Thefreudianslip.pdf

See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/234674125

The Freudian Slip Revisited: A Case of Mistaken Identity in "Finnegan's Wake." Article  in  Language & Communication · October 1995 DOI: 10.1016/0271-5309(95)00013-5

CITATIONS

READS

4

63

1 author: Barbara MacMahon University of Derby 18 PUBLICATIONS   49 CITATIONS    SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Language and Meaning in Literature View project

All content following this page was uploaded by Barbara MacMahon on 26 November 2018. The user has requested enhancement of the downloaded file.

Language & Communication, Vol. 15, No. 4, pp. 289-328, 1995 Copyright © 1995 Elsevier Science Ltd Printed in Great Britain. All rights reserved 0271-5309/95 $9.50 + 0.00

Pergamon

0271-5309(95)00013-5 T H E FREUDIAN SLIP R E V I S I T E D : A C A S E O F M I S T A K E N I D E N T I T Y IN F I N N E G A N S WAKE

BARBARA MacMAHON Introduction In areas of psychoanalytic literary theory, claims are made that unconscious mechanisms are at work both in slips of the tongue and in certain literary word innovations. In this paper I use psycholinguistic models to examine these claims. Recent literary theorists of post-structuralist and deconstructionist orientation have adopted Freudian and Lacanian theory in new approaches to literature. As a legacy of the disillusion with logical positivism, these theorists are concerned to dissociate themselves from empirical methodologies. I aim to show that in refusing to question the scientificity of psychoanalytic work, they paradoxically repeat the ideology of the 'scientific' psychologists who preceded them. Freudian theory purports to scientificity. Interesting hypotheses are raised which have interesting implications for the study of language and literature. These points are obscured, for the linguist, by the unscientific nature of the development of these hypotheses. Those who value Freud nowadays see his ideas as lending themselves to post-structuralism, exempt from the testing procedures of science. Those who reject him ignore potentially interesting areas of research concerning language, literature and the unconscious. This mutual intolerance is particularly counter-productive with regard to the c o m m o n ground between psychoanalysis and psycholinguistics--the slip of the tongue. The slip is frequently referred to in psychoanalysis and psychoanalytic literary theory, and is the object of extensive research into language production models in psycholinguistics. Ignoring the existence of this shared concern results in an impoverishment for both linguistics and literary theory. In the following, I begin by briefly outlining relevant concepts and arguments from psychoanalysis, and giving an example of a counter-argument on the slip of the tongue. I go on to describe psycholinguistic accounts of speech errors, and then show how these accounts can enhance a detailed comparison of three samples of literary and non-literary word substitutions which sheds light on the claims being made in literary theory.

Psychoanalysis and literary theory Freud and the slip of the tongue In Freudian psychoanalysis the unconscious is said to be created in a developmental progression from a pleasure principle to a reality principle. Briefly, Freud states that in order to work--the Correspondence relating to this paper should be addressed to Barbara MacMahon, Nene College, Moulton Park, Northampton NN2 7AL, U.K. 289

290

BARBARA MacMAHON

prerequisite for adult survival---every human being is forced to repress the pleasure drive in favour of the reality principle (Freud, 1915-17, pp. 401--403). This repression takes place in stages during infancy, and is facilitated by the trust that it will ultimately be of benefit to the individual. The journey to repression involves oral, anal and phallic stages--stages of perversity in which the infant takes pleasure in exploring and experimenting with parts of its own body. In order for the child to assume a gendered and otherwise socialized identity, s/he must finally go through the Oedipal stage (Freud, 1915-17, pp. 373-382). Firstly the child recognizes itself and its mother as separate bodies, so assuming an autonomous and subjective identity. Polymorphous perversity, the uninhibited taking of pleasure in all parts of the body, gives way to the desire of another being (the mother). Now the pleasure drive is no longer sustainable within the infant's own body, though it is still 'perverse' in that desire is unharnessed by normal societal restrictions, in particular the prohibition of incest. Gender is assumed through the recognition of genital difference. Finally, a triangular relationship between the child and both parents results in a rivalry between the child and the same sex parent for the affections of the parent of the opposite sex. For a boy child, this means that he perceives his father as a rival for his mother's affections. The threat of castration (the boy is said to perceive women and girls as castrated) forces the boy to give up this rivalry and identify with the father, submitting to the prohibition of incest (and in turn to all other societal prohibitions) in the belief that this repression of his desire for his mother will ultimately be to his advantage; that is, one day he will be grown-up and in a position of power, like his father. (It is not within the scope of this paper to enter into the debate on the obvious problems with the simple reversal of this account for a girl child.) Once the child has come through the Oedipal stage, it is a responsible subject, capable of working for survival, and governed by the reality principle in which gratification is deferred. The infant accepts authority and forms a conscience. S/he has repressed both 'auto-erotic' (self-sustaining) and desirous (parent directed) drives, and in doing so has opened up what Freud called the unconscious. The subject is now split between conscious and unconscious, and will remain so, the unconscious being the receptacle for all subsequent repressions. The original repressed desire for the mother (or father) remains there, and is joined by later wishes inadmissible to consciousness, these latter, according to Freud, usually being sexual or aggressive, like the original Oedipal feelings towards each parent. The individual thus becomes self-censoring. Censorship then, is the barrier separating the conscious and unconscious mind. In most situations censorship operates successfully in preventing unconscious wishes from manifesting themselves in the individual's behaviour, but repressed elements are always waiting for an opportunity to break through this barrier. They do so, according to Freud, in pathological conditions such as psychosis and schizophrenia, as well as in normal symptoms such as dreams and parapraxes--slips of the tongue, bungled actions, etc. These are the subject of Freud's The Psychopathology of Everyday Life (Freud, 1901). It is important to note that in such symptoms the repressed is considered to break through the barrier of censorship and erupt violently into consciousness. This is relevant to Lacanian theories of creativity which I discuss later, in which the repressed is said to re-emerge disruptively in literary language. Freud claims a similarity between the ways in which the repressed re-emerges in different types of symptom, concluding that the unconscious has certain characteristic modes of operation. These modes are also referred to as the primary process, and the two most important types of operation are condensation and displacement. Descriptions and examples of condensation and displacement in Freud's work are numerous, and sometimes conflicting, but the concepts are

THE FREUD1AN SLIP REVISITED

291

immediately comprehensible in Freud's account of one particular type of speech error--the word substitution or malapropism. Here, in the Freudian account, the primary process operates on formal similarities between words (condenses the words), and, on the basis of these similarities, displaces from one word to another, so allowing for the re-emergence of the repressed. The following is a simple example from Freud's work (Freud, 1901, pp. 92-93); I quote from Timpanaro, who summarizes the example: A Jewish convert to Christianity, who found himself the guest of an anti-semitic family with which he nonetheless wanted to keep on good terms, and who feared that his children would reveal the unpalatable truth of their Jewish origin to his hostess who was ignorant of it meant to say to them: 'Go into the garden children', but instead of Jungen('youngsters') says Juden ('Jews')--the very word that he least of all wanted to utter (Timpanaro, 1976,

p. 104). Here then, according to Freud, the repressed element--the fact of Jewish origin--waits for the appropriate opportunity to manifest itself, and finds this opportunity in the sound similarities between 'Jungen' and 'Juden'.

Lacan and psychoanalytic literary theorists In a re-reading/re-writing of Freudian psychoanalysis influenced by Saussurean and Jakobsonian linguistics, Lacan equates transition through the Oedipal stage with the acquisition of or 'entry into' language (Lacan, 1966). Language itself, for Lacanians, is the authoritarian rule-governed system which represses and controls the infant through its naming and imposing of arbitrary distinctions on the world. Without language, the infant is unable to form a sense of self or to distinguish between itself and other beings/objects, yet this system which allows for the construction of identity also restricts and moulds that identity in a predetermined fashion. Language, for Lacan, is to be equated with Freud's law enforcing father figure. Puns and slips of the tongue disrupt what Lacan refers to as the Symbolic--the authoritarian communicative level of language, using the operations of condensation and displacement, which paradoxically are the characteristic operations of language itself (Lacan equates Jakobson's concepts of metaphoric and metonymic poles (Jakobson, 1956) with Freud's condensation and displacement). Lacan provides his own puns as illustrations of his theory, for example referring repeatedly to language as 'Le Nom-de-Pere' and 'Le Non-du-Pere'. The unconscious and language are thus inextricably bound up with one another in Lacanian theory. In a variation on the same argument, Kristeva argues that puns, slips and literary modernist language innovation and experimentation have revolutionary potential in disrupting the Symbolic order (Kristeva, 1974). She distinguishes between Symbolic and semiotic levels of language. The semiotic comprises pleasurable linguistic play associated with auto-eroticism in the babbling phase of early childhood, dominated by the production of sound patterning and rhythmic sequences. With the gradual acquisition of language this instinctive semiotic level is repressed as the child is forced to use language communicatively. Slips of the tongue and forms of wordplay then constitute challenges to the Symbolic in reverting to a level which ignores semantic and syntactic divisions and favours the use of sound similarities between words as an organizing principle. The roots of this idea can also be found in parts of Freud's work (e.g. Freud, 1905, p. 168). Literary theorists have been quick to apply the theories of Lacan and Kristeva to forms of literary language which are innovative and experimental. Their argument is that these forms of writing operate in the same way as the Freudian slip or Lacanian pun, and therefore that they are pleasurable yet unsettling challenges to subjectivity and the production of meaning. The following are quotes from two such theorists who refer specifically to Finnegans Wake:

292

BARBARA MacMAHON

The Wake is a continuous lapsus. It declares itself to be the 'lapse not leashed' and it is the mechanism of the lapsus which allows the work to progress (MacCabe, 1979. p. 143). [the pun in Finnegans Wake] is a characteristic mode of the dream, the witticism, the slip of the tongue: those irruptions of the disorderly world of childhood pleasures and unconscious desires into the clear, linear processes of practical and rational thought (Attridge, 1988, p. 190).

Having given a basic outline of the Freudian account of the slip of the tongue, and of the development and use of this account in psychoanalysis and literary theory, I now go on to provide an example of an objection to the Freudian position.

Timpanaro on the Freudian slip Timpanaro's work on the Freudian slip (Timpanaro, 1976) is concerned with the status of Freudian psychoanalysis as a science, and with its discernible ideology. The Freudian Slip is a critique of the Freudian account of speech errors from an empirical linguistic perspective informed by Timpanaro's experience as a philologist and textual critic. What distinguishes his work from other linguistic accounts of speech errors (apart from the difference in depth of linguistic analysis), is its political orientation. While psycholinguistic accounts of speech errors would seem to offer important implications in contradiction to Freudian theory, these implications are rarely overtly stated, researchers preferring instead to restrict the function of their work to the pursuit of neurological process and language production models. Timpanaro's is a rare combination of politics, psychoanalysis and linguistics. Timpanaro identifies and discusses seven types of slip (Timpanaro, 1976, pp. 135-150), claiming that they are common among copyists and are all attributable to inevitable and insignificant disturbances of neurological processes. This is clearly in contradiction to the Freudian theory of slips as eruptions of repressed wishes. In the following I give two of Timpanaro's examples to illustrate his arguments.

Banalization In the first example Timpanaro rejects Freud's explanation of a slip for three reasons--firstly because it is an insignificant case of banalization, in which case there is no need to appeal to psychoanalysis; secondly because Freud's explanation depends on a non-falsifiable method of analysis (free association); and thirdly because Freud's intervention in the analysis guides it in a certain direction. The slip taken from Freud's work (Timpanaro, 1976, pp. 29--48) concerns the incorrect quotation of a line from V-trgil's Aeneid by a young Austrian Jew in conversation with Freud about the inferior position of Jews in Austria-Hungary. The correct quotation from Virgil is: Exoriare aliquis nostris ex ossibus ultor (Aeneid, IV 625). Let someone arise from my bones as an Avenger (Timpanaro, 1976, p. 29).

The incorrect quotation analysed by Freud is: Exoriare ex nostris ossibus ultor i.e. he omits aliquis and inverts the words nostris ex (Timpanaro, 1976, p. 30).

Timpano suggests that the forgetting of aliquis is a simple case of banalization, which he describes as follows: Forms which have a more archaic, more high-flown, more unusual stylistic expression, and which are therefore more removed from the cultural-linguistic heritage of the person transcribing or reciting, tend to be replaced by forms in more common use (Timpanaro, 1976, p. 30).

THE FREUDIAN SLIP REVISITED

293

In this particular case, the intended construction 'exoriare eliquis . . . u l t o r . . , is highly anomalous' (Timpanaro, 1976, p. 32) and is assimilated by a process of banalization to the speaker's 'own linguistic sensibility' (Timpanaro, 1976, p. 33). In other words, the construction is altered to match a construction common in German, and therefore more familiar to the Germanspeaking Austrian Jew. This is the essence of Timpanaro's argument; banalization involves the substitution of the unfamiliar with the familiar. His explanation of banalization is based on a principle of mental economy (Timpanaro, 1976, pp. 92-95). Given that the human memory is not capable of holding an infinite amount of information, there must be some means of storing the most useful information as a priority, otherwise the memory would become overloaded with superfluities. What is unusual or rare is hence more difficult to fix in the memory than what is regularly used or encountered. In the above case, the intended construction has been erased from the memory in its perfect form on account of its anomalous nature, and the more familiar paraphrase has taken its place. In Freud's explanation of the same case, the Austrian Jew arrives, through a sequence of 'free associations' with the forgotten word (under Freud's guidance) at a source of anxiety allegedly causing the slip (Freud, 1901, Ch. 1). The associations between words/ideas, briefly, are as follows: aliquis (omitted word) ~ reliquien --~ liquidation --4 fltissigkeit --4 fluid --~ San Gennaro (associated in Naples with the liquefying of clotted blood) ~ retardation of liquefying process (also believed, in Naples, to be San Gennaro's responsibility) ~ obsession with absent flow of liquid --~ failure to menstruate of an Italian woman with whom the Austrian had spent time in Naples, and who he feared he had made pregnant (anxiety causing slip). Timpanaro argues that his own account of the slip in terms of banalization is more convincing, in that this chain of associations is not 'cast-iron', as Freud claims it to be. He shows that the omission of any one of the other words in the quotation could, through free association, indicate the same anxiety, and so lead to the same alleged source of the slip. Some of his examples of pseudo slips and the way they can be related to the same anxiety about the possibly pregnant woman are: exoriare (pseudo forgotten word) arising ~ birth; nostris (pseudo forgotten word) ~ noster ---)pater noster --4 paternity; ultor (pseudo forgotten word) ~ eltern (parents) ~ parenthood; ultor (pseudo forgotten word) --, ultimatum. Timpanaro does not question the presence of the anxiety, but rather its designation as the cause of the slip. The implication is that this 'free' association is so free as to allow a focus on any word, not necessarily the word involved in a slip, to indicate a present anxiety. While this may be a useful tool in enabling people/patients to identify an anxiety they might otherwise find difficult to express, it does not in itself support the claim that slips of the tongue are symptoms of neurosis caused by the repression of such anxieties. As well as arguing that conclusions reached through free association are non-falsifiable, Timpanaro questions the spontaneous nature of the Austrian's associations, and the 'unconscious' nature of his anxiety. Firstly, the analysand was familiar with Freudian theory and keen (as well as fearful) that a significance for his slip should be discovered (Timpanaro, 1976, p. 52). Timpanaro suggests that his 'consent' is coerced by Freud's presence and authority, and also by his frequent and explicit interventions and suggestions during the course of the associations. Secondly, Timpanaro argues, the anxiety was clearly a conscious preoccupation rather than a repressed and therefore unconscious thought. Assimilation

Timpanaro examines another episode from The Psychopathology o f Everyday Life in which Freud again gives a repressed thought explanation for an instance of forgetting, this time his own. In

294

BARBARA MacMAHON

trying to recall the name of the artist 'Signorelli', the words which come to Freud's mind are first 'Botticelli' and second 'Boltraffio'. Freud claims that the names 'Signorelli' and 'Botticelli' were equally familiar to him (Freud, 1901, p. 7), thus dismissing the possibility of banalization. His explanation involves association with a conversation immediately preceding the instance of forgetting. In this conversation, thoughts of death and sexuality had remained unspoken, and were associated with the place names 'Bosnia', 'Herzegovina' and 'Traffoi'. In Freud's account, the association between 'Signor' and 'Her(r)zegovina', as well as sound similarities between 'Bosnia', 'Botticelli' and 'Boltraffio', and between 'Boltraffio' and 'Traffoi' account for the associations allowing for the re-emergence of his repressed thoughts. Here again, Timpanaro objects that this explanation is over-tortuous, and suggests that in spite of Freud's protestations, it is likely that the first substitution here is a simple case of banalization. Even discounting the possibility of banalization, Timpanaro offers a more credible account in terms of a substitution based on sound and semantic similarity alone (Timpanaro, 1976, pp. 63-76). He suggests that the initial move from 'Signorelli' to 'Botticelli' involves the substitution of one word for another in the same narrowly defined semantic category (Italian artists), with an identical stress pattern, equal number of syllables, and high proportion of identical phonemes. The move from 'Botticelli' to 'Boltraffio' is then seen as 'secondary disprovement' in that the correct part of the first substitution--elli--is changed, and the incorrect p a r t - - B o - - i s maintained. Timpanaro reinforces his argument here with many similar examples from textual criticism, and also suggests that such secondary disimprovements are much less likely to be banalizations; that 'Boltraffio', despite being comparatively rare, may have been the only available substitution similar in terms of sound and meaning to 'Botticelli'. In this way Timpanaro attributes such slips entirely to minor selectional errors in language production, suggesting that recourse to psychoanalysis is unnecessary and unjustified in such explanations. Timpanaro accepts the therapeutic potential of free association, but argues that it cannot be relied upon as an explanatory account of errors. More importantly, he argues that the techniques of psychoanalysis (including free association) put the patient/analysand in a position of inferiority in relation to the psychoanalyst. The patient either accepts or denies (covertly accepts) Freud's analysis, but may not 'advance an alternative explanation' (Timpanaro, 1976, p. 60). His conclusion is that psychoanalysis is a deterministic theory which took over the consolidatory role of religion in providing an indisputable, authority; the analyst knows the patient better than she knows herself, and is able to extract confessions of sinful repressed thoughts and wishes of which she was previously unaware (Timpanaro, 1976, p. 171). The patient thus loses control and hands over personal responsibility and power to the analyst. The following provides a summing up of Timpanaro's views: The neo-bourgeois of this sort, in contrast to his predecessor, has understood that just as Christ did not c o m e into the world in order to abolish the ancient laws but to accomplish them, so psychoanalysis does not demistify bourgeois values in order to destroy them, but to reinstate and consolidate them. Thus as p s y c h o a n a l y s i s g r a d u a l l y ceased to be a moral scandal and b e c a m e a v o g u e . . . so too, the explanation o f 'slips' b e c a m e a polite pastime (Timpanaro, 1976, pp. 132-133).

Psycholinguistics Timpanaro shares with psychoanalysis the view that slips, in particular word substitutions, are facilitated through the existence of semantic and phonological similarities between words. In this section I show how psycholinguistic models hypothesize on the process by which words are produced. Though there are different types of model, all of them posit moments in stages of

295

THE FREUDIAN SLIP REVISITED

production where words related in one way or another can become confused. This cognitive relation of similar words can be equated with the primary process operations posited by Freud and developed by later psychoanalysts Lacan and Kristeva, and so psycholinguistic insight into lexical production can be used to test theoretical claims that literary innovation resembles the Freudian parapraxis.

Models of lexical selection One model of speech production, described by Aitchison (1987, pp. 165-171) as a steppingstone model, is based on a series of discrete non-interactive stages. In lexical selection, the first step is to choose meaning and word class. Following this, the selector moves on to the next steppingstone, guided by a signpost from the first stone in which a phonological area code is given in order to direct the selector to more or less the right area of the sound-organized part of the mental lexicon. Semantically related words are also present on the first stone, before narrowing down to the target word. Aitchison illustrates this in Fig. 1. In this model, a word selected by mistake on the first stepping-stone would be semantically related; in this example, the relation would be one of hyponymy. On the second stepping-stone, a word could be mistakenly selected because of sounds in common with the intended word. The important point here is that the process involves separate stages, so an error would be caused by either semantic relation or sound similarity, but not both. MEANING

SOUND

badger bearer beggar

beaver . ~ ~

Fig. I. A stepping-stone model with narrowing down (Aitchison, 1987, p. 171). An alternative to the stepping-stone model is the interactive model in which the process of selection does not flow in one direction from one stage to another, but can flow back to previous stages, providing feedback which in turn can affect selection (see Stemberger, 1985). Stemberger refers to this process as 'spreading activation', and the model is based on a system of units and links, analogous to neurons and synapses. Each unit has a resting level of activation, and links carry activation and inhibition (best thought of as a removal of activation) between the units, thus raising or lowering resting activation levels. W h e n one unit is highly activated, it has an inhibitory effect on other units at the same level. The units are units of l a n g u a g e - - p h o n e m e s , words, semantic representations etc,, each of which, through relations of similarity, can trigger activation of other units at a lower or higher level. In lexical selection, activation can spread from a word to a set of phonemes, and back again. Words accumulate or lose activation, and finally one word is activated enough to be selected (Aitchison, 1987, pp. 173-175; Stemberger, 1985, pp. 145-147). Stemberger illustrates and explains the lexical selection process as shown in Fig. 2. In these figures,an arrow denotesan activatinglink, while a filledcircle is an inhibitorylink. A double line represents a large amount of activation, a single solid line somewhat less, and a broken line even less. Some of the inhibitory links have been left out.., for clarity. The exact nature of semantic representation is irrelevant here, beyond the assumption that it is composed of features; I will simply place a word in quotation marks to represent its meaning (Stemberger, 1985, p. 148). In this example, the semantic representation (in quote marks) strongly activates the word 'feather'

296

BARBARA MacMAHON "FEATHER"

Fig. 2, Semantic and phonological effects on lexical access (after Stemberger, 1985, p. 148).

(ringed), which in turn activates the phonemes [f], [e], [~] and [r]. The semantic representation also activates the word 'hair', and weakly activates the word 'leaf'. The phonemes [f], [e], [0] and Jr] all reactivate the word 'feather', phonemes [el and [r] also reactivate the word 'hair'. In addition, phonemes [e], [0] and It] activate the word 'leather', and phonemes If] and [r] activate the word 'favour'. The word 'feather' accumulates a lot of activation through the bold arrow from the semantic representation, and from the arrows from each of the phonemes. Activation, coming from above and below, causes the word 'feather' to inhibit other words, also inhibited by one another, and the build-up of activation in 'feather' finally reaches a level which results in its selection. The interactive type of model is theoretically better than the stepping-stone model because the system is simpler; there is no need to posit two mental lexicons based, respectively, on phonological and semantic modes of organization, and there is less need for a final monitoring device following selection. More importantly, this model can account for the many speech errors which show a relation of both sound and meaning between uttered and intended words.

Lexical selection models and the slip of the tongue In the stepping-stone model, the account of speech errors is as follows. In sound-related word substitution errors, mistaken selection occurs on the second stepping-stone, where, in the previous example, 'beaker' might be chosen in place of 'beaver' because of its close phonological correspondence with the signpost. Semantic associations between uttered and intended words, as in, for example, the mistaken selection of 'badger' for 'beaver', would occur here only by coincidence. Semantically associated word substitution errors would occur on the first steppingstone, where another lexicon is organized into semantic categories. In this example the words are all names for small woodland animals, information presumably contained in the signpost to this stone. As in sound-related substitutions, mistaken selection arises from contiguity in the mental lexicon, but here contiguity is determined by semantic similarity. On the first stone, the organizing principle is something like that of a thesaurus, and on the second something like that of a dictionary. In Stemberger's model the source of errors is 'noise'. Resting activation levels can fluctuate. A unit might have a higher than normal resting level of activation for three reasons (1) random variation, (2) variation in frequency of usage (coinciding with Timpanaro's banalization argument), and (3) 'systematic spread of activation to non-target units from semantic units or from feedback' (Stemberger, 1985, p. 150). The raised resting activation level of non-target units in addition to

THE FREUDIANSLIP REVISITED

297

"PROPOSAL"

Fig. 3. Interactive model for sound-related substitutions.

activation in the selection process can lead to a very high level which results in selection. In soundbased substitutions, the process would be something like that shown in Fig. 3, a real error in which 'routine promotion' is chosen instead of 'routine proposal'. Here, frequency of usage may cause the word 'promotion' to have a higher resting activation than normal; 'routine promotion' is arguably more frequently used in general than the intended 'routine proposal'. Or the resting level of 'promotion' may simply be randomly higher than normal. The higher than normal level plus activation from the phonemes [pr], [ ~], [ ~ ] and [~], as well as from the similar sound shape of the word in terms of syllable number and stress pattern (not shown in the diagram) are enough to activate the word to a level which results in its selection. In this type of substitution, a high degree of activation comes to the non-target word from the phonological level, but none comes from the semantic level. "ASPECTS"

CSY"TAX3

Fig. 4. Interactive model for semantically and phonologically related substitutions.

Figure 4 shows another example, this time of a semantic associative word substitution which also shows some sound similarity. Here the non-target word 'Structures' (as a linguistics book title) is activated by the semantic representation as a hyponym of the target word 'Aspects' (also a linguistics book title). The model is different from the stepping-stone model, but the principles determining mistaken selection are the same. Activation of words means that they are more likely to be mistakenly selected, and activation depends on similarities--phonological or semantic. In this model, though, a unit

298

BARBARAMacMAHON "DEATH AND SEXUALITY" / / / / / "BOSNIA HERZOGOVINA"

\ \ \ \ N. "TRAFFOI"

HE

.

.

.

"SIGNORELLI"

"SIGNOR"

"HERR"

Fig. 5. Unlikely interactive model for Freud's 'Signorelli'/'Botticelli"slip. can be activated by a high degree of sound similarity alone, or by a mixture of semantic similarity and sound similarity. In Fig. 4 we can see that 'respects' is activated through a fairly high degree of sound similarity with 'aspects', but there is no semantic relation, making it less likely to be erroneously selected than 'structures', whose initial semantic activation is reinforced by the phonemes it shares with 'aspects'. Psycholinguistic models, then, posit a stage in the production in which the formal and semantic similarities between words can result in erroneous selection, and in this sense they can be seen to coincide with psychoanalytic theories which posit a primary process which operates on similarities between words in slips of the tongue. Where they differ is in their hypotheses of the sources of errors. Kristevan theory is not so much of a problem here, since the primary process operations themselves constitute the re-emerging of the repressed desire to babble, so the speech

THE FREUDIAN SLIP REVISITED

299

production mechanism which operates on sound similarities can be seen as evocative of this repressed desire. In the classical psychoanalytic account of slips, though, while the psycholinguistic account sees banalization and random variation accounting for slips, Freud claims that mistakenly selected words can be traced through a sequence of associations to an underlying anxiety. In terms of the interactive model, Freud's explanation of the 'Signorelli'/'Botticelli' substitution would have to work as shown in Fig. 5. This model is unlikely. Apart from the unlikely activating links, even if 'Botticelli' had been at a higher than normal level of activation before the process of selection, the distance between the uttered word and its alleged source is great, and there are many units in between. This means that activation is dispersing through the units rather than concentrating in one or a few units. The further activation spreads, the less effect it is likely to have. This slip can in fact be explained much more satisfactorily in terms of the interactive model in the same way as other sound- and meaning-related word substitutions, like the 'Aspects'/'Structures' example given earlier. This would be a result of the activation of 'Botticelli' through its close hyponymous relation with 'Signorelli', reinforced by activation through phonological similarity and a likely higher resting level of activation caused by greater frequency of use or random variation, as shown in Fig. 6. The psycholinguistic models, then, seem to be reconcilable with aspects of post-Freudian "SIGNORELLr'

Fig. 6. Likely interactive model for Freud's 'Signorelli'/'Botticelli' slip.

psychoanalysis rather than with Freud's own account of slips of the tongue. The psycholinguistic models also go some way towards defining the types of similarity which can be influential in producing word substitutions. In following sections I show that, using the interactive model as a basis, it is possible to observe naturally occurring word substitutions and use observations for predictive purposes. If we can predict the degree and type of similarity between words necessary to enable a substitution, we can then test the claims outlined at the beginning of this paper on the nature of literary innovation. /

The analyses Freud's claim is that the primary process operates on similarities between words, and though there are problems with his account of the origin of speech errors, the Kristevan and Lacanian development of psychoanalytic theory appears to offer a more reasonable proposal that the reemerging repressed element in slips of the tongue is precisely this tendency to associate similar sounding words. This coincides with psycholinguistic models of speech production in which sound similarities and semantic relations between words lead to their activation in the selection process.

3~

BARBARA MacMAHON

In this way primary process operations might be seen as equivalent to the operations of the speech production mechanism, so examining errors produced by this mechanism should provide insights into the operations of the primary process. In the following sections I describe the design, method and results of the analysis of the first 20 of a collection of naturally occurring word substitutions taken from Fromkin's corpus (Fromkin, 1973). This analysis is an attempt to find out exactly what degree of sound similarity or semantic relation (or what combination of the two) there has to be between uttered and intended words to enable a substitution. The results are used to make predictions for naturally occurring substitutions in general. These predictions are then applied to 20 more substitutions from the same corpus to see if they are accurate. Modifications based on the second analysis and on the findings of other studies are added to the set of predictions, which are then applied to a sample of word substitutions from Finnegans Wake. The aim of this is to see whether or not the same type and degree of similarity exists between written and implied words in Finnegans Wake, and so whether the same operations can be said to be at work in slips of the tongue and in Joyce's writing. Following the analysis of Joyce's word substitutions I discuss the issue of interpretability, considering how interpretable, and in what way, results of primary process operations can be expected to be. Speculations about interpretability in Joyce's word substitutions lead to the construction of another test, again designed to show similarities and/or differences between naturally occurring substitutions and substitutions from Finnegans Wake. I then go on to apply the same predictions and interpretability test to substitutions (the original 'malapropisms') from Sheridan's The Rivals in order to compare them with both naturally occurring and Joycean substitutions. The purpose of this part of the analysis is to see whether the substitutions from a supposedly 'revolutionary' text differ from those in what I presume would not be called a revolutionary text, and again to see whether either of them has anything to do with discernibly unconscious processes. Finally I draw conclusions about differences between all three samples in terms of production and reception, and I discuss the implications of these findings for the literary theory outlined in the introduction.

Naturally occurring word substitution The sample The sample of forty naturally occurring word substitutions was taken from Fromkin's corpus (Fromkin, 1973, pp. 243-269). I took the first 40 on her list of word substitutions, with the exception of the following: 'A' for 'B', '3' for '59' and 'B' for 'A'; these were excluded as letters and numbers rather than 'words' as such. A list of the 40 substitutions is given in Appendix 1. Of these, the first 20 were the object of the first analysis. The analysis In the analysis of the first 20 naturally occurring substitutions the aim was to identify the types of similarity which can lead to the activation and selection of the unintended word by the intended word. Stemberger (1985) suggests that activation can spread from semantic, syntactic and phonological representations. The analysis was thus based on the need to identify similarity at these levels. Phonemes Phonemic similarity was measured in terms of identical phonemes in the same position in each

THE FREUDIAN SLIP REVISITED

301

word. Transcriptions were made in the international phonetic alphabet. In this part of the analysis a maximized onset model of the syllable was used (see Clements and Keyser, 1983, pp. 37-52). This means that syllable divisions were made by assigning all consonants to syllable initial rather than syllable final position, unless this resulted in a cluster not possible in English. For example, with the word 'cluster', phonemically/klAstzr/, the syllable boundaries would be as follows:

$1

S2

onset

rime

kl

A

I

onse~nme

I

I

I

st

or

and not: $1

on

I

kl

$2

I

e

onset

rime

t

ar

I

As

I

In other words, both consonants [s] and [t] can be put into the onset position of $2 because [st] is a permissible cluster in English. This means that S 1 (and here $2 as well) ends with a vowel, or zero consonant, and so is an open syllable (see Fudge, 1984, p. 19). An example of a closed syllable occurs in the word 'signify'--/slgmfal/--which, according to this model, cannot be segmented as follows:

.,•.1 onset I

$2

r|me

I

s

onset

I

I

$3

rime

I

gn

onse~nme

I

l

I

f

a]

because [gn] is not a possible English cluster, so the maximum number of phonemes, here [n], is placed under onset in $2, and the remaining phoneme(s), here [g], constitute the end of the rime in SI: $1

$2

onTet'~r'me s

I

Ig

onse~nme

I

n

I

I

onset f

rime

I

a[

S 1 is thus a closed syllable here. This model of the syllable is a conventional one based on perceived syllable boundary; these are the divisions people make intuitively when asked to articulate syllables separately (see Kahn, 1980, p. 39). The maximized onset rule is the subject of some debate (see Fudge, 1984; Kahn, 1980; Malmberg, 1967). Malmberg argues that in connected English speech the maximized onset rule does not apply across word boundaries: 'an aim' remains phonetically distinguishable from 'a name' (Malmberg, 1967, p. 80), and therefore that the use of a consonant in the rime rather than the onset position probably helps to distinguish between words. Despite this, the maximized onset rule was used for ascertaining syllable boundaries in this analysis, firstly because it was concerned with single words rather than connected speech, and secondly because it was felt that as long as all the samples were analysed in the same way, any significant differences between them would be discernible regardless of the accuracy of the syllable model.

302

BARBARA MacMAHON

In the above example 'cluster', the cluster [kl] fills only one slot, the onset position, with two phonemes. This means that in the analysis, onset identity where clusters were present had to consist of identical clusters (two of three phonemes) rather than of just one identical consonant. Where only one phoneme was shared between clusters in two words, this was considered a 'half' identity. As well as noting identical phonemes and clusters in identical positions, 'zero-consonanr and 'zero-vowel' were also considered. Zero-consonant means that there is no phoneme at all in a position where it is possible to have a consonant, that is at the beginning and end of every syllable. For example, in the word 'edifice', phonemically/~&fis/, there are three syllables and therefore six potential consonant positions, occupied by three consonants and three zeroconsontants (N and C stand for nucleus and coda, units within the rime; zero-consonant is noted as Oc):

Sl onset

ec

$2

on~J'~nme

nme

s

S3

ec

d

I

onse~rlme

ec

f

I

s

The idea of zero-consonants used here is adapted from Kiparsky's concept (Kiparsky, 1970). He discusses the Finnish epic poem The KaIevala, in which all the words of each line generally alliterate. There are some lines, however, which superficially appear to deviate from this pattern in containing words which begin with different vowels. Kiparsky claims that this apparent anomaly can be explained if we consider two words which begin with different vowels to be initially identical in that neither of them have an initial consonant, or both have a zero-consonant. Hockett also argues for the existence, at a representational level, of a zero-consonant (Hockett, 1967). The analysis shown above extends this idea to positions within the word. Zero-consonants are very frequent in English words of more than one syllable. Zero-vowels occur only in syllabic clusters ending with [1], [m] and [n], as in 'table', 'rhythm' and 'Britain':

$1

$2

onse~rlm e

onse~nme

0c

b

Ov I

r

l

ec

~

ev m

br

I

ec

t

ev n

In this way the uttered and intended words were compared for phonemic similarity, each syllable having three potential points of identity (onset, nucleus and coda), the total number of identical phonemes being a proportion of the total possible. Each pair of words was analysed as in the following example substitution, where 'promotion', as mentioned previously, is uttered in place of intended 'proposal'. Here O, N and C stand for onset, nucleus and coda, and points of identity are shown in bold.

303

THE FREUDIAN SLIP REVISITED

O N C O N C O N C p r ~ 0c m

auec ~

pr e ec p e U e c 1

1

1

1

ev n

z ev I 1

1

identical phonemes/clusters total: 6/9

Cases were also notes where the same phoneme occurred in both words, but in different positions, as in, for example, the substitution 'prostitute' for 'protestant':

O N C O N C O N C pr lo 0c st [ pr b 0c t ~ ~ s t

Oc t

ju

t

ant

It is generally thought that word substitution errors occur at a level of underlying representation rather than surface realization (e.g. Durand, 1990, p. 5). However, underlying phonological forms were not used in the analysis since it was felt that the analysis of surface forms would provide enough information to show what constitutes similarity. Often surface forms which were different from underlying forms had undergone the same phonological changes in both words. For example, in the substitution 'generation' for 'generalization', the vowel in the third syllable has undergone diphthongization in both cases, and the vowel in the final syllable has undergone vowel reduction, also in both cases (Chomsky and Halle, 1968, pp. 178-190). This means that underlying identity has not been lost in my analysis by considering surface forms only. In the few cases where there was underlying identity but surface difference, this was also noted in the analysis, as in, for example, the underlying identity of the final 's' in 'structures' ([z]) and 'aspects' ([s]). Features As well as single phoneme and cluster identity, identity between features was also considered. Stemberger suggests that identity between features can cause activation (Stemberger, 1985, p. 151). In the example 'gone' for 'come' shown in Fig. 7, features shared between phonemes [g] and [k] might contribute to the spreading activation process previously described (here the abbreviations are 'vel' for velar, 'plo' for plosive, '-vo' for voiceless). Because of this possibility, identity between features was also noted in the analysis. Syllable number and stress Where uttered and intended words had the same number of syllables, this was also noted, as were identical stress patterns, again on the assumption that identity here could cause activation. Where the uttered and intended words differed in syllable number, that is where identical stress patterns were impossible, the words were analysed in terms of whether or not the stress patterns could be seen as similar. For example, in the substitution 'instamatic' for 'immediate'--/lnstom~et~k/ for l m i d j ot/--stress patterns were considered similar in that both words have secondary and primary stress, carried by the first and penultimate syllables, respectively, in both words.

3~

BARBARA MacMAHON

f---

"COME"

k

Fig. 7. lnfluence of phonological features in lexical selection.

Word class In Stemberger's model, words activated by phonological and semantic relations with the target word do not differ from it syntactically. It is likely that the target word would strongly inhibit words of a different class. In addition, syntactic structure partly predicts and therefore activates appropriate word class(es) in lexical selection (Aitchison, 1987, p. 100). For this reason the sample was also analysed for the unlikely possibility that uttered words might differ in word class from their intended counterparts. Semantic relations Semantic relations between words are more difficult to analyse. Aitchison suggests that words might be linked semantically in the mental lexicon through synonymic, hypnoymic (or hyponym-superordinate), antonymic and collocational relations (Aitchison, 1987, Ch. 7). The sample was checked for synonymic, hyponymic and antonymic relations. Collocational pairs such as 'agonizing' and 'decision', while they are probably linked in the lexicon, are unlikely to substitute for one another because of lack of word class agreement, so this type of relation was not included in the analysis. Twenty word substitutions from Fromkin's corpus Activating relations Figure 8 shows the analysis of the first 20 naturally occurring substitutions. All pairs of uttered and intended words showed identity of word class, and all but one--'Hawaii' for 'Illinois'--showed an identical or similar stress pattern. Eighteen out of twenty shared the same number of syllables. In other respects the degree of semantic and phonological similarity between uttered and intended words varied. Predictably, if a pair was lacking or low in one type of similarity this would be compensated for by the presence of another type of similarity. Those with no semantic relation at all (there were five of these) had a high degree of phonemic identity. For example, the pair 'prostitute'/'protestant' has six and a half phonemes and clusters in common out of a possible nine.

305

THE FREUD1AN SLIP REVISITED

prostitute/protestant rent/check ,,dialect/dialogue belt/collar 'instamatic~mmediate cows/cars beer/wine vowels/values eyes/hair first/last Hawaii/Illinois , Promo!ion/proposal there/here Steve/George tree/roof Italy/Boston unarmed/unharmed shortJlong 'Structures'/'Aspects' Purdue/Peru

6.5/9 3 1/3 t/k 1 7/9 3 0 1 &2 4/9 k/t 4 & 3 2/3 1 0 1 2/6 2 0 1 1/3 1 3/9 3 6/9 3 1.5/3 1 0 t .5/3 1 0 2 516 2 1/3 1 .5/6 k/p/t 2 4.5/6 2

4. V 4" ./" ,/ ,t / 4" / 4" x 4. , 4. ,/ / .,t V" ~/ 4" 4"

/ 4 r 4" 4. / 4" 4" ¢" 4" 4" 4. 4" ,/ 4. t 4. 4. 4" /

st

I t'

/

/ /. 4"

I 4" / 4.

t

s,t,k

I

4 /* 4.

4" 4"

"in this and following figures indicates semantic relation with eadier word in utterance Fig. 8. First 20 substitutions f r o m F r o m k i n ' s corpus.

Conversely, those with no phonemic identity had semantic relations, as in hyponymic pairs 'Steve'/'George' and 'belt'/'collar'. Examples with both phonological and semantic similarities are 'there'/'here'--one and a half identical phonemes out of a possible three, same syllable number and antonymic relation, 'Hawaii'/'Illinois'--three out of nine possible identical phonemes (here zero-consonants), same syllable number and hyponymic relation. Lack of phoneme or cluster identity in identical positions was sometimes compensated for by the presence of identical phonemes in different positions, as in 'structures' for 'aspects'. Some uttered words showed a semantic relation with a previously uttered word, as in 'beer' for ' w i n e ' - - i n hyponymic relation to previously uttered 'ale' noted by Fromkin. Some cases of banalization are evident. For example, given that the work environment of the collector of these substitutions is a university linguistics department, the substitution 'vowels' for 'values' probably constitutes a banalization since 'vowels' is likely to be a frequently used word in this environment. Similarly in 'Purdue' for 'Peru', Fromkin notes that the speaker bad recently been to Purdue. Predictions

Six definite predictions can be made from these examples: (1) Uttered and intended words will always share the same word class. (2) Uttered and intended words will almost always have identical or similar stress patterns. (3) Uttered and intended words are likely to have an equal number of syllables. (4) Where uttered and intended words have no semantic relation they will have a high degree of phonological similarity.

306

BARBARA M a c M A H O N

(5) Where uttered and intended words have little or no phonological similarity, they will have semantic relation. (6) Many substitution pairs will share phonological and semantic characteristics: phonemic/ cluster identity can vary from 0 to 7/9; semantic relation can be one of three types. The problem with this set of predictions is that the looseness of (6) will predict possible substitution between words with almost any semantic or phonological relationship. The problem in (6) arises from the nature of the activation process. Activation can come from any of the types of similarity, it does not matter which ones. The important thing is that activation can reach a level high enough to trigger selection (see also Morton, 1979). For this reason, instead of trying to continue with individual predictions about types of similarity between words, it was decided to devise a points scheme, based on the analysis of the first 20 substitutions, in which points are given for each type of similarity; the total number of points then has to be over a certain level to allow for the substitution of one word by the other. In the devising of the points scheme, observation and intuitive speculation were the basis for assigning a given number of points to a certain type of similarity. The scheme was reapplied several times to the analysis of the first 20 substitutions, each time raising or lowering points in order to get the totals of all the word pairs over the selection level; since they are all naturally occurring errors, they should all be shown by the scheme to have the potential to substitute for one another. To illustrate, the number of points given for an antonymic relation was at first 3, for the same number of syllables it was 2, and for phonemic identity it was the number of identical phonemes as a proportion of 9. The number of points required to allow for a possible substitution was 9. This excluded 'first' for 'last', which would have 3 points for phonemic/cluster identity, 2 points for the same number of syllables and 3 points for an antonymic relationship, leading to a total of 8. Because of this and other exclusions the number of points for an antonymic relation was finally raised to 8. Modifications were made in this way until a satisfactory degree of accuracy had been achieved for this half of the sample. The final points scheme at this stage was as follows:' Relations between uttered and intended words: same syllable number identical phonemes/clusters as a proportion of identical phonemes in different positions (each) feature identities antonymic relation hyponymic relation synonymic relation obvious banalization

+ points 2 9 1

-

lack of word class agreement lack of similar/identical stress pattern

0.5 8 8 3 3 points 10 5

Relations between uttered and previously uttered words: antonymic relation hyponymic relation synonymic relation Points needed to enable a substitution

+ points 4 3 1.5 9 or more

THE FREUDIAN SLIP REVISITED

307

In this points scheme most of the substitutions in the first sample get 9 points or more. Those which get less than 9 are explainable in other terms. The substitution of 'eyes' for 'hair', where the uttered phrase was 'blond eyes', gets only 8 points. It seems probable that this may have been an elision error for intended 'blond hair and blue eyes'. The substitution 'tree' for 'roof', which gets only 6.5 points, could be a blend of the two phrases 'a branch falling off the tree' and 'a branch falling on the r o o f ' - - t h e uttered phrase is 'a branch falling on the tree'. 'Rent' for 'check', which gets 6 points, could be an elison and transposition of nouns in intended 'pay the rent by c h e c k ' - the uttered phrase is 'pay by rent'. The predictive accuracy of the points scheme was then tested by applying it to the second 20 naturally occurring substitutions and comparing it with the findings of other studies. Twenty more word substitutions from Fromkin's corpus Figure 9 shows the analysis of the second 20 naturally occurring substitutions. The points scheme scores are also given for each pair. All except two of the word pairs in this second sample scored 9 or more points, suggesting that

.

generalisation/generation beer/dinner

_ ivy/ivory

Chinese/Japanese answers/questions high/low west/east closes/opens slower/smaller foot/back neck/nose dissent/content included/excluded blind/deaf

thumb/finger round/back

brother/father nose/head distressed/impressed toes/fingers

! 9/12 0 4/6

514 1&2 2

4. 4. 4"

4. 4. 4.

4/6 tJ'/clj 2&3 1.5/61 s/J" 2 1/3 1 1/3 1 3.5/61 2 3.5/6 w/I 2 0 k/t 1 113 1

4 4" 4" 4" 4" 4" 4. 4"

4. 4" 4" 4" 4" 4. "/ 4"

2/6 s/t 2 7.5/9 3 0 1 0 mlrJ9/f11&2 0 1 4/6 2 I 0 I, 1 3.5/61 p/t 2 i 0 i 1 &2

4" 4" 4. 4. 4. 4" 4" 4" 4"

4" 4. 4 4. *~ 4" 4" 4" 4"

I

?

n,r

er 4

? 4.,to

8.75 13 12

4.

15 4" 15.25 4" 13 4 13 4 15.25 9.25 11 13

v

z I 4" 4"

d

st z

4. / ./ 4" 4" ~"

6.5 17.5 11 11 10 16 10 9.25 9

Fig. 9. Second 20 substitutions from Fromkin's corpus.

the points scheme has a fairly accurate predictive value. The pairs which scored less than 9 points were 'generalization' for 'generation' (8.75), and 'dissent' for 'content' (6.5). With 'generalization' for 'generation' it may be that banalization or hyponymy can account for the error. The utterance is 'the phrase structure generalization'; the substitution is possibly hyponymic in that generalization and generation are both things which grammatical theory does. 'Dissent' for 'content' is very probably activated by preceding 'dissertation' in the uttered phrase 'you can rewrite your dissertation to your heart's dissent'. Phonemic perseveration like this was not included in these first analyses, although semantic perseveration was. If a previously activated word can have an

308

BARBARA MacMAHON

activating effect on a candidate word through semantic relation, it should be able to activate through phonological similarity too. The discovery of this example led to the addition of sound perseveration to the points scheme, with a value of 3 points. Other studies of naturally occurring word substitution differ from this one in terms of focus and categories of analysis, but there are no discrepancies of findings important enough to invalidate the points scheme so far. In research into the structure of the mental lexicon, Fay and Cutler (1977) analyse 183 examples of production errors----~at is inadvertent slips of production rather than errors of encoding (errors which speakers believe to be correct). Their findings on the relationship between uttered and intended words are as follows: grammatical category agrees 99% stressed vowel agrees 54% consonantal onset agrees 50% 87% syllable number agrees stress pattern agrees 98% In my analysis no distinction was made between inadvertent slips and encoded errors. Zwicky's study of encoded errors (Zwicky, 1978) uses similar categories, and his results are as follows: grammatical category agrees 91% stressed vowel agrees 70% consonantal onset agrees 77% syllable number agrees 71% stress pattern agrees 93% Zwicky argues that the high percentage of stressed vowel agreement here is typical of encoded errors, higher than it would be in inadvertent slips. This type of similarity was not noted in my analysis, but in fact only six of the first twenty and seven of the second twenty pairs show stressed vowel agreement. This coincides with Zwicky's finding that the word substitutions in Fromkin's corpus are probably all inadvertent slips. It does not, however, coincide closely with Fay and Cutler's figure of 54% stressed vowel agreement for inadvertent slips. Because of this slight discrepancy, the possible similarity of stressed vowels was not written into the points scheme. Inadvertent slips and encoded errors also differ in frequency of identity of consonantal word onset, another type of similarity not specified separately from phonemic/cluster identity in my analysis. Fay and Cutler's proportion of consonantal onset agreement (50%) leads them to the conclusion that words are stored contiguously in the mental lexicon on the basis of left-to-right ordering. This means that the mental lexicon would be structured along the same lines as a dictionary, with words grouped together according to identical initial phoneme. Hurford (1981) contests the left-to-right hypothesis by taking Fay and Cutler's corpus and constructing a pseudo corpus of words from a dictionary. These were chosen if they maintained identity of phonemes in a sequence at the beginning of the words. Phonemes to the right of the first point of difference were ignored. Hurford shows that identity to the right of the first point of difference was far more frequent in Fay and Cutler's real corpus than in his pseudo corpus, suggesting that phonemic identity to the right is as important as identity to the left. Aitchison and Straf (1982) also address the question of left-to-right ordering in a comparison of child and adult substitutions. Their analysis shows that

309

THE FREUDIANSLIP REVISITED

The majority of malapropisms preserved both a like initial consonant, and the same number of syllables: but when a change was made in one of these~ adults preferred to change the number of syllables, and children the initial consonant (Aichison and Straf, 1982, p. 765).

Part of their conclusion is that Greater familiarity with the written word is likely to lead adults to concentrate on word initials,

and As children get older increase in vocabulary size may lead to the necessity for a more parsimonious system for storage and retrieval, since in English, word beginnings distinguish between words more efficiently than word endings (Aichison and Straf, 1982, p. 773).

In my sample only 8 of the first 20 pairs and 4 of the second 20 share the same initial consonant. Since this type of identity is the subject of some debate, and since the sample used here appears to support Hurford's view that consonantal onset identity is not highly significant in word substitutions, it was not added to the predictive points scheme. General studies of speech errors and word association tests suggest that activation spreads easily and frequently between hyponyms, particularly co-ordinate pair hyponyms such as 'fmger'/'thumb', 'knife'/'fork', as well as between antonyms (Aitchison, 1987, pp. 75-77), corresponding with, the high number of points assigned to these types of semantic relation in the points scheme, now adapted to give four rather than three for co-ordinate pair preservation. Synonyms and superordinate-hyponym relations, excluded or given fairly low scores in the points scheme, seem to form fairly weak links (Aitchison, 1987, pp. 79-82), though in the case of synonyms, of course, this may simply be an indication that the substitution of one for another is likely to go unnoticed. Aitchison and Straf (1982) argue, as does Hurford (1981), that words are stored in the lexicon with derivational affixes. Evidence for this is that errors of derivational affixation such as 'a happy unboy' for 'an unhappy boy' do not occur. In agreement with this, affixes were not removed in my analysis. In fact if affixes were removed this would leave a number of pairs such as 'smaller'/'slower' with less phonemic identity and total scores of less than 9. A feature of substitutions noted in all of the above studies, and absent from my own initial analysis, was that words affected are almost always content words. This is also a feature of aphasic neologisms (Ellis, 1985). Since, like word class agreement, this is almost always the case, this was also written into the points scheme as a prediction with an inhibitory value of -10; that is, if one or both of a possible substitution pair is not a content word, 10 points are deducted from the total score.

Modified predictions The points scheme was finally modified to the following:

Relations between uttered and intended words: same syllable number identical phonemes/clusters as a proportion of 9 identical phonemes in different positions (each) feature identities antonymic relation hyponymic relation synonymic relation obvious banalization

+ points 2 9 1

0.5 8 8 3 3

310

BARBARA MacMAHON

-

lack of word class agreement lack of similar/identical stress pattern non-content words

points 10 5 10

Relations between uttered and previously uttered words: antonymic relation co-ordinate relation hyponymic relation synonymic relation sound perseveration (per phoneme) Points needed to enable a substitution

+ points 4 4 3 1.5 1

9 or more

The predictive power of this points scheme also accounts for the Freudian example 'Signorelli'/'Botticelli' discussed at the beginning of this paper, which scores 18.25. This high score again indicates that activating links between these words are numerous and strong enough not to require recourse to a Freudian interpretation. Word substitution

in

Finnegans Wake

The sample This part of the analysis used a sample of 40 word substitutions taken from Finnegans Wake, listed in Appendix 2. The list contains my interpretation of implied words, plus page references for the 1964 Faber and Faber edition, along with some co-text for each example. The selection of examples was not easy, since whole word substitution is only one of many types of lexical innovation present in Finnegans Wake. The examples were selected by scanning from p. 410 to p. 491. The scanning method may have resulted in bias towards easily recognizable substitutions, unrepresentative of all the word substitutions in this work. This bias would/will have resulted in favour of psychoanalytic literary theorists in that I would have been likely to recognize more substitutions which could be activated automatically rather than consciously interpreted substitutions. Since results showed great differences between the Joycean and naturally occurring samples, this possible bias was disregarded. It does mean though, that if anything, Joyce's innovations are even more different from naturally occurring substitutions than this sample suggests. In collecting words for this sample, only complete English words were chosen. It may be that substitutions occur naturally across languages where the speaker knows more than one language, but I have not seen any integration of this possibility into language production models. More importantly, if I had included foreign words, they would only have been those foreign words with which I am familiar, and it was thought that this might skew results and introduce unmanageable complexity. Neologisms such as 'spuk' and 'ruching' (indistinguishable from unknown foreign words) were excluded, as were words with additions or inserts such as 'trestraversed', 'rignewreck', and blends such as 'whizzling' ('whizzing'/'whistling'), and 'styearcases' ('year'/'staircases'). The selection principle then was to take non-deviant English words which clearly substituted for other English words. The purpose of the analysis here was to apply the points scheme to the Joycean sample in order to see whether the same activating links are in operation here.

311

THE FREUDIAN SLIP REVISITED

Word substitutions from

Flnnegsns Wake

poppy/body .goosemothsrlgrandmother nocturnal/maternal shocked/shook .... yes/us _ betrayal/trial curlew/Carlow foIlyifollow il'" grunt/grant yap/up baddylPaddy • boost/hest him/home token/taken wore/were belaboured/beloved ham/him brothel/brother errsd/her, rd off/of

4/3 p/h ' 6.5/9 6/9 m/n 1.5/3 k/t 1/3 5.5/6 4/6 5/6 2/3 1/3 5/6 pro 2/3 2/3 5/6 2/3 6/9 I:~ 2/3 3/6 0/~. 2/3 2)3 f/v

4" 4" 4. 4. 4" 4"

¢ 4" 4" (4") x ¢"

/ / 4" 4. x 4"

9 8.5 9 7 -15 8.25

4" ¢ ¢ 4.

2

4"

(4")

4"

8

2 t 1 2 1 1 2

4" v / 4" 4" 4" 4"

x 4" x (4") x x x

4" ," x / 4" V" /

-.5 8 -15 10,5 -2 -2 -.5

4" 4. ¢

1

x

4"

4.

3

3 1 2

4. 7 4" ¢' 4"

4" x 4" 4" x

,t 4" ¢

9 -/, 8.5 8 -11

I 1

"~/ x

.....

Word substitutions from Flnnegans Wake (continued) dearth/earth grocery/rosary naught/not along/alone farmer/father gost/god quad/quid tesrs/dears chaste/jest insects/lncost commencs/commit deed/dead pinchably/possibly brunt/hrand sPat!el/special swarming/warming whipt/wept ignite/unite quaith/with load/law

2 3 3 t 1 3&2

~3 7.5/9 "~z 2/3 4/6 ,.nlrl

4,5/6 1.5/3 2/3 2/3 lt.'5/3 ' 5.5/6 I 4/6 2.5/31 5/9 1.5/31 5,5/6 5.5/t 2/3 3/6 1/3 1/3

t/d t/d ~1~

t/d

g/~

~/

4" 4. 4" 4"

6 5.5 6 4 -15 5.25

5 -3.5 5 -18 7.5 -5 -5 -.5 3

'4" 4" 4" 4"

6 .2 55 5 -14

I

1

¢

4.

¢"

3 1 2. 2 1 1 1 1 2 2 1 3 1 2 2 1

4" 4. 4." 4. ~r

¢" x x

/ x ,t

/

4"

4. 4. / 4. 4. / 4" 4"

2

4"

¢

4"

6.5

/

3.5

1

4.

x

x

-14

4"

1

4"

4"

/

5

4.

-17 2

4. ¢" 4.

4.

4.

4.

4.

,'4. 4.

4. 4. 4.

4.

x (4.)

¢

¢

¢ ,r / ~t ./

x 4. 4. 4" 4.

8

10.5 -12 -1 8.75! 7.5 8

i

[

9 s

-2.5 11.25 8 -.5 7 7 9.75 10.25 8

"4.

,t 4. 4. 4. 4.

4. 4.

I 5

7.5 -15 -4 5.75 4.5 5

6

4. 4"

-5.5 e.25 ¢ 5 4. -3.5 4. 4 4" 4 4" 6.'15 4. 7.25 ,t 5

Fig. I0. Word substitutions from Finnegans Wake.

Linking relations Figure 10 shows the analysis of substitutions from Finnegans Wake. Initially only eight of the forty examples taken from Finnegans Wake scored 9 or more in the points scheme designed to

312

BARBARA MacMAHON

predict the necessary degree of sound similarity and semantic relation between words in naturally occurring substitutions. I will go on to argue that these eight are also dissimilar to naturally occurring errors, but even the initial result suggests that the majority of Joyce's literary innovations do not operate on the types of similarity found in naturally occurring substitutions. The eight substitutions which fulfilled prediction criteria were 'grocery' for 'rosary' in 'recite my grocery beans' (10.5 points), 'tears' for 'dears' in 'this, my tears, i s . . . ' (9 points), 'insects' for 'incest' in 'commence incest' (11.25 points), 'swarming' for 'wanning' in 'swarming himself in his sunnyroom' (10.25 points), 'poppy' for 'body' in 'tis an ill wind blows no poppy good' (9 points), 'baddy' for 'Paddy' in 'come back baddy wrily' (10.5 points), 'belaboured' for 'beloved' in 'tearly belaboured' (9 points) and 'nocturnal' for 'maternal' in 'nocturnal goosemother' (9 points). These substitutions could also be discounted on the grounds that they show the opposite tendency to banalization in deviating from a familiar collocation such as 'dearly beloved', or in using what is generally likely to be a less frequently used word in place of a more frequently used one, as in 'swarming' for 'warming'. In fact nearly all of the Joycean substitutions show this tendency. Banalization is a reversion to the familiar, explainable in terms of the spreading activation model in that frequently used, more familiar lexical items have a higher resting activation level than others, making them more likely candidates for substitution. In the case of familiar collocations, it is also likely that sequences of more than one word are stored as units in the lexicon, in which case the sequences might substitute for less familiar sequences. This seems to be the case, as argued previously, when the familiar collocation 'routine promotion' substitutes for the less familiar but intended 'routine proposal' in the naturally occurring sample (see also Matthei and Roeper, 1983, pp. 184-185). It seems unlikely, then, that the reverse should happen as a naturally occurring phenomenon. Examples like the Joycean ones discussed here did not occur in the naturally occurring sample, and thus were not considered in the devising of the points scheme. However, given that this was such a prominent distinguishing characteristic of the Joycean sample, and given the unlikelihood of the same thing happening in naturally occurring errors, reverse banalization (I would use the term defamiliarization if it were not already used within literary theory in a rather different sense) was written into the predictive points scheme with a negative value of -3. This adaptation is added as a second score in Fig. 10. 'Insects' for 'incest' is also unlikely as a naturally occurring substitution in that 'insects' is a countable noun, and here is in the plural, while 'incest' is an uncountable noun. Similarly, 'curlew' for 'Carlow' and 'baddy' for 'Paddy' are unlikely in that they substitute proper names with ordinary nouns. These distinctions were not included in the analysis, but it seems reasonable to predict that activation will not spread between different noun types, given that this did not happen in any of the substitutions from the naturally occurring sample. Taking into account the improbability of reverse banalization and substitution between different noun types, none of the examples from Finnegans Wake can be seen to make connections between words on the basis of similarities of the same type as those which cause activation to spread in lexical selection. The main reasons for the low score of the Joycean pairs were: (1) reverse banalization (in 35 pairs), (2) lack of word class agreement (in 12 pairs), (3) lack of semantic relation (in all pairs) combined with lack of sufficient compensatory sound similarity, (4) presence of non-content words (in 3 pairs). The fact that a large number of Joycean substitutions showed reverse banalization means they have a tendency which is not only different from, but contrary to, that of naturally occurring

THE FREUDIAN SLIP REVISITED

313

substitutions. In addition, word class agreement was seen to be an important criterion in my own analysis of naturally occurring substitutions as well as in all other studies consulted. Word class agreement is the first and strongest prediction made by psycholinguists for word substitution pairs. Both the syntactic anticipation of word class in speech production, and the word class of the activated intended word, activate other words of the same class, and strongly inhibit words of different classes. This means that over a quarter of Joyce's substitutions, such as 'folly' for 'follow' and 'token' for 'taken', are not operating on a fundamental similarity recognized by the lexical selection mechanism or primary process. If we include those cases where substitutions were made between countable and uncountable nouns, or between ordinary nouns and proper names, the proportion of lack of word class agreement increases to 15 out of 40. Also, very strong is the prediction that only content words will be affected in speech errors. Non-content or function words are thought to be stored in a separate lexicon (Aitchison, 1987, pp. 99-100), and are therefore impossible to activate through similarities with content words. Joyce's examples, then, again show a different mode of operation with substitutions such as 'off' for 'of' and 'quaith' for 'with'. While phonemic identity was fairly high between written and implied words in the Joycean sample (all pairs had at least one-third of possible phonemes/clusters in common), there were no cases of semantic relation of the types observed between uttered and intended words in naturally occurring substitutions. Semantic relations of hyponymy, antonymy and synonymy were seen to be important, though not crucial, as activating links in the naturally occurring sample. Since the activating links at work in naturally occurring pairs did not seem to be at work at all here (with the exception of a degree of phonological similarity), an attempt was made to see what kind of connections do exist in Joyce's substitution pairs. There are apparent relations of semantic field in the following passage: By gum, but you have resin! Of these tallworts are yielded out juices for joinloils and pappasses for paynims. Listeneth! 'Tis a tree story. How olave thai ririle, was aplantad in her liveside. How t a n n o b o o m . . .

Here many of the words and word innovations have to do with trees: 'gum', 'resin', 'tallworts' (if 'wort' is a plant, 'tallwort' could be a tree), 'tree', 'olave' (like 'olive'), 'tannoboom' (like 'tannenbanm'). The concepts of words having to do with one another or being of the same semantic field are loose, however, and this type of connection was not evident in the naturally occurring sample. In that sample it was seen to be possible that a previously uttered word could have an activating effect on mistakenly selected words, either through sound similarity, as in the 'dissertafion'/'dissent'/'content' example, or through one of the three possible semantic relations, as in the 'ale'/'beer'/'wine' example. However, the only possible semantic relation of this kind at work in the passage quoted from Finnegans Wake is a superordinate-hyponym relation from 'tree' to 'olave' and 'tannoboom'--likely to be a weak relation, as previously discussed, and one which did not seem to be at work in the sample of naturally occurring errors. Semantic relations here are not those which are significant in the speech production mechanism, but are consciously contrived. It is clear, then, that Joyce's substitutions are not produced in the same way as naturally occurring slips. However, although the relations between the Joycean substitution pairs are different from those between naturally occurring substitution pairs, it is possible that a reader's recovery of implied words might be guided by sound similarity. Although most pairs were discounted as plausible manifestations of unconscious processes because of reverse banalization, lack of word class agreement, lack of semantic relation and presence of function words, it is possible that sound similarity, which was present to some degree, is used at an unconscious level to

314

BARBARA MacMAHON

activate other words of different word classes in a special reversal of the banalization tendency and relaxation of the word class constraint. With regard to word class, this might correspond with the 'repressed babble' theory, in that a reversion to the mechanism of associating through sound alone could be seen as disruptive of semantic categories and syntactic constraints--both of which are assigned by psychoanalytic literary theorists to the repressive order of the Symbolic. However, it would be difficult to see how such reversal and relaxation could be brought about. In the following section I argue that although there is a degree of phonemic identity between words in Joyce's substitution pairs, this plays little or no part in the recovery of implied words, and that interpretation is in fact guided by the use of collocational phrases in the reverse banalization tendency described above.

Inference, recovery, interpretability and disruption So far I have claimed that the semantic relations between words vaguely associated with one another are not of the kind which could have an effect in producing speech errors, and therefore that the loose semantic field relation between word innovations in Finnegans Wake cannot be ascribed to the operations of the speech production mechanism or the primary process. In this section I show that the other form of relation apparent between written and implied words in Finnegans Wake--sound similarity--is not crucial in the recovery of implied words. This means that none of the possible connections at work in the production of speech errors are at work in the production or reception of Joyce's substitutions (see also MacMahon, forthcoming). The lack of phonological influence is important; operation on this type of similarity is arguably the most revolutionary in the theorists' terms, since it can produce semantically absurd results. My claim is that the opposite happens in Finnegans Wake; that Joyce's writing involves a reinforcement of the Symbolic order/communicative function of language in its consciously contrived production of meaning.

Inference and recovery In the recovery of intended words on hearing the type of substitution which is related only in sound to the target word, it is likely that the disjunction of meaning triggers a return to the decoding process, using the same process of activation as in normal recovery, though this is probably not the case in examples with synonymous or hyponymous relations. If someone says 'he's not happy in Hawaii' for intended 'he's not happy in Illinois', the mistake may not be obvious to a hearer who does not already know that 'he' is in Illinois; Hawaii is syntactically and semantically appropriate. In the other types of substitution, particularly those based on a high degree of sound similarity alone, the special error mode of retrieval seems likely. This means that where the slip is noticed by a hearer, the mistakenly uttered word is essential as a source of activation in the second process of retrieval. This is not the case with word substitutions in Finnegans Wake. As already discussed, in a reversal of the usual tendency to banalize, Joyce substitutes words in set collocations and familiar phrases. One or more words are distorted in these phrases, but the reader is able to recover the implied word because it is the one she usually hears or reads in that phrase. For example, in the phrase 'mothers-in-load', 'load' has one phoneme in common with implied 'law', and no other relation apart from identity of word class and syllable number. The reader is able to recover 'law' because it is collocationally anticipated by 'mother-in'. This happens with most of the examples from the Joycean sample. The set phrases and collocations are often common expressions such as 'where on earth' (distorted to 'where on dearth'), lines from religious texts such as 'honour thy father'

THE FREUDIAN SLIP REVISITED

315

('hinour thy farmer'), and lines from Irish songs such as 'come back Paddy Riley' ('come back baddy wrily') and 'follow me up to Carlow' ('folly me yap to curlew'). To test this claim about recovery, a questionnaire was devised, including examples of naturally occurring substitutions, Joycean substitutions and substitutions from The Rivals (to be discussed, and listed in Appendix 3). A copy of the questionnaire is given in Appendix 4. The questionnaire was answererd by 20 respondents. In the first section, co-texts were given with a gap at the point of the substitution. Respondents were asked to supply a suitable word for each gap, the prediction being that the collocational phrases from Finnegans Wake would lead to a high proportion of identical guesses from different respondents, while the lack of uttered word as activator in the naturally occurring examples would lead to more failures to identify the intended word. In the second section, respondents were given complete sequences including underlined substitutions from Fromkin's corpus and The Rivals, and asked to supply the word they thought the speaker had really intended. The prediction here for the naturally occurring substitutions was that the presence of the uttered words would allow easier recovery of the intended words. Overall, the test confirmed the above predictions. In the first section, where no written or uttered word was supplied to aid retrieval, respondents on average guessed 12-14 of the 20 substitutions from Finnegans Wake. The two lines from Irish songs given above were included. Not surprisingly, these were guessed uniformly by five Irish respondents and not by others. Excluding these two lines for non-Irish respondents, the proportion of correct guesses was higher, becoming on average 12-14 out of 18. Words not guessed were frequently those implied in lines with other distortions such as 'shook' in 'he shocked me big the hamd'/'he shook me by the hand'. Correctly guessed errors in the naturally occurring examples were very rare. Examples of incorrect guesses were 'Anglo-Saxon male' where the error was 'Anglo-Saxon prostitute' for 'Anglo-Saxon protestant'; 'three, five and eight are the worst years for tantrums' where the error was 'three, five and eight are the worst years for b e e r ' - - ' b e e r ' substituting for intended 'wine'; 'to work with enjoyment' where the error was 'to work with Steve' instead of 'to work with George'. Most people guessed 'cheque' for the space in 'pay by ', but as I have already argued, this error probably involves transposition and elision rather than straightforward substitution. In the second section of the questionnaire, where respondents were given the error and asked to supply the intended word, accurate guesses were made, on average, 11 times out of 15. Where guesses were wrong they often showed near accuracy in supplying a hyponym of the intended word this was not the case in the gap filling exercise. For example, one respondent suggested 'America' or 'New York' to replace 'Italy' in 'the mafia moved into Italy', where the intended was 'the mafia moved into Boston'. Similarly, 'Chinese' in 'there's a small Chinese restaurant' was replaced with 'Italian', where the intended was 'Japanese'. The test showed that implied words in many of Joyce's substitutions are recoverable from the highly predictive surrounding co-texts, while naturally occurring substitutions cannot usually be recovered without the help of the uttered word as a clue. This supports my claim about the coded retrieval of intended words in slips of the tongue, and the conscious non-coded interpretation of Joyce's substitutions. Certain slips lose their uttered meaning in the second retrieval process, because the hearer realizes that the disjunctive uttered meaning cannot have been intended. The hearer is then able to recover intended words automatically, using the same linking relations used in nonerroneous access and retrieval. In the examples described here Joyce reverses the usual tendency to banalize, already identifying his innovation as a non-automatic process. In general language use, familiar words and phrases

316

BARBARA MacMAHON

carry high resting activation levels and so can easily replace less familiar targets, while in Joyce's examples the familiar is replaced by the unusual. Because of this, and unlike the disjunctive type of slip mentioned above, Joyce's collocational substitutions maintain both meanings, and the reader, assuming that both meanings are in fact intended, will try to arrive at a final interpretation which integrates both meanings. Of course it is possible that meanings would be so disparate as to lead to the kind of disjunctive absurdity typical of the type of speech error already discussed, but my final point in relation to Joyce's substitutions is that they are constructed in such a way as to allow for the integration of meanings in one coherent interpretation. An interpretation of 'mothers-inload', for example, could retain the meaning of both 'law' and 'load', drawing on encyclopaedic information to do with mothers-in-law, traditionally represented as irritating and burdensome-the connections with the word 'load' do not need spelling out here. Appropriacy of both uttered and intended meanings is not usually a feature of naturally occurring word substitutions, except where synonymy and hyponymy are activating links. Where appropriacy does occur it is clearly by accident rather than design and, again, would presumably be interpreted differently. Even in cases of synonym and hyponym substitution, there is no integration of meanings in naturally occurring errors. Joyce's innovations, then, add appropriate, non-disjunctive meanings to the texts, not disrupting but reinforcing the Symbolic order. Lacan's pun 'non-du-pere', discussed previously, operates on the same principle as Joyce's distorted collocations. Here the word 'nom' in the more familiar collocation 'nom-du-pere' is substituted by the unexpected word 'non'. The attempt to integrate both meanings in interpretation is facilitated by a knowledge of Lacanian theory. The reader familiar with Lacan's work is able to infer that the word 'non' is appropriate to Lacan's concept of the authoritarian god/father figure who represents the order of the Symbolic which controls and represses (says 'non') through the process of naming ('nora').

Interpretability and disruption I have argued that Joyce's word innovations require inferential rather than coded modes of recovery, and that they are constructed so as to allow for an integration of meanings which reinforces the Symbolic order. Naturally occurring substitutions may not always disrupt meaning. In the case of hyponym and synonym substitution, errors are more likely to go unnoticed by the hearer, and both may result in an appropriateness of meaning in context. This is not the same, however, as the retention of more than one meaning in Joyce's innovations. Other types of naturally occurring substitution do disrupt communicative goals, providing disjunctions such as the one between 'prostitute' and 'protestant', or between 'imminent' and 'eminent' in uttered 'he's an imminent Cambridge professor'. The more disjunctive the meanings, the funnier the substitution, and this is the relationship with wordplay jokes. While Lacan's not very funny pun relies on inferential procedures and encyclopaedic knowledge associated with areas of Freudian and Lacanian theory, and leads to an integration of meanings, funnier wordplay jokes, which may rely either on inferencing processes or automatic retrieval in the recovery of implied words, produce absurd disjunctions of meaning. An example of a play on words which resembles Joyce's distortion of familiar collocations is 'Relax in an Ajax bath."- Unlike Joyce's innovations, however, the play on words here produces one meaning which is entirely inappropriate--it would be the opposite of relaxing to bath in Ajax. The 'mother-in-load' and 'non-du-pere' examples reflect a non-disruptive attempt to reconcile sound and sense levels of lexical organization, while the 'Ajax'/'Radox' example is truly disruptive and mocks any desire to believe in a relation between sound and meaning.

THE F R E U D I A N SLIP REVISITED

317

Word substitution in The Rivals

The sample As in the analysis of word substitution from Finnegans Wake, the predictive points scheme devised on the basis of naturally occurring substitutions was applied to a sample of substitutions taken from The Rivals, given in Appendix 3. The sample was again of 40 substitutions. There are 56 malapropisms/related slips/innovations in The Rivals, of which 40 have obvious intended counterparts. (The number of easily recoverable malapropisms in The Rivals was the lowest in all three sources of substitutions, and so was used as the number of examples selected for all three samples.) Sixteen problematic cases were excluded from the analysis. This selection was again likely to bias results in possibly excluding unrealistic substitutions, since the difficulty in recovering an intended counterpart (one of the reasons for exclusion) may be caused by a lack of activating relation between uttered and target words. However, since targets could not be agreed in these cases, it was impossible to include them in the analysis. Since the purpose of this analysis was to see if primary process operations/activating relations were at work in any of Sheridan's substitutions, the problem was disregarded. An example of a difficult to recover substitution is the word 'Simony' in 'never let her meddle with Simony, or Fluxions, or Paradoxes'. There are different opinions on the source of 'Simony' here. The Oxford edition of The Rivals notes the intended word as 'cyclometry', the Mermaid as either 'cyclometry' or 'ciphering'. 'Cyclometry' is more likely, despite differing in syllable number, because it was conventionally paired with the subsequent 'parallaxes'--the target for 'paradoxes'. Other words were excluded because their semantic similarity raised the question of whether they were supposed to represent malapropisms at all. For example, in the line 'prepare Lydia to meet your son's invocations' there are three opinions on the intended word: according to Mermaid it is a substitution for 'protestations', in the Oxford edition it is not noted as a malapropism, and my own initial interpretation was that the intended word was applications. 'Applications' bears a high degree of sound similarity, but seems only slightly more appropriate in context than the uttered 'invocations'. Both 'invocations' and 'applications' come close to the predictable intended meaning without feeling like the right words. It may be that there is a tendency to regard all Mrs Malaprop's long words as malapropisms, where some of them are just semantically close misuses, the intended and uttered word being the same, but wrong. Other examples of near appropriacy are 'the ingenuity of your appearance' where, according to Mermaid, 'ingenuity' is a substitute for intended 'ingeniousness', but 'ingeniousness' also seems inapproriate, and 'the similitude struck me directly' where 'similitude', according to Mermaid, is not a malapropism, and according to Oxford, is a substitution for 'simile'. A third type of substitution excluded from the analysis was that in which there was suggestion of a representation of blending rather than simple substitution. An example is 'ineffectual' as a possible blend of 'intellectual' and 'unaffected', and 'intricate' as a blend of 'insolent' and 'obstinate'. The points scheme used in the analysis was based on a corpus of likely inadvertent slips rather than errors of encoding. However, Mrs Malaprop's examples are presumably encoding errors; as a character she shows no awareness of her mistakes, suggesting that Sheridan wants us to believe that she believes she is using the words correctly. For this reason, the stressed vowel agreement, claimed by Zwicky to be more frequent in encoded errors, was also taken into account.

318

BARBARA MacMAHON

As previously mentioned, substitutions from The Rivals were also included in the questionnaire along with those from Fromkin's corpus and Finnegans Wake. This was again in order to see whether relations between uttered/written and intended/implied words in these substitutions played a part in recovery.

Activating and linking relations Figure 11 shows the analysis of malapropisms from The Rivals. Of the 40 substitutions, 25 reached a score of 9 or more in the predictive points scheme. All of the plausible malapropisms showed a high degree of sound similarity with intended words-all pairs except one had at least two-thirds of possible single phonemes or clusters in common, while only four showed semantic relations. Of the implausible malapropisms, the failure to reach high enough levels for possible substitution is attributable to the lack of semantic relation. Unlike the Joycean sample, none of the pairs showed reverse banalization, difference of word class or presence of non-content words. Two pairs of words differed in stress pattern. As predicted by Zwicky for encoded errors, a high proportion, 3 l out of 40, showed identity of stressed vowel (this compares with only 13 of the 40 Joycean examples). If a score of 2 points is added for stressed vowel identity, 33 of the 40 substitutions reach a score of 9 or more. This adaptation is added as second score in Fig. 11. This means that a large proportion of the malapropisms in The Rivals are similar to naturally occurring substitutions and, therefore, that these are produced and retrieved using the activation connections used in the production and retrieval of real speech errors. Their tendency to activate on high levels of sound similarity with no semantic relation identifies them as what can be argued to be the most disruptive or 'revolutionary' type of error. Phonemic activation (in addition to the usual word class identity and content words requirement) causes the activation links between words in this type of substitution, resulting in semantic disparity and absurdity. Examples are 'persisted' for 'desisted' in the line 'I thought she had persisted from corresponding with him', and 'preposition' for 'proposition' in the line 'I have since laid Sir Anthony's preposition before her'. Incidentally, after completing this analysis I came across the 'preposition'/'proposition' example as a naturally occurring encoded error in a student essay-a fact which supports the claim that many of Sheridan's substitutions resemble real substitutions.

Recovery, absurdity and disruption Sheridan's plausible substitutions (the 33 with a score of 9 or more) highlight the issue of interpretability and disruptive value. My claim is that the type of innovation used by this writer, which relies on a high degree of sound similarity in producing disjunctive meanings, is more disruptive of the Symbolic order than Joycean innovations. As in naturally occurring substitutions, in Sheridan's malapropisms the repressed babble is not compromised by a forced compatibility with semantic divisions. If they are truly like naturally occurring substitutions, implied/intended words should also be recoverable automatically. In the questionnaire designed to test recoverability, as with the naturally occurring examples, none of the substitutions from The Rivals were guessed correctly in the section where co-texts and gaps were given. In the second section, where Mrs Malaprop's uttered word was also given, 9 out of 10 intended words were guessed correctly. This proportion is even higher than the proportion of correct guesses in the naturally occurring examples (11 out of 15--probably a reflection of the high degree of sound similarity between pairs in The Rivals sample. Recovery here does not depend on the recognition of distorted familiar phrases, and recovered target words cannot usually be reconciled with uttered counterparts in interpretation. The lack of meaning

319

THE FREUD1AN SLIP REVISITED

substitutions from The RivMa

Word

accomodation/recommendation illiterate/obliterate laconically/ironically paradoxes/parallaxes malevolence/becevolence Iocality/Ioq uacity preposition/proposition interceded/intemept ed extirpate/extricate progeny/prodigy conjunctions/injunctions persisted/desisted particle/article geometry/geography controvertible/incontrovertible supercilious/superficial illegible/ineligible intuition/tuition artiticlal/artful hydrostatics/hysterics

111i1

11o.5/I.21 i 1~1at 110/121

ig/l l

10/12 115/151 I ~i~ I nn 10/121i ["~ n .

.

I

-,.~,.,--~'~ =~---

- . m u m u m

10.5/1'2] 11/12 10112 t/d 715/9 'ph< 7/9 d~/d /~i

.

jUmm]IUlB;=.:]lNlm---mUm, a--

.

:

;llmlU =~ llllilm-;lllmm=---gllmlBBUm--

:

I

l m m = a m = . ' -

_

mumuummm, J m m - B m m N m l l iZlmllzlmmllll

m--: --=

= ,----[]mmm==--

i,~,

mlmm-)ram==

_Mm

~)lmNmmrl-,,m

•

W [] B ilZ

-

=~:".,,~*~,['-,,--,,---=

E I~ E [] B []

--,,,,__ -,--- --

[] E

,:-= I ,,,,.1_ i m _-.,-: m,m ' ---

im [] III

substitutions from The Rivals (continued)

Word

physlogno m)qph raseoiogy compilation/appelation oracular/vernacular allegory/alligator caparisons/comparisons antis!ropha/catast rophe pa~icipata/precipit ate felicity/velocity put refactions/pet rificatio ns epitaphs/epithets afffluence/influence paracide/parricide enveloped/develoPed delustons/allusio ns reprehend/comprehend exhort/escort envoy/convoy i precedelproceed Vand~fke/vandal perpendiculars/particulars

10.5/15[ 8/12 9112 8/12 11/12 9/12 7,5/12 10112 f/v 9/1 2 6.5/9 7/9 8/9 7/9 7/9 6/9 4/6 s/z 4/6 5/6 4/6 10/12 d/t

5 4 4 4 I 4 4 4 4 m4&5 I 33 3 3 3 3 2 2 2 2 5&4

/ / / r / / / V" V" / / / ,/ / / / / / x /

,/ / / / ~ V" / /" V" V" / / / / / / / / x /

,/ / / / / / / / V" / / / 4 / / V" / / ,/ /

r,s k t

8.3 8 8.75 8 10.25 8.75 9.6 10.5 9.75 9.5 9 10 9 9 8 8.5 8 9.5 -7 8.5

/ / / / ,/ /

/ / / ~ / / / ,/

10.3 10 10.75 10 12.25 8.75 11.6 10.5 9.75 11.5 9 12 11 11 '10 10.5 8 11.5 -7 10.5

Fig. 11. W o r d substitutions f r o m The Rivals.

compatibility between uttered and target words and the probability of automatic retrieval coincides with the disjunctive type of naturally occurring substitution in which the momentary absurdity in the attempt to interpret a decoded word can also be the basis for wordplay jokes.

320

BARBARA M a c M A H O N

I do not extend my claim of disruption to one of revolution here. Sheridan's realistic malapropisms are made safe in the mouth of Mrs Malaprop. The fact that she is unaware of her mistakes and that they are probably representative of encoded errors suggests that she is undereducated and over-ambitious. Everyone probably has badly encoded word forms in the lexicon, particularly when a word is new or infrequently used, but only people like Mrs Malaprop have the naive confidence to use them with such regularity. All of these points make her an object of ridicule rather than a proponent of targeted disruption or revolution.

Conclusion The predictive points scheme devised on the basis of naturally occurring substitutions shows that relations between uttered and intended words will either have a combination of sound similarity and semantic similarity, or a high level of sound similarity alone. An essential point of similarity needed between words in order to enable a substitution is word class agreement. A very important similarity is that of stress pattern, and both words have to be content words. On the possibility of the repressed desire to babble being the source of slips, the kind of slip with a high degree of sound similarity and no semantic relation could certainly be explained in this way. This should disrupt the Symbolic order in producing absurdity. The erupting repressed from this perspective is the tendency to group and associate words according to sound similarity, as in the pre-signifying, pre-Oedipal, babbling stage. In terms of the disruption of the Symbolic order, a type of error based on semantic relation--that in which antonyms are exchanged----can also be seen as 'revolutionary'. The exchange of antonyms can result in a disparity of meaning similar to, though perhaps not quite as absurd as, sound similarity type substitutions. If antonyms are substituted, the speaker says exactly the opposite of what s/he intended. This would be a triumph for the unconscious, since it shows that the Symbolic order--the order in which semantic distinctions are made--is not a watertight system, and can produce its own errors which disrupt the production of meaning. The psycholinguistic account of naturally occurring word substitutions predicts Freud's slip of 'Signorelli' for 'Botticelli' without needing the unlikely and elaborate associations given by Freud himself. Given the uniformity of the results of the predictive points scheme, it is not necessary to appeal to activating relations other than those shown in the analyses--there is no reason to accept Freud's explanation when this slip is satisfactorily accounted for in the same way as all the others. Semantic links were limited to antonymy, hyponymy and synonymy (fairly weak). If semantic associations other than these had an activating influence, as Freud suggests, results would have been much less uniform, since relations of this type were not considered in the analysis. Joyce's word substitutions resemble the naturally occurring sample only in that there is some degree of sound similarity between written and implied words. Joyce's substitutions are constructed rather than spontaneously produced, and this is reflected in their characteristics. They are constructed on the basis of loose associations such as semantic field connections between several lexical deviations within the same passage. These connections resemble the non-coded, loose and distant kinds of connection used in free association, a technique called into question as a reflection of unconscious processes by Timpanaro. The reverse of banalization also occurs in Joyce's manipulations--something that seems unlikely to happen in slips of the tongue. Joyce uses reworked familiar collocations to introduce extra meanings which are appropriate or conjunctive in exactly the same way as Lacan's pun 'non-du-pere'. If Joyce's word substitutions were really based on the same primary process operations or

THE FREUDIAN SLIP REVISITED

321

activating relations as slips of the tongue, one possibility is that they would be uninterpretable and produce absurdity. This is what often happens in slips based on a high degree of sound similarity alone--as in, for example, the 'imminent professor' example. Alternatively, they could produce contextually appropriate single meanings, as in 'the four blind children' for 'the four deaf children'. In this case, and depending on context, they could be as intelligible as if there were no word substitution at all. In fact, Joyce's innovations show neither of these extremes of interpretability. Unlike Joyce's substitutions, Sheridan's malapropisms from The Rivals exploit the (disruptive) possibility of associating words of similar sound and disparate meaning, thus producing comic absurdity in substitutions such as 'conjunctions' for 'injunctions' in 'Long ago I laid my positive conjunctions on her never to think on the fellow again'. While the sample from The Rivals shows that it is possible for a writer to simulate the operations of the speech production mechanism in producing realistic slips, it is important to note that the disruptive value of the innovations is minimized by attributing them to a character who is the object of ridicule. Interestingly, the portrayal of characters prone to malapropisms continues in the same vein in modem-day soap operas, where slips of the tongue are written into the scripts of middle-aged to elderly, undereducated, pretentious and usually female characters. With regard to text evaluation and location, the findings of these analyses suggest that Finnegans Wake is not greatly different from any other text which invites associative inferences through using words which contain similar sounds. Finnegans Wake may contain a higher proportion of associations than other texts, and associations may be made within one word substitution rather than between two words, but this cannot be claimed to represent a modernist rupture with previous literary devices. Chiming effects such as rhyme and alliteration associate meanings of words in the same way as Joyce's word substitutions--that is, in a non-absurd way, and these effects are abundantly evident in pre-modern literature, particularly literature of the romantic period. This, from Chatterton's Song from Aella, is one of many possible examples: 'My love is dead, Gone to his death bed.' Here, as in Joyce's examples, the rhyme between 'dead' and 'bed' invites a common inference based on associations between sleep and death. Rather than representing a rupture in the literary tradition, Joyce's writing should be seen as a predictable continuation and development of previous poetic trends, in that associations made suggest a compatibility of sound and meaning. The anti-romantic use of sound similarities between words is arguably more disruptive, as in the following lines from Byron's Don Juan: • . . charms in her as natural As sweetness to the flower or salt to Ocean Her zone to Venus or his bow to Cupid. But this last simile is trite and stupid.

Here, the high degree of sound similarity between 'stupid' and 'Cupid' is in sharp contrast to the disparity of meaning between the words, and the effect is to ridicule the typically romantic tendency of Byron's contemporaries to evoke a sense of harmony between sound and meaning. Neither of the two types of sound association-----disruptive or harmonizing--are confined in their use to the area of literature. The 'stupid'/'Cupid' rhyme, for example, is also used in a song by Connie Francis, and sound/meaning harmony effects are common in advertising. The examples given here are fairly arbitrary given the possibilities. My point in including them is to show that Joyce's use of language in Finnegans Wake, apart from not being based on primary process operations, is neither disruptive, nor particularly modernist or uniquely literary.

322

BARBARA MacMAHON NOTES

'This and subsequent points schemes in this paper are designed as a heuristic means of showing differences between naturally occurring and literary substitutions. They are rudimentary, and while they serve the above purpose, they would clearly have to undergo more testing and modification to be produced as sound psycholinguistic models. -'Ajax is the brand name of a scouring powder. Thc line here plays on an advertisement for bath salts: 'Relax in a Radox bath'. REFERENCES AITCHISON, J. 1987 Words in the Mind, An hztroduction to the Mental Lexicon. Blackwell, Oxford. AITCHISON, J. and STRAF, M. 1982 Lexical storage and retrieval: a developing skill? In Cutler, A. (Ed.), Slips r~fthe Tongue and Language Production. Mouton, Amsterdam. ATTRIDGE, D. 1988 Peculiar Language. Methuen, London. CHOMSKY, N. and HALLE, M. 1968 The Sound Pattern of English. Harper and Row, New York. CLEMENTS, G. N. and KEYSER, S. J. 1983 CVPhonology: a Generative Theor), of the Syllable. MIT Press, Cambridge, Massachusetts. DURAND, J. 1990 Generative and Non-linear Phonolog3: Longman, London. ELLIS, A. 1985 The production of spoken words: a cognitive neuropsychological approach. In Ellis, A. (Ed.), Progress in the Psychology of Language. Lawrence Erlbaum, London. FAY, D. and CUTLER, A. 1977 Malapropisms and the structure of the mental lexicon. Linguistic' Inquiry 8, 505-520. FREUD, S. 1901 (1966) The Psychopathology r~fEveo'day Life. Benn, London. FREUD, S. 1905 (1976)Jokes and their Relation to the Unconscious. Penguin, Harmondsworth. FREUD, S. 1915-17 (1973) Introductory Lectures in Psychoanalysis. Penguin, Harmondsworth. FROMKIN, V. A. (Ed.) 1973 Speech Errors as Linguistic" Evidence. Mouton, The Hague. FUDGE, E. 1984 English Word Stress. Allen and Unwin, London. HOCKETT, C. E 1967 Where the tongue slips there slip I. In Fromkin, V. A. (Ed.), Speech Errors as Linguistic Evidence. Mouton, The Hague. HURFORD, J. 1981 Malapropisms, left-to-right listing and lexicalism. Linguistic lnqui~ 12, 419-423. JAKOBSON, R. 1956 Fundamentals of Language. Mouton, The Hague. JOYCE, J. 1939 (1964) Finnegans Wake. Faber and Faber, London. KAHN, D. 1980 Syllable-based Generalization,~ in English Phonology. Garland, New York. KIPARSKY, P. 1970 Metrics and morphophonemics in The Kalevala. In Freeman, D. C. (Ed.), Linguistics and Literary Style. Holt, Rinehart and Winston, New York. KRISTEVA, J. 1974 Revolution in Poetic Language (1984 trans.). Columbia University Press, New York. LACAN, J. 1966 Ecrits, A Selection (1977 trans.). Tavistock, London. MacCABE, C. 1979 James Joyce and the Revolution of the World. Macmillan, London. MacMAHON, B. (lbrthcoming) Interpreting literary word innovtion and language errors. In Green, K. (Ed.), Radical S~listics. MALMBERG, B. 1967 The phonetic basis for syllable divisions. In Lehiste, I. (Ed.), Readings in Acoustic Phonetics. MIT Press, Cambridge, Massachusetts. MATTHEI, W. and ROEPER, T. 1983 Understanding and Producing Speech. Fontana, Suffolk. MORTON, J. 1979 Word recognition. In Morton, J. and Marshall, J. C. (Eds), Psycholinguistics Series, Vol. 2: Structures and Processes. Elek, Cambridge. SHERIDAN, R. B. 1775 (1968) The Rivals. Oxlord University Press, Oxford. SHERIDAN, R. B. 1775 (1979) The Rivals. Mermaid, London. STEMBERGER, J. P. 1985 An interactive activation model of language production. In Ellis, A. (Ed.), Progress in the Psychology c~fLanguage. Lawrence Erlbaum, London. TIMPANARO, S. 1976 The Freudian Slip. New Left Books, London. ZWlCKY, A. 1978 Classical malapropisms. Language Sciences 1, 339-348.

THE FREUDIAN SLIP REVISITED

Appendix 1 Word substitutions from Fromkin's corpus Uttered

Intended

Co-text

prostitute rent dialect belt instamatie cows beer vowels eyes first Hawaii promotion there Steve tree Italy unarmed short

protestant check dialogue collar immediate cars wine values hair last Illinois proposal here George roof Boston unharmed long

Structures

Aspects

Purdue generalisation beer

Peru generation dinner

ivy league Chinese answers high west closes slower world foot neck dissent included blind thumb round brother nose fingers

ivory tower Japanese questions low east opens smaller sun back nose content excluded deaf finger back father head toes

white Anglo-Saxon Protestant pay by rent a book with almost magnificent dialect he got hot under the belt a fairly instamatic thing I thought Westerns were where people ride horses instead of cows three, five and eight are the worst years for beer the native vowels blond eyes take him to the lab first he's not that happy in Illinois a routine promotion he's there tonight to work with Steve a branch falling on the tree the Mafia moved into Italy I urgently request you to release the hostages unarmed my dissertation is too short Don't forget to return Structures the conquest of Purdue phrase structure generalisation Q: When are you going to have the ale? A: With the beer in our academic ivy league There's a small Chinese restaurant there are a lot of answers he's a high grader When were you last on the west coast before the place closes if I write any slower everything under the world I couldn't get it off my foot bridge of the neck rewrite your dissertation to your heart's dissent this must be included the four blind children bang my thumb (with a hammer) the feature round I knew his brother-in-law l should stand on my nose don't burn your fingers

323

324

BARBARA MacMAHON

Appendix 2 Word substitutions from

Finnegans Wake

(Page references for 1964 Faber and Faber edition.) Written

Implied

Page

Co-text

dearth grocery naught along farmer goat quad tears chaste insects commence deed pinchably brunt spatial swarming whipt ignite quaith load poppy goosemother nocturnal shocked yes betrayal curlew folly grunt yap baddy boost him token wore belaboured ham brothel erred off

earth rosary not alone father god quid dears jest incest commit dead possibly brand special warming wept unite with law body grandmother maternal shook us trial Carlow follow grant up Paddy best home taken were beloved him brother heard of

410 411 412 413 413 413 413 413 414 414 414 416 417 417 417 417 417 446 447 448 448 449 449 464 465 466 466 466 466 466 485 485 485 487 487 489 489 489 49 I 491

where on dearth recite my grocery beans whither.., or naught goat along knows how we shall complete that white paper hinour thy farmer goat along nose worth twenty thousand quad this, my tears is even if only in chaste to commence insects with him to commence insects with him sees of the deed as entomate as intimate could pinchably be smolking his spatial brunt smolking a spatial brunt of Hovana cigals swarming of himself in his sunnyroom jadcses whipt let u s . . . all ignite along quaith mothers-in-load tis an ill wind blows no poppy good my nocturnal goosemother nocturnal goosemother he shocked me big the hamd be introduced to yes betrayal buy jury folly me yap to Curlew folly me yap to Curlew grunt unto us I pray folly me yap to Curlew come back baddy wrily his boost friend cum him buddy rowly with me the answer appears to have been token from the writings 1 wore trying on my garden substisuit tearly bleaboured remember ham to me you . . . remind me of my natural saywhen brothel have you erred of Van Homper have you erred off Van Homper

THE FREUDIANSLIP REVISITED

325

Appendix 3 Word substitutions from The Rivals (Page references for Mermaid edition.) Written

Implied

accommodation

recommendation

Page 61

illiterate

obliterate

24

laconically paradoxes

ironically parallaxes

26 27

malevolence locality preposition interceded extirpate progeny

benevolence loquacity proposition intercepted extricate prodigy

29 29 62 62 25 27

conjunctions

injunctions

62

persisted particle geometry

desisted article geography

62 62 27

controvertible supercilious illegible

incontrovertible superficial eligible

25 27 28

intuition artificial hydrostatics physiognomy compilation

tuition artful hysterics physiology appelation

28 29 62 82 87

oracular

vernacular

64

allegory caparisons antistrophe participate felicity putrefactions epitaphs affluence paracide enveloped delusions reprehend exhort envoy precede Vanyke perpendiculars

alligator comparisons catastrophe precipitate velocity petrifications epithets influence parricide developed allusions comprehend escort convoy proceed vandal particulars

68 82 102 104 104 104 64 83 102 102 116 64 105 105 105 117 103

Co-text Your being Sir Anthony's son, Captain, would itself be a sufficient accommodation you will promise to forget this fellow--to illiterate him, I say, quite from your memory Fie, fie, Sir Anthony, you surely speak laconically! I would never let her meddle with Greek, or Hebrew, or Algebra, or Simony, or Fluxions, or Paradoxes, you forfeit my malevolence for ever your being a simpleton shall be no excuse for your locality 1 have since laid Sir Anthony's preposition before her 1 have interceded another letter Now don't attempt to extirpate yourself from the matter I would by no means wish a daughter of mine to be a progeny of learning Long ago 1 laid my positive conjunctions on her never to think on the fellow again I thought she had persisted from corresponding with him She seems resolved to decline every particle that I enjoin her I would have her instructed in geometry, that she might know something of the contagious countries 1 have proof controvertible of it she should have a supercilious knowledge in accounts I hope you will represent her to the Captain as an object not altogether illegible I shall be glad to get her from under my intuition Had she been one of your artificial ones, I should never have trusted her Oht It gives me the hydrostatics to such a degreet His physiognomy is so grammatical I am to thank you for the elegant compilation of 'an old weather-beaten she-dragon'--hey ? Sure if I reprehend anything in this world, it is the use of my oracular tongue She's as headstrong as an allegory on the banks of the Nile No caparisons, Miss, if you pleaser Sir Anthony not to be found to prevent the antistrophe! we should only participate things Why, fly with the utmost felicity to be sure you have no more feeling than one of the Derbyshire putrefactions! a nice derangement of epitaphs my affluence over my niece is small Here's fine suicide, paracide, 'twas he enveloped the affair to me Nay, no delusions to the past If I reprehend anything in this world, it is the use of my oracular tongue Come girls! This gentleman will exhort us Come, Sir, you're our envoy Lead the way, and we'll precede Why, thou barbarous Vandyke He can tell you the perpendiculars

326

BARBARA MacMAHON

Appendix 4 Sample substitutions questionnaire In the following lines, * indicates a missing word. If you think you know what it is, write it in the space below the question. If nothing occurs to you, leave it blank. 1. For

*

2. Tis a

and ever. *

story.

3. white Anglo-Saxon

*

4. Why, fly with the utmost 5. Honour thy 6. pay by

to be sure.

*

*

7. I thought she had 8. No

*

*

*

from corresponding with him.

, Miss, if you please!

9. a book with a most magnificent 10. He

*

m e big the hamd.

11. His

*

friend.

12. he got hot under the 13. Origin of

*

*

14. Sir Anthony not to be found to prevent the 15. Poles

*

!

*

16. a fairly

*

17. Folly me

thing. *

to Curlew.

18. I thought Westerns were where people ride horses instead of 19.

*

unto us, I pray.

20. Nay, no

*

to the past.

21. three, five and eight are the worst years for 22. the native

*

23. Mothers-in24. A 25.

* *

26. blond

*

*

and trap. Crusoe. *

27. That gentleman can tell y o u - - ' t w a s he 28. Garden of

*

*

the affair to me.

*

THE FREUDIAN SLIP REVISITED 29. Son o f a

327

*

30. C o m e b a c k

*

Wrily.

31. She should have a 32. Let there

*

k n o w l e d g e in accounts.

*

33. I hope y o u will represent her to the Captain as an object not altogether 34. W h e r e on

*

35. take him to the lab 36. I have

*

*

another letter.

37. I m m a c u l a t e

*

38. So that she m i g h t

*

the true m e a n i n g o f w h a t she is saying.

39. h e ' s not that h a p p y in 40.

*

buy jury.

41. a routine

*

42. h e ' s

tonight.

*

43. to w o r k with 44.

*

*

Redbreast.

45. a branch falling on the

*

In the following sentences, the underlined work is a mistake. In the space below the sentence, say what you think the intended w o r d was. I. The M a f i a m o v e d into Italy 2. Why, fly with the utmost felicity to be sure 3. I urgently request y o u to release the hostages u n a r m e d 4. m y dissertation is too short 5. 1 thought she had persisted f r o m c o r r e s p o n d i n g with him 6. Q: W h e n are y o u g o i n g to h a v e the ale? A: With the beer 7. No

hydrostatics

Miss, if y o u please

8. in our a c a d e m i c ivy league 9. Sir A n t h o n y not to be f o u n d to prevent the antistrophe 10. Nay, no delusions to the past 11. T h e r e ' s a small Chinese restaurant

328

BARBARA MacMAHON

12. there are a lot of answers 13. That gentleman can tell you--'twas he exploded the affair to me 14. he's the high grader 15. She should have a supercilious knowledge in accounts 16. When were you last on the west coast? 17. I hope you will represent her to the captain as an object not altogether illegible 18. if I write any slower 19. I have interceded another letter 20. the conquest of Purdue 21. So that she might reprehend the true meaning of what she is saying 22. bridge of the neck 23. rewrite your dissertation to your heart's dissent 24. the four blind children 25. before the place closes

View publication stats