The mission of this work is to commend a method for describing the valid and invalid procedures of natural sentences. That method is based upon the proposal that there are only three models of the sentence, and only four styles of using them. This proposal contains no prescription. It is a neutral observation about how we make statements. The method uses Traditional Grammar (TG) terms selectively, engaging only those that are useful for describing the unit `sentence', and for analysing instances of it. Exposition of TG terms is conventional, with one notable exception: the term `clause' is eschewed on ground that it is at best a vacuous distinction, and at worst a saboteur in TG's analytical scheme. 

That `clause' is a vacuous term is immediately apparent on this perspective: Two sentences make the same sense. The first has an adjective sequence that TG would call `clause' because it contains the verb-like element `had instigated':

 He talked about the revolution that he had instigated.

The second has an adjective sequence that TG would call `phrase' because it does not contain a `verb': 

He talked about the revolution of his making.

What, other than an accident of word choice, is the difference between these two adjective sequences? Their senses are alike to the point of being inter-changeable, and neither does anything more or less than describe `revolution'. A description of how these sentences make sense gains exactly nothing with the distinction `clause/phrase'. It is therefore pointless and officious to make that distinction. More condemning criticism of TG's phrase/clause distinction is that the words it calls `verbs' for the purpose of that distinction are not verbs at all: What verbial function does `had instigated' perform? That it performs none is the inevitable conclusion, for reason alone that what it does do is take part in an adjective sequence that describes `revolution'. In the light of TG's healthy contemporary practice of classifying words in parts-of-speech categories in accordance with their function in a sentence, it is surprising that it continues to tolerate the possibility of there being a verb in a sequence that does not have a verbial function. Tolerance of the clause/phrase distinction perpetuates the misconception that a complex sentence can contain more than one verbial functor. In practice, the writer of a complex  sentence who allows a second verbial element to be operative in it admits the procedural error that inevitably has him say something nonsensical. (See examples.)

Traditionally, but quite wrongly, the various `clauses' that are not part of TG's `main clause' are its `subordinate clauses'. This hierarchy exists on two defective pieces of conceptualisation. The first defect is in the view that the `subordinate clauses' of complex sentences contain verbs. They do not: They are noun, adjective, adverb or participle sequences – `phrases' is a nice term for them – that attach with descriptive, definitive or particularising intent to some part of the basic sentence. To do that is their function. The verbial function is performed in the basic sentence to which they attach, and only there, because only a basic sentence has a subject and object or complement, so it alone is capable of containing a verbial element. The tenet that there may be a verbial element that is not a verbial functor (one that is not the determinant of the relationship between the subject and its object/complement) makes the TG concept `verb' incomprehensible. The distinction `main clause/subordinate clause' has spawned this obtuse tenet.

The other defect in TG's `main clause/subordinate clause' distinction is in that it does not acknowledge the different procedures of the complex and the joined sentences. Yet they are very different: The complex sentence is one basic sentence (TG calls it `the main clause') embedded by sequences (TG calls them `the subordinate clauses' and phrases) that extend the scope of its statement. Being a sentence, it has a subject and an object or complement. The sequences that embed it do not. For this reason those sequences are functionally and structurally quite unlike a sentence. TG's hierarchical `main clause'/`subordinate clause' ignores the patently obvious fact that there can be no hierarchy of unlike elements.

The conjoined sentence (compound or composite) is a combination of several sentences regulated by compounding operators (conjunctives, disjunctives, etc.) for the purpose of making a statement. There are as many verbial functors in it as there are sentences being joined. The particular framework of logic on which any set of sentences is joined is the one that serves its composer's statement-making intent. No sentence on this framework is more important than another. To call them `main' and `subordinate' is to wildly misrepresent the logical relationship that holds between them. 

Another issue on which this book confronts TG is its traditional claim that a word or phrase that locates the place or the direction of the subject's activity by naming that place or direction is `an adverb of place'. It is not. Rather, is a locative noun, and the word that represents it is a locative pronoun. This is so for the perfectly simple reason that a place or a direction has been named: the syntactic functor that names is necessarily a noun. To call such nouns `adverbs', as TG often does, is to muddy the waters gratuitously. 

No-one needs to be told how to compose a sentence. Composing one is the natural and unavoidable act of reasoning we perform whenever we want to say something. Albeit natural, this reasoning does not inevitably express itself as validly constructed sentences. Like any other sequence of reasoning, it can be invalid. This book provides the criteria for distinguishing the valid from the invalid procedures of sentence construction. The validity in question is not a creature of arbitrary rules. It is the procedural validity that achieves sense by avoiding ambiguity and 'no sense'. Thus, when this book notes of a sentence-part that it is wrongly placed, the noting is done on the basis that procedure for making sense, one sense, and the sense the writer seeks, has been wrongly implemented. Criteria of procedural validity are never stylistic considerations of the `it sounds better' or `ought to be' kind. 

A statement, by nature, has a truth value: it says something that is either true or false. (The statement `John arrived yesterday' is true if and only if John did arrive yesterday.) But, as every experienced reader knows, not all written or spoken sense is made as statement. Naturally then, not all sequences of words that make sense are sentences in the `statement' mode. Writers have made sense – constructed portraits, represented states of mind, tipped perspectives, etc. – outside the procedures of the statement-mode `sentence'. Such `literary' senses are personalist expressions to which neither truth conditions nor precepts of procedural validity can apply. It is, for instance, not useful to ask of: 

Altarwise by owl-light in the half-way house
The gentleman lay graveward with his furies;
Dylan Thomas 

whether it is true or false, nor whether its procedure is valid. Rather, we make sense of it in the private ways that our linguistic and aesthetic sensibilities render us capable. Concerned with the procedural validity that yields a true or false statement, this book does not examine `sense' on this literary level. 

The modes `question', `order' and `exclamation' are related to the statement mode in that they raise a subject for a purpose. That purpose is not to say something about it, as the statement does, but to enquire or order or exclaim on the basis of it. These modes are not subject to truth conditions for reason alone that they are these modes: `John is here!' can be either true or false if it is used to exclaim that John is here. But the identical expression can also deride the suggestion that John is here: `John is here!' Derisive remark, like expostulation, is one-dimensional: by nature: it does not have `true' and `false' facets. These modes have an a-syntactic element that is quite unlike the statement's. For instance, `Door!' shouted at someone who has just walked through a door and failed to shut it raises the subject 'your shutting of the door' every bit as much as `Shut the door, please!' or `Would you shut the door?' raise it. That the latter is constructed and marked as a question, and the former as an exclamation, is immaterial: `subject raised and order issued' is resplendent in both constructions. 

No-one could mistake `Would you shut the door?' for a question if it is uttered in the context described above. But, in the same context, nobody would mistake `Would you swim in a crocodile-infested river?' for an order or a request. It is clearly an enquiry about the subject 'your swimming in a crocodile-infested river'. And both `Swim in a crocodile-infested river!' and `Swim in a crocodile-infested river?' are clearly exclamations (of disbelief, horror, protest, etc.), despite the question mark that punctuates one of them. Why one subject – 'your shutting of the door' – does not yield a question and another subject – 'your swimming in this crocodile-infested river' – does is something to do with the nature of the subject and the situation in which it is raised, and nothing much to do with syntax (procedural validity). Investigation of this interesting issue is beyond the scope of this book. 

Sophie Johnson
at English Grammar Tutor

Return to The Well Bred Sentence Table of Contents