The Articulate Mammal (45 page)

The most slippable units, incidentally, are words and phonemes (significant sounds), after these come morphemes (meaningful chunks of word) (Bock 1991).

What (if anything) can we learn from this seemingly strange array of errors? In fact, quite a lot. First, we can suggest what the units of planning are – in other words, the size of chunk we prepare in advance ready for utterance. Second, we can look at the process of word selection. Third, we can make hypotheses as to how words and syntax are planned and assembled.

THE UNIT OF PLANNING

The unit of planning appears to be what is sometimes called a
tone group
, or
phonemic clause
– a short stretch of speech spoken with a single intonation contour. For example:

WHAT TIME IS IT?

DEBORAH BOUGHT SOME SNAILS.

MAX TOOK A BATH.

Note, by the way, that a so-called
phonemic
clause (or tone group) must not be confused with a
syntactic
clause (or sentoid). The two quite often coincide, but do not necessarily do so. For example:

I WANT TO BUY SOME BUNS.

is a single phonemic clause, though is usually regarded as containing two underlying syntactic clauses. In this chapter the word
clause
refers to a phonemic clause, unless otherwise stated.

The main reason for confidently asserting that the tone group is the unit of planning is that slips of the tongue usually occur within a single tone group. For example:

WE’LL GO TO TAXI IN A CHOMSKY (We’ll go to Chomsky in a taxi).

WE FORGED THIS CONGRESS … CONTRACT IN OUR OWN CONGRESSES (We forged this contract in our own congresses).

This strongly suggests that each tone group is planned and executed as a whole. If larger units were prepared, we would expect to find frequent contamination between clauses. As it is, such interference is rare, so much so that Boomer and Laver (1968) regard it as a tongue slip ‘law’ that ‘The target and the origin of a tongue-slip are both located in the same tone-group’ (with ‘law’ to be understood in a statistical rather than in an absolute sense).

On the rare occasions when this ‘law’ is broken, whole
words
can slip into the preceding clause. That is, words can cross clause boundaries, whereas sounds generally do not. For example:

WHEN YOU BUY THE LAUNDRY … (When you take the laundry, please buy me some cigarettes).

WHEN YOU TAKE THE ROSES OUT, ADMIRE … (When you take the garbage out, admire the roses).

EXTINGUISH YOUR SEATBELTS … (Extinguish your cigarettes and fasten your seatbelts).

Compare these with the following sound transpositions and anticipations, which all occur within the same clause:

SHE WROTE ME A YETTER … (letter yesterday).

TWAPTER CHELVE (chapter twelve).

A COP OF CUFFEE (a cup of coffee).

DOG WAS … (Doug was a doctor).

This phenomenon indicates that key words are thought out while the preceding clause is being uttered – whereas the detailed organization of a tone group is probably left till later.

WORD SELECTION

Moving on therefore to word selection, our most direct information comes from a famous ‘tip of the tongue’ (TOT) experiment (Brown and McNeill 1966). Less direct evidence comes from selection errors.

The TOT experiment was a simple one. The researchers assembled a group of students, and read them out definitions of relatively uncommon words. For example, when the ‘target’ word was SEXTANT, the students heard the definition: ‘A navigational instrument used in measuring angular distances, especially the altitude of sun, moon and stars at sea.’ Some of the students recognized the right word immediately. But others went into a TOT (‘tip of the tongue’) state. They felt they were on the verge of getting the word, but not quite there. In this state the researchers asked them to fill in a questionnaire about their mental search. To their surprise, they found that the students could provide quite a lot of information about the elusive missing name. Sometimes the information was semantic, and sometimes it was phonetic. For example, in response to the definition of SEXTANT, several students provided the similar meaning words ASTROLABE, COMPASS and PROTRACTOR. Others remembered that it had two syllables and began with an S, and made guesses such as SECANT, SEXTON or SEXTET.

Semantically, this suggests that similar meaning words are linked together in the mind. We probably activate a number of them, before pinpointing one in particular. When errors occur, we have been insufficiently precise in
locating the exact one needed – as with YESTERDAY instead of ‘tomorrow,’ SHIRT instead of ‘blouse,’ and (another example from the TOT experiment) BARGE, HOUSE-BOAT, JUNK instead of ‘sampan’.

Phonetically, we find a similar picture. People seem to activate several similar-sounding words, before narrowing down the field to one. Malapropisms such as COMPETENCE for ‘confidence’ and NATIVE APE for ‘naked ape’ suggest that people look for words with certain outline characteristics, such as similar initial consonant and number of syllables before they finally select one. Adults give higher priority to the initial consonant than to the number of syllables, so that they often produce malapropisms such as CONDESCENDING for ‘condensing’, and SEGREGATED for ‘serrated’. Children, on the other hand, seem to pay extra attention to the number of syllables, and produce comparatively more malapropisms with a wrong initial consonant, as in ICE CREAM TOILET for ‘ice cream cornet’ (cornet = cone), MISTAKE CAR for ‘estate car’, LEPRECHAUN for ‘unicorn’ (Aitchison and Straf 1981). The situation is not quite as straightforward as suggested above, because a number of other factors play a role in memory, such as the presence of a rhyming suffix, as in PERISCOPE for ‘stethoscope’, PORCUPINE for ‘concubine’. And sometimes a word can get ‘blocked’ by a similar-sounding one: ‘His name begins with an R. I know it’s not Rupert, but that’s the name I keep thinking of.’ The target was
Robert
. As with all psycholinguistic phenomena, a large number of intertwined variables need to be considered (Aitchison 2003a).

The mechanism involved when words are selected is becoming clearer. We probably start with the ‘idea of a word’, then only later fit it to a phonetic form. This is shown by cases when we cannot remember a key word, even though it is clearly ‘there’ in some sense:

HE TOOK A LOT OF … WHAT’S THE WORD I WANT? ER … PERSUASION.

But in fluent speech, selecting the meaning and fitting on the sounds are processes which overlap. People probably begin to find possible phonetic forms while they are still finalizing their choice of word. This is shown by slips in which the word uttered has some meaning and some sound similarity to the target, as in HE WAS IN THE NEXT TRAIN COMPONENT (compartment).

A ‘spreading activation’ or ‘interactive activation’ model is a plausible explanation (Roelofs 1992; Aitchison 2003a). In this model, activation of similar words spreads out and diffuses in a chain reaction. If someone was trying to say MONDAY, all the days of the week would be strongly activated, which would in turn activate the months of the year, though less strongly. Each meaning would stimulate a sound pattern which in turn would rouse further sound patterns. So the ‘idea’ of MONDAY might trigger MONDAY, MAYDAY or MIDDAY, SATURDAY and SUNDAY would trigger each other, and so on. The
task of the speaker is not only to select the word she wants, but to suppress the ones she does not require, though sometimes this process goes wrong:

IT’S AN EXOTIC PLANT, AN ASPIDISTRA NO, AN AMARYLLIS, ER, GLADIOLI, AH – CHAMELEON (CAMELLIA).

Such word-searches are normal. They usually take place fast and privately, but occasionally slowly and openly, as here. The speaker has activated various polysyllabic plant names, homed in on those with a stressed syllable before L, then at the last moment substituted an animal, CHAMELEON, for the shrub ‘camellia’.

Aphasics in particular have problems over suppression. They let through a far wider range of inappropriate words than normal speakers, though there is usually some link with the target, as in DRIVING for ‘swimming’, caused by DRIVING for ‘diving’ and DIVING for ‘swimming’. Sufferers from Alzheimer’s disease have even greater problems (Astell and Harley 1996; Blanken 1998).

The general picture is clear. Words which are relevant both in sound and meaning get more and more excited. Finally, one wins out over the others – though a TOT state may occur if the word has been only partially activated (Harley and Bown 1998).

PLANNING AND ASSEMBLAGE

Let us now consider how the words and syntax are planned and assembled. We can divide this into two main stages: first,
outline planning
, which begins while the previous clause is being uttered. Second,
detailed planning
, which takes place while the clause is actually in progress. Outline planning means the
choice of key words, syntax and intonation pattern, whereas detailed planning involves the fitting together of previously chosen words and syntax.

We know that outline planning includes the choice of intonation pattern, because errors which occur within the tone group (the unit of planning) do not normally disrupt the intonation pattern, as in:

TAKE THE FREEZES OUT OF THE STEAKER.

We are now faced with a tricky and once much disputed question: which comes first, the words or the syntactic pattern? Those who argued that the words come first pointed out quite simply that ‘key’ words determine the choice of syntax, and by ‘key’ words they mean above all nouns, verbs and sometimes adjectives. Clearly, verbs influence the choice of syntax more than the nouns – but the noun may, in some cases, influence the choice of verb.

Those who suggest that the syntax comes first note that when a speaker makes a word selection error, she almost always picks a wrong word belonging to the same word class as the target word. That is, nouns are confused with other nouns, verbs with other verbs, and adjectives with other adjectives. Even aphasic speech, which is often quite garbled, tends to follow this pattern (though exceptions do occur). People say UP instead of ‘down’, JELLY instead of ‘blancmange’, TRANSLATION instead of ‘transformation’. But there is no reason for parts of speech to cling together like this. Why shouldn’t verbs and nouns get confused? The fictional Mrs Malaprop gets her word classes confused much of the time, which is why many of her malapropisms are implausible. She says things such as:

YOU WILL PROMISE TO FORGET THIS FELLOW – TO ILLITERATE HIM, I SAY, QUITE FROM YOUR MEMORY (You will promise … to obliterate him … from your memory).

But in real life, it is extremely unusual to find adjective–verb confusions of the ILLITERATE for ‘obliterate’ type uttered by Mrs Malaprop. Even malapropisms uttered by children generally follow this similar word-class pattern:

YOU TAKE AN ANTELOPE IF YOU SWALLOW POISON (You take an antidote if you swallow poison).

I’M LEARNING TO PLAY THE ELBOW (I’m learning to play the oboe).

According to the ‘syntax first’ supporters, the most likely explanation for this phenomenon is that the syntax has already been chosen, and the words are then slotted in: ‘Unless the syntactic structure is already constructed, word selection would not be constrained to proper word classes’ (Fromkin 1973: 30).

How are we to solve this controversy between the ‘words first’ and ‘syntax first’ supporters? Who is right? Possibly both sides, to some extent, and the controversy seems fairly old-fashioned. We now know that the human brain is capable of complex parallel processing, so possibly, the speaker is thinking up both at the same time. On the one hand, it is unlikely that the key-word advocates are entirely correct. There is no evidence that we assemble
all
the key words, and then bind them together with joining words. On the other hand, it is quite impossible to plan the syntax with no idea of the lexical items which are going to be used. For example, the syntax of:

JOHN CLAIMED TO BE ABLE TO EAT A LIVE FROG.

must depend to some extent on the word CLAIM since other words with a similar meaning take a different construction. We cannot say:

*JOHN ASSERTED TO BE ABLE TO EAT A LIVE FROG.

or

*JOHN DECLARED TO BE ABLE TO EAT A LIVE FROG.