In Table I, as in those which follow, the ordinary figures show the actual observed number of reactions; the italic figures show the number, computed to the nearest integer, which we should expect to find if chance only were at work. This last number is obtained as follows: if we have N objects of which n1 belong to class p1, na to class p2, na to class p, etc. (so that Σn = N) and of which m1 also belong to class 91, m2 to class q2, mg to class 93 etc. (so that Σm = N), then by the ordinary theory of probability we should expect the number belonging to both class p and q, to be nx X My Thus 287 reactions out of 1658 fall in class I and 301 out of 1658 in 287 × 301 1658 class G; we should therefore expect to find that = 52, very nearly, of the members of class I were also members of class TG. It will be noticed that although in many cases the agreement between the actual numbers and the 'probable' numbers is very close, there are others in which there is a marked difference; these are the cases in which the connection between affective tone and reaction form shows itself. In view of the evidence which I brought forward in the preceding paper, I regard it as incontestable that the affective classes G and TG chiefly contain positively toned words, classes O and T mainly indifferent words and classes R, GR, TR and TGR mainly negatively toned words. I do not consider, however, that it is practicable to discriminate further than this at present, or to avail ourselves of the quantitative differences which I gave reason for supposing to exist between the classes which make up these three main groups. I therefore simplify Table I by classifying the reactions, with regard to their affective tone, into 'positively toned,' 'neutral' and 'negatively toned.' The result is shown in Table II. The behaviour of the various classes can be more clearly seen here than in the original table. I regard the indications afforded by this table as reliable; in most cases we have a good number of reactions in a class and it must be remembered that the crudity and liability to fortuitous interference which made the memory test, used in the preceding paper, so insensitive a criterion, so to speak, of the quality of affective tone, no longer apply here. Once we have determined the qualitative properties of the different indicator classes we can say with J. of Psych. (Med, Sect.) I 20 considerable assurance that the reactions belonging to them possess those affective properties. Class I (Co-ordination) shows a slight but distinct tendency towards toned as opposed to neutral reactions; the actual figures (98 and 79) for both positively and negatively toned reactions are greater than those indicated by probability (91 and 71 respectively), while the actual figure for neutral reactions is well below the probable figure (110 to 124). Class II (Predicates) is worth considering in some detail especially in view of the comments I made about it above. Sub-class (a), consisting of 'simple' predicates, shows a slight tendency to favour neutral (105 'actual' to 99 'probable'), at the expense of negatively toned reactions (52 'actual' to 57 'probable'). II (b)—predicates implying personal opinions or judgments of valuehas a marked excess of negatively toned and a marked deficiency of neutral reactions (33 'actual' to 19 'probable' and 16 'actual' to 32 'probable' respectively). II (c) subject relationship-conforms exactly to the probable values. II (d)-object relationship-like II (a), somewhat favours the neutral reactions at the expense of the negatively toned. II (e)-definition of time, place, means etc.-is a very small class and its deviations from the probable values appear to me to be insignifi cant. In fact II (b) shows a characteristic tendency not found in any other form of predicate reaction. It should be regarded, in my opinion, as essentially an 'inner' association, to which general type its affective properties conform, while II (a) is psychologically indistinguishable from the emphatically 'outer' association of co-existence. II (c) and (d) are less obviously 'outer' but in the majority of cases they conform much more nearly to this group than to 'inner' associations. On the whole I consider that they ought to be classed as 'outer.' II (e) I think should be retained in the 'inner' group. To insist on such widely differing types of reaction as II (a) and II (b) being kept in the same class simply because they are both grammatical predicates is, surely, mere pedantry. Class III (Causal dependence) is again rather small; its tendency is to favour the positively toned reactions at the expense of the neutral. Class IV (Co-existence) is the first of the indisputably 'outer' types. It is a large class and shows an unmistakable tendency towards neutral reactions at the expense of the negatively toned. Class V (Paraphrases, synonyms, etc.) shows a slight and probably negligible tendency in favour of negatively toned reactions at the expense of the other two. Class VI (a) (Verbal reactions depending on common phrases, etc.) is again large and shows a very marked tendency towards neutral reactions, mainly achieved at the expense of the negatively toned. Classes VI (b) and (c) are exiguous and their divergences from 'probable' values are small. They should probably be included in class VI (a). In Class VII (Indirect reactions) the tendency is unmistakable1; there is a great preponderance of negatively toned reactions at the expense of both the positively toned and the neutral, especially the latter. Class VIII (Freaks') is very small, but I think that the marked excess of negatively toned reactions (16 'actual' to 8 'probable') is almost certainly significant. We may now simplify the classification still further and compare the whole of the inner associations with the outer. It is clear that inner associations contain a marked preponderance of positively and negatively toned reactions and a marked lack of neutral reactions; outer associations favour the neutral reactions chiefly at the expense of the negatively toned. As I have already observed, I consider that class VII should be included among inner associations. I have kept it distinct up to this point, partly because its type of association is, by definition, somewhat obscure and partly because I wanted to show the tendencies of inner associations without there being any question of their being unduly influenced by the inclusion of reaction forms which might appear of dubious eligibility. When class VII is thus included the figures become: 1 In this class there are 23 reactions actually observed in the 'indicator' class TR; the 'probable' number is 6. The probability of this discrepancy being due to chance is about 2.3 x 10-7. Inner associations show the same characteristics as before but more markedly with regard to negatively toned reaction and less so with regard to positively toned. Finally, I shall assume that my contentions as regards Predicate forms are warranted and shall transfer classes II (a), II (c) and II (d) to the outer associations. The figures then become: This again greatly increases the relative predominance of negatively toned reactions among the inner associations; it slightly reduces the relative differences between actual and probable figures for outer associations of all three kinds-they are, in fact, slightly diluted by the addition of a number of reactions distributed in close accordance with probability. It may be convenient to keep these predicate classes II (a), II (c) and II (d) with the other predicates for certain purposes, but I think there can be no doubt that if we are considering reactions from the affective point of view, their proper place is with the outer associations. And after all it is the affective tone which we are seeking in all practical applications of association methods; reactions unaccompanied by it are of no great value, they do not lead to significant complexes of pathological importance or even to constellations of theoretical interest. 1 The chance of this difference between the actual and probable figures being accidental is about one in two million. The affectively toned reactions are the important reactions, especiallyfor clinical work-those which are negatively toned. If therefore we are desirous of 'summing up' a subject in the way which is sometimes attempted by study of the 'reaction-type,' it is important that we should adopt the system of classification which will most clearly show the relative number of reactions constellated by 'complexes'-i.e. which are negatively toned-and that we should know which classes are likely to contain the greatest proportion of such reactions. The best way to do this would be to use all three complex indicators, viz.: reaction time, psycho-galvanic reflex and the reproduction test. A complete analysis into the 'indicator classes' can then be made. But it may well be that external circumstances may not permit of the application of all, or indeed of any, of these tests. In such a case we have only the form of the reactions to fall back on and I think it is clear that the relative proportion of complex-determined reactions will be much more clearly shown if we adopt the system of classification which I have here advocated (viz. separation of predicate forms into 'outer' and 'inner' and the inclusion of all the very indirect and 'personally' constellated reactions of my class VII-under the head of 'inner associations'), than if we adhere to the scheme used by Jung. The proportion of inner associations to outer will then afford some measure of the subject's 'complexity'--if I may coin a word to denote possession of complexes. My figures show that for normal subjects the proportion of inner to outer associations is almost precisely 1 to 2 and any proportion much greater than this is likely to mean a correspondingly large number of negatively-toned, complex-determined reactions and therefore to be significant. The most important classes from this point of view are II (b) and VII. Class I is somewhat significant, although much less so, and the figures for class VIII show that 'freaks' are very noteworthy. I do not claim that this method is anything but very rough, only that it is likely to be less misleading than existing methods. At the risk of prolixity and repetition I wish to emphasise the point of view indicated in the preceding paragraphs. My contention is that no system of classifying reactions can be of any value unless it is based on the nature of the psychical processes which determine those reactions rather than on the verbal or grammatical form which they may take. The different forms are only of interest in so far as they can be correlated with significant psychical conditions of one kind or another; apart from this they are merely academic and sterile. |