already, or, at any rate, has advanced half way towards knowing it-a result which is a special help in mastering irregularities (p. 119). Thus in the grammar to First Steps I give under the phonology, among other examples of the vowels, twa handa, two hands,' twegen fēt, 'two feet,' twēgen menn, 'two men,' so that when the learner comes to the numerals, he finds that the paradigm

twegen twä twä

twam twēgra

offers hardly any new difficulties; for he finds the above examples repeated with a reference to the place where they occur, but without any translation, together with mid twam handum as an example of the dative, whose ending -m is already familiar to him, from the nouns and adjectives. The only remaining form twēgra is sufficiently illustrated by the parallel genitives préora and pritigra, of which I proceed at once to give examples in sentences. The form twegra is only added to complete the paradigm, as it does not occur in the texts in First Steps, for which reason no special example is given of it.

It will be seen that after the learner has gone through such a book as First Steps, in which the grammar is kept strictly within the limits of the recognition-stage, a great part of his grammatical knowledge will be unconscious instead of analytic and systematic. Thus he will know a good many individual forms of strong verbs, but will know nothing of the distinctions of class. Thus he may know that brecan has preterite brac and preterite participle brocen, but he has not learnt to refer it to the bear-class, although he may have noticed the parallelism between bræc, brocen and bær, boren, and may have strengthened this association by remembering the further parallel stelan, stal, stolen. In this way he will be well prepared for the classification of the strong verbs. A few weeks' work at the Anglo-Saxon Primer, which is constructed on the rigorous grammar-andglossary historical method-though otherwise made as simple and easy as possible-will then, as I have said in the preface to First Steps, enable him to 'systematize his knowledge and round it off, and he will proceed to the elements of historical and comparative grammar with all the more zest through not having had them crammed into him prematurely.'

Evils of the Separation of Syntax from

Accidence

The evils of the separation of syntax from accidence are well shown in the way in which the dead languages are taught in schools. Boys are made to learn paradigms by heart, and are then set to read the classical authors with the help of a dictionary before they have acquired any real knowledge of the meanings of the inflections they are expected to recognize in their texts-much as if they were taught the names of tools without being taught their uses. Thus in learning Greek they are taught to recognize the optative mood entirely by its form without having any idea of its meanings and functions as distinguished from those of the subjunctive, of which, indeed, they come to regard it as an arbitrary and unmeaning variation; to which may be added that their ideas about the meaning and function of the subjunctive mood itself are vague enough. When they are afterwards made to learn the rules of syntax, they are unable to apply these rules to what they are reading, and in most cases the possibility of doing so never enters their minds: they prefer to go on as before, and to guess at the meaning from the context without paying any regard to the moods. It is not very long ago that the rules of Greek syntax were learnt in Latin-an effectual bar to any intelligent application of them.

Examples

It is now generally admitted that a grammatical rule without an example is of no practical use: it is an abstraction which is incapable of entering into any direct associations with anything in the language itself. The example, on the other hand, is concrete: it can be imprinted firmly on the memory by the mere force of the mechanical associations involved in carefully reading it and carefully pronouncing it aloud; while, on the other hand, it is logically associated with the rule, which it explains, illustrates, and justifies. The example serves also as a standard or pattern by which the learner can recognize other examples of the rule as they occur in his reading. The example is thus a link between these other examples and the rule itself.

Many of the older grammarians, while expending much

thought and care on elaborating their statement of the rules, considered the choice of examples as of subordinate importance. They forgot that the first object of grammatical study is not the acquisition of rules, but of a practical command of the language itself; so that instead of the examples being intended solely to illustrate the rules, the true relation is almost the reverse: the rules are mere stepping-stones to the understanding of the examples; so when the latter are once thoroughly understood, the rules become superfluous and may be forgotten.

These considerations have led some reformers to advocate putting the example before the rule, the idea being that the learner is thereby led to study the example carefully and then deduce the rule for himself, and finally compare his deduction with the rule as formulated in the grammar. This is the old inventional fallacy (p. 116) over again. Experience shows that when the learner knows that the work of deducing the rule from the examples has been already done for him, he naturally declines to do it again, so that, if the rule is put after the example, he simply reads the rule first, and then returns to the example. If, however, he prefers to read the example first, there is nothing to prevent him from doing so, whether it precedes the rule or not. Most learners prefer to read the rule first in order to know what the example is about, and what to look for in it-for a sentence may be, in itself, an example of a dozen rules of grammar-and if they do not understand the rule, they then read the example and return to the rule again, and when they finally understand the rule, they concentrate their attention on the example. We may say, in short, that the order of rule and example is of no importance compared with their mutual relations.

The number of examples depends partly on the nature of the rule, partly on the scope and size of the grammar.

Some rules hardly require any example at all through being practically of universal application, or self-evident, or because they are of no intrinsic importance, and are added only for the sake of completeness. But it is a safe principle never to take for granted that a rule does not require an example: if adding a few words in parentheses will make the statement or rule any easier to grasp, or prevent some misunderstanding that the

writer never thought of, they certainly ought to be added. If they are superfluous, no harm is done. Besides, what is superfluous to one reader may be helpful or even necessary to another. German writers often exasperate the reader by giving half a page of examples of some pet truism that requires only two words to illustrate and prove it, and then make a series of abstract generalizations expressed in unfamiliar and arbitrarily defined terminology without any help in the way of example, so that they often become unintelligible even to their own countrymen.

If every rule is to have an example, it follows that a compound rule ought to have an example of each division of the rule. Thus, such a rule as that 'verbs expressing joy, desire, memory govern the genitive' requires at least three examples. But in such a case as this many short grammars would give only one, on the mechanical principle that each paragraph is to have only one example. Even in the shortest grammar space may generally be found for a full number of examples by omitting some of the irrelevant matter of which such illplanned books are generally full.

If there is not room for more than one example to those rules which really seem to require it, additional examples to those rules that most require it may be given in a separate book.

A good example must fulfil two conditions: (1) It must illustrate and confirm the rule unambiguously. Thus, as already remarked (p. 107), examples of the use of the ablative in Latin should, if possible, be forms which cannot be taken for datives. (2) The example must be intelligible as it stands, without any further context. If the example is a sentence or is contained in a sentence, the sentence should be one which will bear isolation from the context. In dealing with separate words, it is often a great help to the learner to give them in natural groups such as hands and feet, buy and sell, past, present, and future, dead or alive, neither here nor there. The more concrete a word is, the better it will bear isolation. It is mere waste of space to give bare lists of prepositions, conjunctions, and other form-words in an elementary practical grammar.

It need scarcely be said that the examples must be in the language with which the grammar deals. Thus no one would

think of illustrating a rule of Spanish syntax by a Portuguese example. But it is almost as great an absurdity to illustrate rules of modern English syntax by examples taken from Shakespeare, except in special cases where the earlier constructions have been imitated by modern writers. All of this would, however, be quite out of place in a practical grammar for beginners.

Carrying this principle a little further, we must be careful that our examples in an elementary grammar do not contain any specially difficult or rare words or irregularities of construction which do not directly illustrate the rule.

Examples made up extempore for the purpose of illustrating a rule are not so good as those which have been collected from a variety of writings. There is, first of all, the danger of monotonous repetition of words, ideas, and constructions. In the effort to frame collocations of words to illustrate some rule, the grammarian is apt to produce unnatural, trivial, or otherwise objectionable sentences, such as the golden sun shines brightly the happy children of our teacher sing sweetly enough from their book of hymns, both taken from an English grammar of some repute in its time.

Every example ought to be explained-even in the phonology. The translation of a new word not only gives a useful piece of information, but serves also to identify the word.

But the explanation need not necessarily take the form of translation. There is one objection to translating the examples in a grammar: the learner is tempted to read them carelessly, and so not get all the benefit that would result from a conscientious analysis of them. In my First Steps in Anglo-Saxon I have therefore tried the experiment of putting the explanation of the examples in the grammar on the same footing as the words in the texts, as far as possible, so as to oblige the learner to read the examples with the same care as the texts themselves. At the beginning of the grammar each example is translated in full. When a word or word-group or sentence is repeated in the grammar, it is not translated, but the learner is referred back to the place where it is translated; and after the first few pages each new word in the examples is explained in the notes at the end of the book. Hence the reader is obliged to study each example carefully, and with constant comparison of what