To get good writers, first get good sentences

Most teachers have first-hand experience of being asked to explain something beyond their knowledge, and it is their unpleasant memories of this, not their stubbornness or cultural Marxism, which often makes them resistant to teaching the grammar of ‘good sentences’. The many structures that constitute language interlock in complex and fascinating ways, so that the impulse to use ‘grammar’ to teach students ‘sentence level skills’ can, in so many cases, gallop uncontrollably ahead of understanding. Many teachers know that they don’t fully understand, and feel uncomfortable putting themselves forward as experts. This type of discomfort is usually quite an alien experience for the politicians, consultants and managers who push grammar or ‘sentence level work’ as a solution to writing ills: few of these characters are actual experts in language, but most are oddly unwilling to consider that that fact may be of significance.

My advice for anyone who is asked by someone with a larger pay packet than theirs to use ‘grammar’ to teach students to write ‘good sentences’ is this: ask them to explain the differences between a subject, an agent and a topic. Some may be able to, in which case work with them and learn from them. If they can’t, then feel free to be a little stroppy about it. Or, as a third option, read what follows, and you may be able to be the one doing the explaining.

If the verb-job is the top of the tree of English grammar, the subject is what comes next – it is what the verb is true of, and is a necessary element to form most clauses. But the idea of a subject is rather limited, and fails to capture two essential concepts that, while overlapping with it, are importantly different. These are the agent – what does the verb, if it is an action, and the topic, which can be thought of what the clause is about – in English the topic always comes first, so it can also help to think of it as the clause’s launching point.

In a basic, bog standard clause, all three of these roles are fulfilled by the same words:

The dog chased the rabbit through the backyard

Here, ‘chased’ is true of the dog, so ‘the dog’ is subject. The dog is also what actually did the chasing, so ‘the dog’ is agent, and the dog is, of course, at the start, so ‘the dog’ is topic – the launching point. This clause therefore exemplifies bog-standard grammar. But ‘bog-standard’ is of course only the beginning: in a large number of clauses, this alignment of the three roles in one word or phrase is broken, and it is the alternating adherence to, and manipulation of, the ‘typical’ structure which is the essence of grammatical choice in both speech and writing. We can, then, mess with our example clause so that the dog loses one or more of these roles, and we can do some of this without even changing the clause’s core meaning:

Through the back-yard the dog chased the rabbit   (dog is agent and subject, but not topic)

The rabbit was chased though the backyard by the dog  (dog is agent, but not topic or subject)

And by changing the meaning, and yet maintaining the same content words, we can also separate the dog from the action, so that it is no longer even the agent:

The rabbit chased the dog through the backyard. (rabbit is now topic, agent and subject)

And there are also many sentences in which the verb used does not even describe an action, so there is no agent at all:

The dog’s teeth scared me

The collar seems expensive

Of these three roles – subject, agent, topic – the topic is in many ways the most important in terms of explaining how texts work. Regrettably, it is also almost certainly the one which is least understood and taught. The topic’s importance to good communication lies in the fact that it is the most psychological of the concepts – it’s all about the way that speaker and audience actually manage and process the message that’s being conveyed. The reason the topic always comes first is that it is, by definition, the launching point: what the clause is about, rather than why it is happening. Conversely, what follows the topic (what, following one tradition, I shall call the comment) is the information which actually motivates the clause psychologically, and which is the focus of attention at that point in the flow of information.

This disentangling of what the clause is about (topic) from what is for (comment) can seem quite confusing, and I think one reason for this is the unfortunate, if understandable, habit we have of presenting and thinking about clauses in isolation. The truth, of course, is that a single clause is very rarely all there is of a text. Most texts include many, and it is the grammatical management of topics and clauses across clauses that makes for good communication. Let’s return to our dog to see why:

Standing aimlessly at the kitchen window, I caught sight of a strange dog. The dog chased a rabbit through the back yard.

Now we can feel more clearly why the dog was the topic in the underlined clause: we already knew about him. He popped up as the comment of the previous sentence, and so was fresh in our audience’s mind when we moved to the next sentence. As fresh, but familiar, material, he makes for a good topic, and since the point of the clause is to let us know what he was up to (chasing rabbits), we leave this bit towards the end, making it the comment. This ordering of information means that the audience has a much better chance of properly processing the news about the rabbit-chasing: they’ve already been given time to get prepped for it by our reminder about the dog (that is, the topic).

Language thus surfs on an expectation that clauses will generally move in waves from topics – given launching points – to comments – focussed information that is the actual point of our message. The term ‘wave’ signals the way that the topic-comment distinction lacks the mathematical precision that some other syntactic categories possess – or at least are typically seen as possessing. It is less easy to draw a distinct boundary between these two elements than it is, say, to identify that in the clause, that dog has bitten everyone, that dog is clearly the subject and has bitten the verb phrase that applies to it. Instead, the start of the topic may be thought of as the peak of one wave of importance, and the start of the comment as the peak of the next, with the space in between a kind of fluid movement from one peak to the other.

This may seem complex, and in some ways it is, but there is at least one way in which the topic-comment structure is simpler than other elements of grammar: it contains just those two terms, and no more. The only catch to this simplicity is that, since they can apply to clauses as well as words and phrases, topic-comment structures can nestle inside each other. In the following sentence, the second, underlined clause is all a comment with regard to what precedes it, whilst the two sets of bolded words constitute lower level comments within each of the clauses:

Because that man left the gate open, I fell into the hole

With this notion of waves of importance in mind, we can start to see why managing the topic-comment distinction is so important to clear communication.  When the expectation is thwarted, messages often become less easy for audiences to manage:

Standing aimlessly at the kitchen window, I caught sight of a strange dog. A rabbit was chased by the dog through the back yard.

While not awful, the underlined clause is somehow odd. As topic, the rabbit jumps in and confuses us, and after that it’s slightly irritating to have to hear about the dog afterwards when we already knew about him. Of course, awkwardness is not always a mistake: in literary texts, for example, some writers make a great virtue out of sometimes making us feel this way. But in the run of the mill situation, we strive not to irritate or awkwardify our language. Managing a constant and broadly predictable flow from topic to comment is the essential way that we do this.

This insight can also shed light on a number of pedagogically fetishes that have recently taken hold in the writing instruction of many schools and curricula. Taken up by some teachers, and some curriculum designers, they are each examples of efforts to ‘add grammar’ (and, therefore, presumably, that holy of holies of contemporary education, ‘rigour’) to the curriculum, usually without much genuine consideration as to what this grammatical feature actually achieves.

Fronted Adverbials

Many students, teachers and, increasingly, parents have been put through the wringer over this in recent years. The terminology sounds fantastic, and is a great way to impress some parents (and managers) with what you’ve been teaching, but I am sceptical that many people involved in the fronted adverbial racket really understand where the loot’s coming from, or where it’s going.

Just so it’s clear, here are a few examples of fronted adverbials – they’ve been bolded for clarity:

in the meadow, they played as if there would be no tomorrow

at noon, the town fell into a heat-induced torpor

because it was raining, Joan decided not to put the washing out

In each case, you can see that the subject of the main clause (‘they’, ‘the town’ and ‘Joan’) which typically comes first, and so is ordinarily the topic, does not, and so is not. Instead, another bit of grammar (in the first two cases prepositional phrases, in the last, a subordinate clause) has been ‘fronted’ – put first – so that it instead forms the main topic of each sentence.

The fronted adverbial ought thus to be understood as a special case of what is called a ‘marked topic’. Linguists use the term ‘marked’ to talk about anything that’s not ordinary – that breaks the bog-standard expectations. A marked topic, then, is a topic which fails to be the other sorts of thing topics typically are: the agent, the subject, and given information. Since, by definition, an adverbial is not the subject of the clause, when placed at the front, it forms a marked topic.

But most English teachers lack this grammatical terminology so, when pressed to explain WHY students – or anyone else for that matter – should use fronted adverbials, they will often reach for that English faculty catch-all, ‘emphasis’. Emphasis is a crude instrument, though, when dealing with the subtle relationship between topic and comment. Marked topics do draw attention to themselves, simply by virtue of being something a little (though not very) unusual. In some contexts, then, ‘emphasis’ works as a partial explanation. But it is very misleading if it is assumed to mean that the rest of the sentence does not have emphasis, or, worse, that the material would lack emphasis if it was put at the end. Consider this pair of sentences:

On Saturday, I saw my best friend

I saw him on Saturday

What are these sentences about, and what are they for? Roughly, the first seems to be about Saturday, while its purpose is to get over the idea of seeing the best friend, but the second is about seeing the friend, and its purpose is to make it clear that this was on Saturday. The first sentence is perhaps an answer to the question, ‘tell me what you did on Saturday?’, while the second is more likely the answer to ‘when did you see him?’.

It’s potentially very confusing, then, to say that fronted adverbials add ‘emphasis’ to the information they carry.  In many cases it is the opposite: they effectively build up suspense, casting their content as a launching point, and so create greater focus on what comes after them. What is more, as example 2 shows, if the reason for our message is the adverbial, we tend to put it at the end, not the beginning.

A similar story can be told about a particular variety of fronted adverbial – the fronted non-finite clause:

Wiping the sweat from his brow, he carefully approached the safe

He carefully approached the safe, wiping the sweat from his brow

Here, the adverbial non-finite clause (wiping…) is more about the character’s feelings and reactions whereas the finite clause (approached…) is about the things that happened (as indeed one might expect given its finite status – see ‘A verb is a doing word’). So, what is the effect of the choice here? Again, fronting the adverbial is not really about putting ‘emphasis’ on it. It is rather that, in having a marked topic, the first is more suspenseful: it drives the action of the story forward by placing the event at the end, and establishing it as the comment of the message. The second option, on the other hand, is less about action, and feels about more about our character, casting the psychological reaction as a comment on the topic of the action. Critically, the question of ’which is ‘better’’ will very much depend on what is going on either side of this sentence, and on what one is trying to achieve.

Passive Voice

There is a mass of confusion regarding this term. I shall not go into huge detail here about what it is or how this confusion happens – for a fuller and far more authoritative explanation, I’d recommend the youtube videos of Geoffrey Pullum, one of the foremost living grammarians of English (this is an entertaining series of informal tutorials explaining both what the passive is, as well as the embarrassing number of writing ‘experts’ who don’t remotely understand it). For the purposes of this essay, though, the key point is the extent to which the whole point of the passive is to shape the syntax of clauses so that the flow of one’s message most effectively follows the expected topic-comment structure. In my experience, even teachers who understand the passive from a structural point of view rarely understand this key pattern.

The effect of the passive is to separate the ‘agent’ role from the subject and topic roles (as explained above, in a typical ‘unmarked’, ‘bog-standard’ clause, the three go together). We’ve already seen an example of this, in these two alternatives:

The dog chased the rabbit

The rabbit was chased by the dog

In both versions, the dog remains the agent – the thing that ‘does’ the action, in this case ‘chasing’. But the second example, by using the extra, helping verb ‘was’, along with the preposition ‘by’, effectively ‘shifts’ the dog to the end of the clause, and brings the rabbit to the beginning. The distinction may seem simple here but, as Pullum illustrates in his videos, identifying the passive in the wild can actually be tricky, and many students find it a challenge. Added to this confusion is deeply unfortunate, and muddled, tradition of seeing the passive voice as somehow inevitably ‘wrong’ – a feature of poor writing to be studiously avoided.

In truth, there is nothing whatsoever wrong with the passive. We use it all the time in speech and writing, and probably the most important use it has is to help manage information – to ensure that the topic-comment structure of our clauses conveys the clearest possible message given what has already happened in the text. Essentially, we use the passive when we want the recipient of the action – whatever the verb is happening to – to be the message’s topic (remember, this means the ‘launching point’ – what the clause it about, rather than what it is for).

So, ‘the rabbit was chased by the dog’ is a passive clause which is about the rabbit, and which is intended to get over the idea that a dog chased it. Conversely, ‘the dog chased the rabbit’ is about the dog, but its purpose is to get over the idea that this dog went and chased a rabbit. Thus, the frequent classroom explanation – that the passive adds ‘emphasis’ to the recipient, is once again misleading. Witness the following examples, all of which ‘push’ the agent into the comment of the clause, signalling its importance as the actual purpose of the clause (the agent is bolded for clarity):

she was helped by a kindly stranger

this was painted by Michelangelo

he was arrested by an elite team of detectives

the meal was prepared by a highly experienced cook

As subject, the recipients of the actions described here receive less focus in terms of the flow of information, and it is significant that in the first three examples the subjects are pronouns (she, this, he) which refer to previously mentioned items (or in the case of ‘this’, to an item which is readily available from context). This pattern feels natural precisely because the passive’s role is to ‘relegate’ the recipient to ‘topic’. I hesitate slightly to use the term ‘relegate‘ –  the topic of course is important in its own way, it’s what the clause is about, after all – but ‘relegate’ does in the end confirm the key point – that in passive structures such as these it is in fact the agent which receives the most attention – it, and not the recipient, is the purpose of the message. Once again, the structure concerned is a complex, two-part relation for which the simple notion of ‘emphasis’ is really rather ill-suited.

The following passive clauses (bolded) serve to further illustrate the point:

A satisfied man came in. The building had finally been built by him

The boys looked guilty. The window had been broken by them

These constructions are not outright ‘ungrammatical’ in the sense that they violate the usual rules of English syntax, but they are certainly awkward. This is because they push at the conventions of typical topic-comment management. In both cases, pronouns – him and them – which refer back to previously mentioned items – a man and the boys – are ‘pushed’ right to the end of the comment ‘slot’ of the second clause, implying that they are new content worthy of comment. But they are not new – they are old information which was well established in the previous clause. In each case, a more natural choice would be to use active voice, ensuring that it is the action, along with its recipient, which receives the ‘comment’ status:

A satisfied man came in. He had finally built the building

The boys looked guilty. They had broken the window.

I should stress here that I consider it very much a moot point as to which students – if any – need to be explicitly taught this terminology. Of course, for older students who are interested, it’s fascinating, and I personally wish more had the chance to learn about it. But I have seen little evidence that it would, for example, have much impact on the general writing skills of the average student. My feeling is that teachers should know what the passive is and what it I for, if only so that they can stop giving advice that it is a bad idea. Probably more important, though, is the realisation that the passive is another example, along with the fronted adverbials, of the perils of positing such a thing as a single, syntactically ‘good’ sentence. The ‘grammar’ of any sentence is inextricably bound up both with both what has gone before and what is soon to come


Cohesion is yet another term which English teaching likes to draw down from linguistics without due regard to either its complexity, or the controversy concerning its precise meaning. For some, cohesion is a pretty broad notion, encompassing almost any feature that helps unify a string of clauses into a single text. Other experts, though, prefer to refer to this notion as ‘coherence’, and reserve ‘cohesion’ to describe only the way that word meaning is used to create connections between various places in a text. Here’s a short text to illustrate this divergence:

Sophie made her garlands of the lovely flowers. She made her dwellings of the shady bowers.

In this case, the pronoun ‘she’ in the second sentence is clearly creating, by virtue of its referential meaning, a link to the word ‘Sophie’ in the first. This is a pretty uncontroversial example of a cohesive tie. But what of the rhyme between ‘flowers’ and ‘bowers’? Or the parallel structure of the two sentences (‘she made her Xs of the Ys’)? These features also serve to draw the text together into a single unified whole, and so are sometimes also referred to as cohesive ties, despite the fact that they have nothing to do with word meaning.

I personally prefer the technical precision of the narrow, word-meaning focussed notion of cohesion, but this example does demonstrate that cohesion is a long way from being the only way in which coherence – ultimately more important in crafting clear writing -is achieved. Unfortunately, though, probably because they can be more readily ‘planned’ (which is to say, embedded into management-pleasing lesson designs) – teachers too often fixate upon cohesive ties – various words and phrases which they believe are crucial in creating coherent texts. All too often, these are taught in an explicit fashion and, depressingly, students are increasingly graded on whether, and indeed how much, they have used these terms. I once sat in a meeting where a student’s text was presented as an example of successful teaching and writing. It was something like the following (the bolded words are the ones we were being invited in particular to celebrate):

Many people reject the celebration of Australia day on the 26th January. Namely, many Australians do not associate the day with careful reflection on their national identity. Moreover, for many indigenous Australians, the day marks what they consider to be the invasion of Australia by the British. For example, it would be better for a date which the whole community can support to be chosen.

Quite obviously this is bad writing, and it is bad because students have been encouraged to write in a bad way. They have been taught some cohesive words and phrases, some of which at least they are insufficiently familiar with from real life examples, they have been taught them in a top-down, ‘explicit’ fashion, and then crudely assessed on their ability to use them in a superficial manner (an assessment style typically constructed chiefly so as to be able to produce numerical data which, entered into a spreadsheet, can stand as proof of the existence of that much desired, but pretty much mythical beast ‘measurable progress’).

Of course, phrases and words of this kind are important ways that cohesion, and therefore coherence, can be achieved. However, as with any linguistic capacity, they are best learnt through an informal process of genuine interaction. We acquire them through reading and hearing them, many times, in many genuine contexts, and then, as we seek to express ourselves, we end up drawing on them to clarify our thoughts and ideas.

What is more, the preoccupation with words of this sort also fails to acknowledge the vital role played by the topic-comment structure in achieving coherence. Consider the following short text, which is replete with cohesive ties:

John and Jim were friendly. A girl called Joanna liked them. Thus, a funny advert for a bag inspired her to buy one for them. A deep embarassment was felt by them therefore.  

Here, adverbs like ‘thus’ and ‘therefore’ and pronouns like ‘them’ and ‘her’ all help to knit together the sentences and create a sense of continuity of subject matter. But while it may be cohesive, this text is far from being fully coherent. This is because the information management swings wildly about: information is thrown at us in a lurching and unpredictable fashion, awkwardly losing touch with the rhythmic ‘wave’ pattern that we expect texts to follow. By stripping out the cohesive ties, and addressing the poor information management, we can create an alternative which, while a little odd and repetitive, is easier to follow and process:

John and Jim were friendly. John and Jim were well liked by a girl called Joanna. Joanna chose them each a new bag which Joanna had been inspired to buy by a funny advert. John and Jim felt terribly embarrassed.

The drive to improve students writing through the drilling of ‘transition markers’ or ‘cohesive ties’ is another unfortunate symptom of the top-down algorithmic approach to writing which is taking hold in too many schools. And, once again, a key problem is the failure to recognise the way that good sentences are good only because of the way that they work together to manage messages and the flow of information across sentences.


The vital topic-comment waves I have been exploring here apply in roughly the same way to both speech and writing, and yet many people obviously find it harder to write as clearly as they tend to be able to speak. One of the main reasons for this is that writing lacks the key tool used to organise ideas in speech: intonation. We learn about how to use this even before we utter words, and for most of us it comes pretty naturally (though intriguing recent research suggests that a full competence in the intonational system of the language is not typically acquired until the late teens). Lacking anything like as powerful an add-on tool as intonation, the written form of the language is in many ways a rather anaemic mode of communication, missing much the expressive power that speech enjoys.

It is true that writing has a recourse which speech lacks: punctuation. Punctuation does have some advantages over intonation, an example being the way that full stops can create clearly demarcated sentences in a way that intonation cannot. But, generally speaking, the nuances of meaning which punctuation affords are pretty limited when compared to the richness of expression than spoken intonation offers. Many people discover this out the hard way, when they are informed that an e-mail they themselves considered to be the very acme of wit was interpreted by their friends or spouse or employees or boss to be creepy, disturbing, offensive, or all three.

It is thus something of a misconception that the typical patterns of syntax in writing are in themselves importantly different from those of speech. There are differences – the chief one being that speech, especially of the informal variety, is characterised by much longer, strung out syntactic structures that modern written English. But the fundamentally distinctive feature of writing is not so much about how its syntax works, as the sheer absence of intonation. Since the chief role of intonation is to support the topic-comment structure, this does indeed have a knock-on effect on syntax, since syntax, along with punctuation, becomes in writing more central to managing the message. But the ability to do this is, as with any language competence, is acquired through genuine, scaffolded interaction – reading being by far the most important interaction of this sort. In short, if someone has not read a great deal, they are bound to find everything much harder, and there’s sadly not much that even the fanciest of explicit instruction techniques can do about that.

Good grammar in writing, in the fullest sense, is thus about far more than ‘good sentences’. Whichever way we choose to assess the grammar of a single sentence – whether in terms of ‘right’ syntax, the fetishistic insertion, or avoidance, of certain structures, or the mere presence of ‘cohesive’ vocabulary – we will always fall far short of what makes good writing. Rather, good grammatical choices are mostly about how clauses and sentences interact with one another, and with the broader context. Those who struggle to write well struggle with a number of issues, but this is typically one of the most important, and it’s usually a result of them not spent enough time reading. The arrangements they choose for each individual sentence are usually syntactically correct, but confuse or bore (or both) when considered in the context of the sentences around them. This is why the idea that one can break writing down into the ‘grammar’ of ‘good sentences’ is pretty implausible. Texts cannot be broken down in the manner of a mathematical operation, in which it is possible to assess a student’s grasp of each step independently of the others. To put it a little too dogmatically, there is no such thing as a good sentence, only a good text.

This reality is somewhat discomforting for those whose vision of teaching is top-down instruction, in which explicitly directed steps pass from conscious, regulated practice to automaticity. What I have described here is very rudimentary, but linguists have in recent years started to chart how information management works in much more detail. I suppose this work could, in theory, be made explicit, and teachable, but the findings as they stand are complex, confusing, and often controversial. What is more, since the success ‘criteria’ for any single sentence cannot be untangled from what comes after, it seems highly unlikely that a simple step-by-step algorithm could be devised to fill those precision-engineered lesson plans so beloved by the administrative classes.

It is also pretty significant that the great majority of mature writers are not consciously aware of these rules. The topic-comment distinction is a relatively recent innovation in linguistic analysis, dating from the mid-nineteenth century at the earliest (though, of course, actually used in language for thousands of years prior to this). It beggars belief that people today need massively complex explicit instruction in this to write English well, whereas Charles Dickens, Jane Austen, Thomas Jefferson and Edward Gibbon did not. How, then, did Dickens et al do it, if no one drilled them on the rules? Well, they read a lot, and they practised writing a lot. They developed a highly attuned implicit capacity to handle information with clarity and style, and they were able to do this because they engaged meaningfully in the human institution of the written language. But these days, a lot of folk really don’t want to hear this sort of thing, because it implies that high-level literacy is about motivation, inspiration, and human connection – and I defy you to measure those as data points in your spreadsheet.

%d bloggers like this: