There is another way of reporting a character's speech, or, still more commonly, his thoughts, which is especially common in 20th century authors, but which may occasionally be found in much earlier writers. This is neither direct speech, which reproduces the speaker's exact words, as they were uttered, in quotation marks; nor is it indirect speech, which retells the character's words from the author's point of view, and is characterised by such formulas as, He said that... The third way of reporting a character's speech or his thoughts stands apart from those two. It is not direct speech, as it does not reproduce the speaker's words in their original form, and it is not indirect speech, as it does not introduce them by formulas like He said that..., though the changes in the personal pronouns, etc. are made.

In characterising indirect speech as compared with direct, we must dwell on two special cases in which a distinction found in direct speech gets lost in a change into indirect speech.

The term "sequence of tenses", though widely used both in practical language teaching and in theoretical investigation, has still not received a clear and commonly accepted interpretation which might be used as a firm basis for further research in this field.

It would be vain to expect that every sentence we can meet with in a text is bound to be either syndetic or asyndetic, either compound or complex, etc. Several or indeed all of these characteristics may be found in a sentence at the same time. It may, for instance, consist of several clauses, some of them connected with each other syndetically, i. e. by conjunctions or connective words, while others are connected asyndetically, i. e. without any such words; it is also possible that some of the clauses are co-ordinated with each other, so that a certain part of the whole sentence is compound, while others are subordinate, so that another part of the whole sentence is complex, etc.

By an inserted clause we mean a clause appearing within another clause and interrupting its structure. A clause of this kind may either be asyndetic, or it may be introduced by a conjunction, most usually perhaps by the conjunction for. An inserted clause usually contains some information serving to elucidate what is said in the main body of the sentence, or it may be a casual interruption due to the speaker suddenly thinking of something vaguely connected with what he is talking about, etc. There is certainly no reason to term an inserted clause subordinate, since no signs of subordination are to be found. Neither is there any valid reason for saying it is co-ordinate in the sense that clauses are co-ordinate within a compound sentence. Indeed there are no clear signs which would prove that a sentence with an inserted clause is a composite sentence at all — though this of course depends on the exact interpretation we give of the notion of "composite sentence". The question whether a sentence with an inserted clause should or should not be considered a composite sentence is, after all, of little theoretical interest, and we here content ourselves with stating that we will not take it as composite. The sentence with the inserted clause taken out of it is a simple sentence (unless of course it contains coordinate or subordinate clauses) and with the inserted clause it may be reckoned as a special type — a simple sentence with an inserted clause.

As has been pointed out above, by asyndetic we mean composite sentences whose constituent clauses are not joined together either by a conjunction or by any kind of conjunctive word (relative pronoun or relative adverb: who, which, that, when, where, etc.). This does not mean that there is nothing at all at the beginning of the second clause to express some kind of relation between the clauses. The second (or third, etc.) clause may begin with some word that does indicate some relation of this kind: for instance, if it begins with the adverb however,, this indicates a semantic relation between the clauses akin to that expressed by the co-ordinating conjunction but; if it begins by the adverb instead this, too, indicates some sort of relation. Something similar may be said about such words as nevertheless, nonetheless, therefore, notwithstanding, still, etc. It may also be said that a personal or demonstrative pronoun referring back to some person or thing mentioned in the first clause of an asyndetic composite sentence (the so-called anaphoric use) performs some connecting function. Yet we will consider sentences built in this way to be asyndetic, as they do not contain any grammatical link between the clauses.

Of course a complex sentence does not always consist of one main and one subordinate clause. It may contain two, three, or more subordinate clauses, which may or may not be connected with one another. Two subordinate clauses are connected with one another if they belong to the same type (for example, if they are both object clauses). In that case they may be joined by a co-ordinating conjunction (and, but, or). We can see an example in the following sentence: He thought what a handsome pair they made, Lucien and his mother, and how the abbe had no chance at all with them. (R. MACAULAY)

When we set out to characterise the syntactic function of some word, phrase, or clause, we are often at a loss for want of objective criteria which would justify this or that view of their function, and we are often reduced to subjective opinions, often incompatible with one another, instead of reaching conclusions binding on every one. The result is that the whole sphere to which such a question belongs, begins to look like one inaccessible to scientific treatment, and we either reconcile ourselves to this state of things, or else we decide that the question had better be dropped altogether.

From a purely descriptive viewpoint, we can establish a category of clauses beginning with that. An analysis of the surrounding elements (the context) is needed to find out whether that in a given case is a demonstrative pronoun, a relative pronoun, or a conjunction, and what kind of clause is introduced by it. We will not at present dwell on the question how we have found out that the word that opens a clause (that would necessitate some additional investigation which does not belong here); we will consider it as settled that it does come at the beginning of a new clause, and we will limit ourselves to the study of the questions indicated above.