et al.'s concern, cited above, that grammar has a harmful effect because it usually displaces the teaching of writing strikes to the heart of the matter. The notion of grammar teaching is dominantly represented in terms of systematic teaching of the structure of the language, with a focus on identifying and labelling grammatical constructions. No connection is made between knowing the structure of a language and being a user of that language. It is the argument of this paper that purposeful grammar teaching occurs within the teaching of writing, not divorced from it; and that this teaching develops students' metalinguistic understanding of how written texts are crafted and shaped. In this way, grammar is positioned as a resource for learning about writing and one which can support students in becoming increasingly autonomous and agentic decision-makers in writing.

RE-CONNECTING GRAMMAR AND WRITING
Theoretically, our representation of grammar as fundamentally connected to the teaching of writing, and to learning about being a writer, draws on a Hallidayan conceptualisation of grammar as social semiotic. It adopts a fundamentally functionally-oriented perspective on grammar, rather than a form-focused one. Halliday argues that functional grammar goes beyond the mere description of the structure of language: it enables us 'to show the grammar as a meaning-making resource and to describe grammatical categories by reference to what they mean -an insightful mode of entry to the study of discourse' (Halliday and Matthiessen, 2004, p. 10). In other words, the fundamental emphasis of a functionally-oriented grammar is on the relationship between form and meaning. To illustrate this, consider the following extract from an English novel for children based on Arthurian legend:

Extract 1
Sitting alone in the room was a girl -no, rather a woman -and beside her a harp. As I strained to see better, I slipped noiselessly on the wet cobbles. But so intent was she on her playing that she did not hear me and she did not look up. Her fingers plucked effortlessly. It was her fingers, long, white and dancing, that I loved first. Her hair was the colour of honey, of gold washed in milk. It fell over her face so that I could not see her. But I did not need to, for I knew already she would be perfect.
From Arthur, High King of Britain by Michael Morpurgo (1994) In this extract, the reader is introduced to this character (Guinevere) for the first time. Although there are many possibilities offered by this extract for discussing connections between grammatical structures and how they make meaning, one possibility is to focus on character description and how the noun phrases create a strong visual first image of Guinevere: • 'a girl -no, rather a woman'; • 'her fingers, long, white and dancing'; • [her hair is] 'the colour of honey, of gold washed in milk' Not only do these noun phrases establish a visual impression of Guinevere, but they invite the reader to infer what kind of a woman she is, and what the narrator thinks of her. Thus, attention to the noun phrases in this specific context makes a connection between the grammar (the noun phrases) and how they are functioning in the writing (to create a visual image and to invite inference).
Thinking about grammar from a functional perspective in this way represents a re-alignment of attention to grammar with how it is functioning in a text. It stands in contrast to the form-focused view of grammar which is more concerned with grammatical accuracy and compliance to grammatical rules in writing. Derewianka and Jones (2016) contrast these two ways of thinking (Table 1) to draw out the different way that a Hallidayan theorisation of grammar operates, a pattern of contrasts which illustrates the different pedagogical possibilities that arise when grammar is re-connected to writing.

The Traditional (Form-Focused) Approach A Hallidayan (Functional) Approach
Describes language in terms of word classes; nouns, prepositions etc Describes language in terms of the relationship between form and function of language eg an adverb can provide information about an action (He ran swiftly), or provide a comment (Luckily he escaped).
Operates at the level of the sentence and below Deals with language from text to word level, and looks at the interaction of these.

Describes the grammar of written language
Describes how written language differs from spoken language and is used with multi-modal texts.
Sees language as a set of rules to be followed Sees language as a resource and seeks to extend students' potential to make meaning more effectively.

Focuses on grammatical accuracy
Values well-structured sentences, but goes beyond structure to include other functions of language Presents a decontextualized view of language Systematically describes how the choices we make in using language are influenced by factors in the context Uses a pedagogy typically concerned with naming and labelling grammatical forms, often through inauthentic exercises Draws on a scaffolding cycle that relates students' knowledge about language to the kinds of meanings they need to make in various areas of the curriculum and in their daily lives. Halliday and Matthiessen (2004, p. 49) argue that grammar is 'a network of inter-related meaningful choices'. The notion of 'grammar as choice' is central to the idea of re-connecting grammar with writing, and chimes with the distinction which Carter and McCarthy (2006, p. 7) make between the grammar of structure, and grammar as choice, where one is principally concerned with analysing and identifying the language as a system, whereas the other is principally concerned with the consideration of the different effects of different grammatical choices. This way of thinking about grammar is one which has gained increasing traction recently, not only in our own research, but internationally, as evidenced in the arguments made in Table 2.  Maun and Myhill 2005, p. 20 United Kingdom The purpose of grammar is to 'enable pupils to make choices from among a range of linguistic resources, and to be aware of the effects of different choices on the rhetorical power of their writing' Lefstein 2009, p. 382 United Kingdom/Israel.
'We need to understand the possible power effects of our choices. We need to understand how our ideational choices construct participants, processes, and circumstances from a particular perspective; we need to attend to our choices of mood and modality, which encode relations of authority and agency between writers and readers; we need to think about how textual choices work to foreground and background ideas, to construct cause and effect, to position information as old or new. Janks 2009, p. 130 South Africa We can encourage 'writers to recognize and use the grammatical and stylistic choices available to them and to understand the rhetorical effects those choices can have on their readers'

Kolln and Gray 2016: Introduction
We can explore this notion of grammar as choice by returning to the example, quoted earlier, from children's author, Michael Morpurgo. He chose to describe Guinevere's fingers as she played the harp in this way: It was her fingers, long, white and dancing, that I loved first.
His choice to position the three adjectives (long, white and dancing) after the noun is an unusual position in English speech, though it is more common in literary text. He could instead have chosen to write: It was her long, white and dancing fingers that I loved first.
Both versions are grammatically correct, highlighting that grammatical accuracy is not the touchstone of effective writing. More relevant to the argument here, however, is that Morpurgo chose the first version, which raises the question of why. The positioning of the adjectives after the noun emphasises the description of the fingers through altering the rhythm of the sentence: the reader's attention is drawn to the detail of 'long, white and dancing', and also, of course, to what that description makes the reader infer about Guinevere. The point here is not that one sentence is better than the other, but that writers can make choices to match their own authorial intention at that point in the narrative. In other words, we can help writers understand how the language choices they make are not arbitrary, but central to how they communicate their ideas.

EXPLICIT GRAMMATICAL KNOWLEDGE
A functionally-oriented approach to the teaching of grammar, with an emphasis on grammar as choice, thus connects learning about grammar with learning about writing. It also draws on and potentially develops explicit grammatical knowledge. This is another point which has frequently been misunderstood in the contestation of the value of grammar teaching. Those who argue against grammar teaching often rightly point out that native speakers of a language have an immense amount of tacit grammatical knowledge, demonstrated in their capacity to speak and to write, and this tacit knowledge is built through experience as a language user. This position is wholly correct: being able to identify a word as noun always follows being able to use nouns in speech and writing to communicate, and being able to identify nouns does not make us any better users of nouns.
However, this position fails to address some key points about language and learning. Tacit knowledge is knowledge that we cannot articulate (Polanyi, 1966;Hetherington, 2011), whereas explicit knowledge can be verbalised, shared and systematised. In everyday life, tacit knowledge is immensely useful and efficient: for example, we know how to recognise people from their faces, we learn how to ride a bike without knowing the physics of balance and motion, and we learn to speak without knowing the rules of grammar. But in a learning context, tacit knowledge is more problematic. Nonaka and Takeuchi (1995) developed a model of tacit and explicit knowledge for use in organisational learning, which drew attention to the difficulty of sharing tacit knowledge. In the writing classroom, it is difficult to share tacit knowledge about effective writing, because of the inability to verbalise it. Explicit grammatical knowledge, however, is accessible and usable grammatical knowledge: it is 'learning' knowledge, as it can be used to develop greater understanding of how to write, how to solve writing problems, and to share thinking about grammatical choices in writing.
At the same time, it is important to acknowledge that students do develop tacit knowledge about writing from their reading, and will draw on that tacit knowledge when shaping texts. We cannot know whether Morpurgo made a conscious choice to position the adjectives after the noun in the example quoted earlier, or whether this drew on his tacit knowledge as a reader. But Morpurgo is an experienced and professional writer, not a student learning about writing, and in many classrooms in the 21 st century there are substantial numbers of children who do not read the kind of texts that school expects them to write, and thus do not access the models that might build this tacit knowledge.
Thus, in the writing classroom, generating explicit grammatical knowledge about the relationship between grammatical choices and rhetorical effects is pedagogically important. It opens up possibilities for shared discussion about language choices and for the development of a shared language for talking about writing. It shows developing writers how texts are shaped and how meanings are contextually bound, and generates a sense of the repertoire of possibilities that are available to all writers as they write. Like Martin, (1985), we would also argue that explicitness about the linguistic patterns and expectations of different written texts is also socially just, ensuring that socially-disadvantaged students are given access to understanding of how texts work, which more advantaged students have acquired through their different cultural capital.

METALINGUISTIC UNDERSTANDING
Thus far, this paper has made an argument that purposeful grammar teaching occurs within the teaching of writing, making meaningful connections for learners between grammatical knowledge and writerly knowledge. We argue that this is achieved through adopting a functionally-oriented theoretical framing of grammar, and through developing writers' understanding of the possibilities of particular grammatical choices in writing. The learning power of explicit grammatical knowledge about writing, discussed above, rests in its capacity to support students' metalinguistic understanding about writing.
We know from research that metacognition is a predictor of success in writing (Kellogg, 1994;Sharples, 1999) because writing always needs to be self-monitored. Writers need metacognitive knowledge of writing to: • plan, because planning requires strategic advanced thinking about the task and your intentions as a writer (Hayes, 1996, p. 2); • develop 'a model of their audience, for reflecting on rhetorical and content probabilities . . . for monitoring their progress . . .' (Kellogg, 1994, p. 213); • revise, because revision is always a conscious activity, requiring metacognitive decision-making (Alamargot and Chanquoy, 2001, p. 108).
In a nutshell, metacognition in writing refers to our capacity to think about what we know and can do in writing, and to think about ourselves as writers and how we manage the writing process. Crucial to the argument of this paper, the focus of metacognition is more on the 'how' of learning, than the 'what'.
Metalinguistic understanding is generally understood as a subset of metacognition (Gombert, 1992, p. 13), the element that focuses directly on language use. As with metacognition, the 'meta' part of the word refers to moving to a different level of awareness or perspective. So metalinguistic understanding is moving beyond the merely linguistic, to thinking about and understanding the linguistics. In other words, it is about explicit understanding of language structure and choice: it is concerned with looking at language, not just using language. Gombert (1992, p. 13) explains metalinguistic understanding as comprising two parts: firstly, as 'activities of reflection on language and its use' and secondly, as the 'subject's ability intentionally to monitor and plan their own methods of linguistic processing (in both comprehension and production)'. Building on this, our own definition of metalinguistic understanding deliberately draws on different disciplinary perspectives on writing: cognitive, because it is about thinking; linguistic, because it is about language; and socio-cultural, because it is about how writing is a social practice: Metalinguistic understanding is 'the explicit bringing into consciousness of an attention to language as an artifact, and the conscious monitoring and manipulation of language to create desired meanings, grounded in socially shared understandings' (Myhill, 2011, p. 250).
Creating opportunities in the language classroom for learners to talk about, investigate and explore, and reflect on language use in the texts they read and in their own writing builds metalinguistic understanding. Ribas, Fontich and Guasch (2014) argue for the importance of supporting metalinguistic activity in the classroom, fostering the abstract thinking necessary to rich metalinguistic understanding. This supports growing independence as a writer, helping students to make choices and decisions that are informed by an understanding of how texts work, and through an increasingly developed sense of authorial intention.

FROM THEORY TO PEDAGOGY
Our own research (Myhill et al., 2012;Jones et al., 2013;Myhill et al., 2013;Watson and Newman, 2017;Myhill et al., 2018a;Myhill et al., 2018b) has taken this theoretical framing of a productive role for grammar in the writing classroom and the development of writerly metalinguistic understanding to develop a corresponding pedagogy and to research its efficacy. The design principles underpinning the pedagogy were informed by the theoretical ideas discussed thus far: namely, that purposeful teaching adopts a functionally-oriented approach promoting grammar as choice; that explicit grammatical knowledge about writing is addressed; and that fostering metalinguistic understanding for writing is at the learning heart of the pedagogy. This theory-pedagogy relationship is realised through what we now call the LEAD principles, explained in Table 3. To illustrate this pedagogy in practice, let us return to the Morpurgo example used earlier. The example is taken from a teaching unit, devised by the research team, using the LEAD principles. It is designed for primary school children, aged 10-11, and focuses on learning about writing fictional narrative. The Morpurgo text ('Arthur, High King of Britain') is used as the authentic text and the whole book is read with the class. By the end of the unit, the students write their own fictional narrative, a new Arthurian legend, which focuses on the development of character and effective shaping of plot. In several lessons, the class look at creating character descriptions in different ways, and in one lesson, they are focusing on providing detail through post-modification of the noun phrase, and on how the positioning of adjectives after the noun emphasises the visual detail of the adjectives. In the previous lesson, the students had been looking at physical descriptions of characters being created through nouns and adjectives, and they had visualised a character of their own and undertaken some free-writing, playing with initial ideas for their own character descriptions. The learning sequence for one lesson is outlined below, with commentary on the implementation of the LEAD principles.
1. The teacher uses Powerpoint to explain how noun phrases can build description, briefly recapping on pre-modification, before addressing more specifically some of the ways nouns can be post-modified. The teacher leads discussion on how these expanded noun phrases, all taken from the Morpurgo text, build the description of the characters (see Figure 1).

Figure 1. PowerPoint Slide
You can also build noun phrases by adding more description after the noun You could add adjectives: • her fingers, long, white and dancing, • her eyes, wide and intense, • a lady, dark-haired and beautiful, You could add a prepositional phrase: • the colour of honey • the hood of his dark cloak You could add a non-finite clause beginning with an -ing or -ed verb: • gold washed in milk • a lady, dark-haired and beautiful, wearing a gown of wine-red 2. The teacher shows the students a Powerpoint slide with an image of an old man, and a reminder of some of the post-modified noun phrases describing characters in the book. The teacher then invites students to work in pairs, and, looking at the image of the old man on the Powerpoint slide, to talk together about what the character might be like and how he might be described him. The pairs then generate some well-chosen noun phrases to describe what he looks like (see Figure 2).

CREATE YOUR OWN NOUN PHRASES TO DESCRIBE THIS CHARACTER AND PAINT A PIC-TURE OF HIM IN WORDS
Commentary: the teacher invites students to explore in collaborative writing the Link between noun phrases and character description, still using Examples from an Authentic text, and setting up peer-to-peer Discussion.
3. As a whole class, the teacher leads the Joint Composition of a three or four sentence description of the old man's character, taking choices from the class, discussing the choices and how well they capture the visual details. The discussion focuses on the precise choice of nouns, whether the adjectives add more information/precision, and the possibilities of detail provided through postmodification.
Commentary: the teacher builds understanding of the Link between noun phrases and character description through joint composition, drawing on students' own writing, and through whole-class Discussion of the efficacy of the noun phrase choices.
4. The teachers asks students to work individually, going to the previous lesson's freewriting character description. The students are encouraged to read it aloud in their heads and then to rewrite it as a paragraph of no more than 100 words describing the character, trying to improve the quality of the description, perhaps by extending some of the noun phrases with information which follows Commentary: the teacher consolidates individual students' understanding of the Link between noun phrases and character description, giving them time to develop their own character description for their Arthurian story, and focusing their attention on the Link between the noun phrases they choose and the character description they want to create. 5. Plenary: as a whole class, the students play a game based on a British television programme (The Voice). The class have to decide who can stay in the court of Camelot and who has to go, based on how strongly they can picture the character. Six children sit at the front of the class on chairs facing away from the class, playing the role of judges. Four children read their character description aloud to the whole class, and the six judges at the front each decides if they stay or go.

Commentary: this final activity introduces an element of fun, but at the same time helps the student writers to test out the effectiveness, with real readers (or listeners!), of the noun phrase choices they have made to create their character descriptions (Link).
This practical example of teaching, exemplifying the LEAD principles in action, shows how purposeful grammar teaching can be fully integrated within the teaching of writing, with the aim of developing writerly metalinguistic understanding. However, our empirical research has indicated that the quality of the metalinguistic discussion in this pedagogy is critical in enabling effective learning about writing to occur (Myhill and Newman, 2016;Watson and Newman, 2017). Thus, this paper will now spotlight the metalinguistic talk component of this pedagogy, and draw out patterns of effective and less effective practice.

GOING META! METATALK TO PROMOTE METALINGUISTIC THINKING
Central to the argument presented here, that purposeful grammar teaching occurs within the teaching of writing, is the parallel claim that this teaching develops students' metalinguistic understanding of how written texts are crafted and shaped. We argue that high-quality, dialogic talk is a key tool in developing this metalinguistic understanding, and in facilitating transfer of learning into students' writing. There is, of course, a historical recognition of the importance of talk in writing development, particularly Britton's seminal statement that 'reading and writing float on a sea of talk ' (1983, p. 11). But this research and professional materials have always been more attentive to talk for writing, to generate ideas for writing, rather than to talk about writing, which develops more specific understanding of the complex ways in which writing creates meanings. To address this, we have borrowed the concept of 'metatalk' from L2 research to refer to talk about language use in writing.
In L2 research, metatalk (Swain, 1995;1998) describes metalinguistic reflection on language use, and tends to be form-focused; in the L1 context, we retain the idea of metatalk as 'language used for cognitive purposes' (Swain, 1998, p. 69), but are more concerned with understanding linguistic choices in writing as functionally-oriented (Halliday, 2004), rather than form-oriented. So our interest is not in subject-verb agreement or management of tense per se, which are often key concerns for L2 learners, but in supporting growing awareness of how linguistic choices subtly alter the way a text conveys its communicative message. For example, a third person narrative establishes a different relationship with the reader than a first person narrative, and being able to recognise and discriminate between these choices is an important aspect of developing as a writer. Metatalk, therefore, is talk about writing which encourages metalinguistic thinking, and which has the following characteristics or attributes: • A specific kind of (meta)talk about writing with a focus on language use; • It encourages the articulation of thinking about linguistic choices; • It is a way of exploring the relationship between a writer's authorial intention, the linguistic choices which realise that intention, and the intended effect on the reader; • It is a pedagogical tool which, through enabling and encouraging this verbalisation of choice, allows teachers to determine and extend the level of metalinguistic thinking and understanding that students have developed; • It is dialogic: it can be used 'to teach students to think-to make knowledge' (Resnick et al., 2015) and to 'open up discourse space for exploration and varied opinions' (Boyd and Markarian 2015, p. 273).
Through detailed analysis of transcriptions of classroom discourse, we have been able to identify some of the ways in which teachers differentially realise these characteristics and attributes. Sometimes teachers remain driven by a strong sense of their own authority and retain close control of the discussion, with the result that the class discussion is principally about determining what the 'right answer' is in the teacher's head: Sometimes this closed questioning with a pre-determined answer is accompanied by praise with no explanation of why the choice is a good one: The unexplained praise tells students they have the 'right' answer, but in both these examples, the teacher does not probe the student's metalinguistic understanding. In the first example, the idea that noun phrases are more important than adjectives is a misunderstanding, because adjectives are frequently in noun phrases. And in the second example, what does the student think is the impact of the short sentence after this exchange? Closed, teacher-led interactions like these also lead to missed opportunities to follow up on student thinking represented in their responses.
More successful classroom talk opened up the discussion much more, inviting students to think metalinguistically and verbalise their thinking. The nature of the initial question in opening up was significant, as in the exchange below where the teacher first asks a closed question to clarify understanding of the key grammatical point, then asks a question that opens up thinking about the link between a grammatical choice and its rhetorical effect:

Extract 5
Teacher: What is the subject of the sentence?

Student: The sword
Teacher: Why do you think he's chosen to do it this way round? Why has he left 'the shining sword' -the subject -until later in the sentence?
Sometimes, because these are real teachers in real classrooms leading discussion in-the-moment, it is possible to see both good questions which open up thinking, and missed opportunities to explore students' thinking:

Extract 6
Teacher: So the verb is in front, but normally we have it after the subject. Why do you think the writer inversed, changed this around? In this example, the two responses by the students suggest misunderstandings, both grammatical and rhetorical, which could have been brought to the fore and explored as a class. But the teacher pursues her questioning, rather than pausing to explore the metalinguistic thinking implied in their responses.
In other interactions, the teacher not only asks opening up questions, but builds on student responses by probing for more explanation of their thinking:

Extract 7
Teacher: Why is that such a good sentence? Student: They've described it well.
Teacher: Yes he has -but from the reader's point of view, what's just happened? Student: He made the reader wait.
Teacher: Good, but how has he done that, what has he done? Student: He put 'the ring of fine gold' at the end. Not until the end of the sentence do we find out what it is.
Here, the teacher pushes for more justification or elaboration of the answer provided, leading in the first example to a very well-verbalised explanation of the effect of the subject-verb reversal being discussed.
In summary, our analyses of authentic whole class teacher-student interactions, drawn from several different studies, shows that less effective metatalk occurs when the teacher's questioning seeks 'right answers' and limits metalinguistic discussion; when the teacher tends to foster a view that certain grammar forms should be used; and when the teacher misses opportunities to follow through on students' responses to extend metalinguistic understanding. In contrast, more effective metatalk occurs when the teacher asks questions which open up metalinguistic thinking, rather than closing it down; when the teacher's questions probe for more explanation or elaboration of metalinguistic thinking, and when talk opportunities are structured to link grammar and meaning, drawing attention to how grammatical choice relates to the reader-writer awareness, authorial intention and the effect on the reader.

CONCLUSION
This paper has argued that purposeful grammar teaching occurs within the teaching of writing, not divorced from it; and that this teaching develops students' metalinguistic understanding of how written texts are crafted and shaped. In particular, we maintain that adopting a functionally-oriented approach to grammar focuses on meaning -how we write something is as important as what we writeand that showing learners the possibilities of different grammatical choices can enable them to have more conscious control of how their writing communicates their intended message. Alongside this, we suggest that explicit grammatical knowledge is pedagogically usable 'learning' knowledge: developing greater metalinguistic understanding of writing and being a writer, and allowing teachers to access students' metalinguistic thinking about writing. This can be realised through metatalk, enabling and encouraging the verbalisation of choice. However, teachers' skill in managing metatalk about metalinguistic choices in writing is critical in framing students' capacity to think metalinguistically about their writing and to be autonomous writerly decision-makers.