lecture 3 - speech and language

Question 1

language

Answer 1

System of visual/vocal symbols that have meaning to user and recipient

Question 2

use of language

Answer 2

• Communication is one of most important human behaviours - language has evolved through social contacts among our early ancestors
  
  • Speaking and writing are social behaviours - we learn them from other people and use them to communicate with them
  • An effective language system also abides by certain rules
  • Harley 2012 - language can be characterised as a system of visual and/or vocal symbols which have meaning to the user and the recipient
  • Around 6,000 distinct languages in the world - worlds largest language is Chinese and most popular foreign language is English.
    Language allow us to consider complex and abstract issues by encoding them in words and then manipulating the words according to specific rules. These rules are the subject of an area of study called linguistics

Question 3

linguistics

Answer 3

Study of the rules of language

Question 4

psycholinguistics

Answer 4

Study of the role of cognition in language acquisition, production and comprehension including how verbal behaviour develops
How we turn ideas into air and back again

- Study of linguistics involves determining the ‘rules’ of language and the nature and meaning of written and spoken language whereas Psycholinguistics the study of verbal behaviour examines the role of human cognition in language acquisition and comprehension - the interpretation of psych and linguistics 
- Psycholinguists are interested in how we acquire language, how verbal behaviour develops and how we learn to speak from our interactions with others. They are interested in the interaction between the structure and processing of langauge  
- Psycholinguistics = relatively recent but studied since early experimental days eg by Wundt = father of linguistics argued sentence was the most basic element of speech production and comprehension. Speech production involved the transformation of thought process into sequences of speech segments, comprehension was the reverse process. The linguist Hermann Paul argued words, not sentences were the buildings blocks of speech. 
- 1920s and 30s Wundt’s form of psych was challenged by behaviourism that argued psych should concern itself only with observable  behaviour. 1950s psych had renewed interest in the nature of language spurred by linguist Chomsky

Question 5

perception of speech

Answer 5

• Speech is the production of a series of sounds in a continuous stream, punctuated by pauses and modulated by stress and changes in pitch.
  
  • Speech is a more flexible means of communication than writing as the sentences we utter are a string of sounds, some of which are emphasised and some are quickly glided over.
  • We can raise pitch of our voice when uttering some words and lower it when speaking others.
  • We maintain a regular rhythmic pattern of stress.
  • We pause at appropriate times eg between phrases but we do not pause after pronouncing each word
    Speech doesn’t come to us as a series of individual words we must extract the words from a stream of speech.

Question 6

recognition of speech sounds

Answer 6

• Human auditory system performs complex task of enabling us to recognise speech sounds
  
  • Sound system of speech is phonology
  • These sounds vary according to the sounds that precede and follow them, the speakers accent and the stress placed on the syllables in which they occur
  • Phonemes are the elements of speech - the smallest units of sound that contribute to the meaning of a word eg pin has three phonemes p I and n
  • Phonemes are not the same as letters eg ship has 4 letters but three phonemes - sh I and p
  • In linguistics phonemes are indicated by two forward slashes to indicate they are phonemes not letters
    Phonemes is the first step in recognising speech sounds

Question 7

communication model - turning ideas into air and back again

Answer 7

agent produces speech and recipient receives the speech and cognition does speech perception

Question 8

Speech production and perception - making consonants and vowels

Answer 8

Speech is produced by a set of muscles in face, mouth and throat

diagram in notes

Question 9

Changes in air, changes in meaning - Phonemic (or phonetic) contrasts

Answer 9

• Phoneme: smallest unit of speech sound (≠ letters)
  
  • Pin: /p/ + /i/ + /n/
  • Ship: /sh/ + /i/ + /p/
• Group of phonemes: smallest unit of speech that influences meaning
  
  • Bet à Bit
  • Dig à Gig
• Big à Pig
  Not all changes in sound change meaning

Question 10

making consonants (VPM)

Answer 10

voice - whether/when the vocal cords vibrate
Voice
“zip”/”sip”
“bat”/”pat”
“dip”/”tip”

place - where in the vocal tract the constriction takes place
Place
“pat”/”tat”/”cat”
“bot”/”dot”/”got”

manner - how the air moves out of the vocal tract/ what sort of constriction takes place
Manner
“nose”/doze” (nasal/stop)
“dip”/”zip” (stop/fricative)

NB in physical (acoustic) terms, these dimensions are typically continuous, not either/or

Question 11

categorical perception of phonemes

Answer 11

• Pa’ and ‘Ba’ differ in Voice Onset Time (VOT)
• VOT refers to the delay between the start of a speech sound and the onset of the vibration of the vocal cords, i.e. when the lips open relative to when the vocal chords start vibrating
  ‘Pa’ VOT tends to be about 50 ms slower than ‘Ba

Pa’ and ‘Ba’ differ in Voice Onset Time
– when the lips open relative to when the vocal chords start vibrating
‘Pa’ VOT tends to be about 50 ms slower than ‘Ba’
So what do we perceive as we gradually change VOT?
huge variability in actual acoustic changes – due to differences in anatomy and context – perception groups things together into categories

hypothetically we would except a gradual shift from ‘ba’ to ‘pa’

adults should be able to discriminate between each VOT

the actual perception is categorical and there is an abrupt shift typically at about 20-25 ms

Question 12

consequences of categorical perception

Answer 12

We’re good at perceiving changes across category boundaries

We’re bad at perceiving changes within category boundaries

Question 13

production of speech

Answer 13

• Lister and a damson 1970 presented ptps with a series of computer generated sounds consisting of a puff of air followed by an ah. The sound varied only in one way, the amount of time between the puff and the ah. When we speak we make a puff for Pa but not for ba. However even though the computer always produced a puff ptps reported that they heard ba when the delay was short and Pa when it was long. Ptps discriminated between the phonemes /p/ and /b/ strictly according to the delay in voicing. The experiment demonstrates that the auditory system is capable of detecting very subtle differences.
  
  • The fundamental unit of speech, logically and descriptively is the phoneme research suggests that psychologically the fundamental unit is larger eg the two syllables do and Dee each consist of two phonemes. When spoken the same phoneme /d/ is heard at the beginning. However when Lieberman et al 1967 analysed the sounds of the syllables they found at the beginning phonemes were not the same. In fact they could not cut out a section of a tape recording of the two syllables that, would sound like /d/.
  • The results suggest that the fundamental unit of speech consists of groups of phonemes such as syllables.
    The perception of a phoneme is affected by the sounds that follow it (Ganong 1980). Using a computer to synthesise a novel sound that fell between those of the phonemes /g/ and /k/, Ganong reported that when the sound was followed by ift the ptps heard the word gift but when followed by iss they heard kiss. These results sugggest that we recognise speech sounds in pieces larger than individual phonemes.

Question 14

making vowels

Answer 14

Height
Vertical position of tongue in the mouth

Backness
How far back in the mouth the tongue is.
/i/ (“ee”) – front
/u/ (“oo”) – back

Roundedness
Shape of the lips.
Correlated with tongue position in many languages

Question 15

formants

Answer 15

=distinctive frequency components/peaks of the acoustic signal we need to distinguish vowels
F0 – fundamental frequency – frequency of vibration of the vocal chords
= peaks of the acoustic signal
F0 – fundamental frequency – frequency of vibration of the vocal chords
First two formants F1 and F2 are usually sufficient to identify a vowel; each vowel has a unique F1 and F2 difference

Question 16

speech perception

Answer 16

The role of formants (peaks of the acoustic signal) in perception
What else contributes to how we perceive speech?
Is it just the sound?

Question 17

the McGurk Effect

Answer 17

McGurk effect is an example how one sensory modality (vision) can influence another (hearing)
Speech perception is a multi-modal process; not only based on auditory but also visual cues
you hear things that aren’t there

Question 18

Word recognition -the importance of context

Answer 18

Computer generated novel sounds that fell between /g/ and /k/ (Ganong, 1980)
When followed by “ift” -> “gift”
When followed by “iss” -> “kiss”
Speech is full of hesitations, sloppy word productions etc.
47% recognition of isolated words vs 100% recognition within context of original conversation (Pollack & Pickett, 1964)

Question 19

Brief recap - ideas -> air -> ideas

Answer 19

• Producing speech involves movement
  
  • fine control of the vocal tract to shape the sound wave to convey particular meaning
• Perceiving speech involves ‘reconstructing’ the meaning from the sound wave
  
  • range of sources
    ○ key clues (but they’re only ‘clues’) in the wave
    ○ knowledge of language
    ○ Contextual cues
    multimodal sources of information including visual cues

Question 20

procedures to investigate speech perception in infants

Answer 20

1) High amplitude sucking - rate of sucking increases when new sound is detected, then slows down again if sound is repeated

2) Head turn preference -  If infants turn their head and listen for longer (or shorter) to one type of stimulus compared to another, then they must be able to perceptually distinguish them (requires being able to hold head up)

3) Preferential looking - If infants look for longer (or shorter) to one type of stimulus compared to another whilst hearing the names of the stimulus (eg Dad), then they must recognise the name

Question 21

Language learning

Answer 21

• Using the high amplitude sucking procedure, infants as young as 1 month old could tell the difference between consonants ‘ba’ and ‘pa’ (Eimas et al., 1971).
• However, phonemes vary between different languages. E.g. ‘ba’ and ‘pa’ are common to English but are less common in other languages.
  So is this ability to distinguish between phonemes language specific?

Question 22

language learning - categorical perception

Answer 22

Categorical perception
Phonemic categories are language specific
English versus Hindi
/d/ (alveolar) - /D/ (retroflex)
/t/ (alveolar) - /T/ (retroflex)
English versus Japanese
/l/ (alveolar) - /r/ (retroflex)
Infants are sensitive to all categorical boundaries in the first 6 months
By 12 months, they becoming sensitive only to their native categories

Question 23

Segmenting the sound:Finding words in continuous speech

Answer 23

If infants had to rely on being taught individual words, their vocabulary would be a fraction of what it actually is

Statistical properties of language (pattern extraction)
Sequences of sounds that occur within words will occur more frequently and than ones between words
“Thecatsatonthemat”
“Thecatdrankthemilk”
“Thecatkilledthemouse”
Artificial ‘words’
infants listen to: ‘ti-bu-do-pa-bi-ko-mi-la-se-pa-bi-ko-mi-la-se-ti-bu-do-mi-la-se-pa-bi-ko-ti-bu-do…’ for 2 minutes
Afterwards, infants spend longer listening to ‘se-ti-bu’ than to ‘pa-bi-ko’

Question 24

The emergence of languageThe link between production and perception

Answer 24

Babbling
non-linguistic, not really ‘speaking’
emergence of stable, adult-like syllables around 6-8 months
both universal features and individual differences
also shaped to some degree by ambient language
Accuracy of spoken words are related to the number of words in child’s ‘lexicon’ (vocabulary

Production-perception loop
when infants hear (recognises) sounds in adult speech corresponding to their own babbling, they already ‘know’ how to produce those sounds, so their first words are pretty accurate (eg ”papa”, “mama”) -> link between production and perception

Question 25

Language Development Milestones

Answer 25

0-5 months: Smiles, coos (aah, ooh, grr), makes different sounds to interact

6-12 months: recognition of simple words, babbeling (ba ba pa pa pa), production of first words (bye bye, ball)

12–18 months: Understanding ~140-190 words, Production of 75 – 112 words
18-48 months: Vocabulary spurt, basic grammar, increasingly complex sentences, understanding of more abstract concepts
> 48 months: Figurative language, combination of sentences, rhymes, takes listener’s perspective

Question 26

key points

Answer 26

Speaking is movement
Movement has perceptual consequences
changes in movement, changes in air, changes in meaning
Perception
auditory system extracts patterns from sound
Various sources of information, including linguistic knowledge
Language emerges out of ‘basic’ processes of movement and perception

Question 27

errors in speech production

Answer 27

• Speech errors or slips of the tongue are not confined to brain damage. - Fromkin 1988, Dell et al 1997 -
  
  • One obvious error is where the beginnings of words are transposed eg instead of saying dear old queen you might say queer old dean. This is an example of a spoonerism named after the Oxford don William a spooner who was noted for making such mistakes as saying noble tons of soil instead of noble sons of toil.
    Speech errors are interesting as although they are errors they still follow the rules of grammar eg one might confuse nouns in a sentence (would you pass me that cupboard from the pepper) but you would not confuse a noun with a verb (would you cupboard the pass from the pepper). Errors reflect what we had intended to say rather than what we want to say (levelt 1989). Somehow an error occurs between conception and execution.

Question 28

Recognition of words - the importance of context

Answer 28

• The perception of continuous speech involves different mechanisms from those used in the perception of isolated syllables. Because speech is full of hesitations, muffled sounds and sloppy pronunciations, many individual words can be hard to recognise out of context.
  For example, when Pollack and Pickett (1964) isolated individual words from a recording of normal conversations and played them back to other people, those people correctly identified the words only 47 per cent of the time. When they presented the same words in the context of the original conversation, the participants identified and understood almost 100 per cent of them

Question 29

Understanding the meaning of speech

Answer 29

The meaning of a sentence (or of a group of connected sentences that are telling a story) is conveyed by the words that are chosen, the order in which they are combined, the affixes that are attached to the beginnings or ends of the words, the pattern of rhythm and emphasis of the speaker, and knowledge about the world shared by the speaker and the listener.

Question 30

Syntax

Answer 30

• Understanding speech entails following the ‘rules’ of langauge. Words must be familiar and combined in specific ways.
• All languages have syntax or grammar which is a set of rules governing the ways in which words are used to form sentences. They all follow certain principles that linguists call syntactical rules, for combining words to form phrases, clauses or sentences.
  Our understanding of syntax is automatic although learned we are not conscious of this process.

Question 31

word order

Answer 31

• Its important in english.
• In English, the first noun of the sentence is the subject, the second noun is the object and the part in between is usually the verb. This structure is referred to as S–V–O word order (for subject–verb–object) and around 75 per cent of the world’s languages possess this sentence structure (Bernstein and Berko, 1993).
• Other languages, however, have different orders. Japanese, for example, uses the S–O–V order and both Welsh and Arabic use V–S–O.
  The assignation of words into meaningful categories (such as noun, verb, adjective and so on) is called parsing, and parsing involves being able to identify word classes.

Question 32

word class

Answer 32

• Its the grammatical categories such as noun, pronoun, verb and adjective, and words can be classified as function words or content words.
• Function words include determiners, quantifiers, prepositions and words in similar categories: ‘a’, ‘the’, ‘to’, ‘some’, ‘and’, ‘but’, ‘when’ and so on.
• Content words include nouns, verbs and most adjectives and adverbs: ‘apple’, ‘rug’, ‘went’, ‘caught’, ‘heavy’, ‘mysterious’, ‘thoroughly’, ‘sadly’.
  Content words express meaning; function words express the relations between content words and thus are very important syntactical cues.

Question 33

affixes

Answer 33

Affixes are sounds that we add to the beginning (prefixes) or end (suffixes) of words to alter their grammatical function. For example, we add the suffix ‘-ed’ to the end of a regular verb to indicate the past tense (drop/dropped); we add ‘-ing’ to a verb to indicate its use as a noun (sing/singing as in ‘we heard the choir sing’ and ‘the choir’s singing was delightful’); and we add ‘-ly’ to an adjective to indicate its use as an adverb (bright/brightly).
- We are quick to recognise the syntactical function of words with affixes like these. For example, Epstein (1961) presented people with word strings such as the following:
a vap koob desak the citar molent um glox nerf
A vapy koob desaked the citar molently um glox nerfs
The people could more easily remember the second string than the first, even though letters had been added to some of the words. Apparently, the addition of the affixes ‘y’, ‘-ed’ and ‘-ly’ made the words seem more like a sentence and they thus became easier to categorise and recall.

Question 34

semantics

Answer 34

• The meaning of a word – its semantics – provides important cues to the syntax of a sentence

Question 35

Function words and content words

Answer 35

Function words (such as ‘the’, ‘and’, ‘some’) help us determine the syntax of a sentence; content words help us determine its meaning. For example, even with its function words removed the following set of words still makes pretty good sense: ‘man placed wooden ladder tree climbed picked apples’. You can probably fill in the function words yourself and get ‘The man placed the wooden ladder against the tree, climbed it, and picked some apples’

Question 36

prosody

Answer 36

Prosody is a syntactic cue which refers to the use of stress, rhythm and changes in pitch that accompany speech. Prosody can emphasise the syntax of a word or group of words or even serve as the primary source of syntactic information.
For example, in several languages (including English), a declarative sentence can be turned into a question by means of prosody.
- We do this by intonation. In written communication, prosody is emphasised by punctuation marks. For example, a comma indicates a short pause, a full stop indicates a longer one along with a fall in the pitch of voice, and a question mark indicates an upturn in the pitch of voice near the end of the sentence.
- These devices serve as only partial substitutes for the real thing. Because writers cannot rely on the cues provided by prosody, they must be especially careful to see that the syntax of their sentences is conveyed by other cues: word order, word class, function words, affixes and word meaning.

Question 37

relationships between semantics and syntax

Answer 37

• Sentences can be read or heard semantically in more than one way.
• The linguist, Noam Chomsky (1957, 1965), suggested that language can partly be explained by reference to sentence grammar. Although Chomsky’s ideas underwent several revisions, the 1965 version of his theory suggests that there are three grammars. The first – generative grammar – represents the rules by which a speaker’s ideas can be transformed into a final grammatical form. These transformed ideas or thoughts are called deep structures (the second grammar). The final output is the surface grammar or structure which is the end spoken product.
• The deep structure represents the kernel of what the person intended to say. In order to utter a sentence, the brain must transform the deep structure into the appropriate surface structure: the particular form the sentence takes.
• Most psychologists agree that the distinction between surface structure and deep structure is important (Tanenhaus, 1988; Bohannon, 1993; Hulit and Howard, 1993).
  Individuals with a language disorder known as conduction aphasia have difficulty repeating words and phrases, but they can understand them. The deep structure of other people’s speech appears to be retained, but not its surface structure.

Question 38

What is meaning?

Answer 38

• The meaning of a word (its semantics) is defined by particular memories associated with it.
• Memories associated with a word are not stored in the primary speech areas but in other parts of the brain, especially regions of the association cortex.
• Different categories of memories may be stored in particular regions of the brain, but they are linked, so that hearing the word ‘tree’ activates all of them.
• To hear and understand the meaning of a familiar word, we must recognize the sequence of sounds that constitute the word.We must, therefore, have some form of memory store which contains the auditory representations of words. This store forms part of our auditory word-recognition system.
  When we find the auditory entry for the word in our mental lexicon (lexicon means ‘dictionary’), we must be able to access semantic information about this word. The region of the brain responsible for the auditory comprehension of words must somehow communicate with another region (or regions) which allows us to ascribe meaning to what we have just heard.

Question 39

Is there a universal langauge?

Answer 39

• Or are there some features of language that are by most if not all languages? Yes
• For example, all languages have nouns and words to represent states of action or states of being because we all need a way of referring to objects, people and events. Hockett (1960a, b) has suggested that all languages share similar features.
  Zipf’s law of abbreviation (1935/1965) argues that the more frequent words in a language are the shortest, a conclusion that was the result of an analysis of 986 languages from 80 different families. It is called an ‘exceptionless language universal’ (Bentz & Ferrer Cancho, 2016) and enables efficient communication. Pronouns, for example, are commonly used and are short. People learning artificial languages will also choose shorter forms of words for meanings that occur more frequently (Kanwal et al, 2017).

Question 40

gestures

Answer 40

• Gestures can help people think and talk better and help the person you are talking to understand better and aids your own comprehension of what you are saying. This effect may also not depend on the type of gesture you make.
• Gesture has been classified into four types: iconic, metaphoric, deictic and beat (McNeill, 1992).
• The iconic type is a literally representation of an action; a person making a fist and using it to hammer an imaginary nail, for example, saying ‘I hit it repeatedly’.
• A metaphoric gesture might use movement to represent an idea such as waving to signify that a person has lost a lot of money.
• Deictic gestures are direct and involve directing attention to an object or event; for example, pointing to a punnet of strawberries that are wanted on a market stall and saying ‘those strawberries, please’.
• Beat gestures are movements of the hand that are semantically unrelated to the content of the speech being produced or heard
• That we gesture when we speak is not remarkable. What is remarkable is that gestures might improve the comprehension of the person we are speaking to and might even improve our own comprehension. To determine whether this was the case, Dargue et al (2019) undertook a meta-analysis of 83 independent samples of individuals who had participated in gesture and comprehension studies.
• They found that three types of gestures benefitted comprehension; the only one that didn’t was beat gestures, which may be because the gesture is only beneficial when something diÓfficult, rather than something simple, is being described (Gluhareva & Prieto, 2017). Gesturing also helped the gesturer to understand themselves better, a finding consistent with previous studies (Dargue & Sweller, 2018). The effects were much stronger for other-person comprehension, however.
• When the researchers explored the effect of gesture on comprehension in samples of different ages they found an almost universal benefit, except for adolescents (and this may have been because only two studies of adolescents’ gestures were included). However, a study of the effect of inhibiting gestures on speech fluency found no effect of allowing people to gesture on their degree of fluency, or disfluency (Kisa et al, 2022).
• Gesture has been considered the evolutionary forerunner of vocal language (Gentilucci and Corballis, 2006; Kendon, 2016).
  The ability of apes to be taught sign language and, therefore, extract meaning from symbols using motor movement, is thought to provide some evidence for this theory, as is the observation that infants’ first attempts at conveying meaning involve pointing (Kendon, 2016). Ploog (2002) hypothesised that we have two neural systems which mediate vocal behaviour. The first is in the cingulate cortex (and is found in non-humans) and the second is neocortical (seen in humans), which controls contralateral voluntary motor movement

Question 41

Language acquisition by children
Perception of speech sounds by infants

Answer 41

• Language development begins even before birth. Although the sounds that reach a foetus are somewhat muffled, speech sounds can still be heard. And some learning appears to take place prenatally.
  
  • The voice that a foetus hears best and most often is obviously that of its mother. Consequently, a newborn infant prefers its mother’s voice to that of others (DeCasper and Fifer, 1980). DeCasper and Spence (1986) even found that newborn infants preferred hearing their mothers reading a passage they had read aloud several times before their babies were born to hearing them read a passage they had never read before.
  • Homae et al (2011) using near infrared spectroscopy looked at infants’ brain activity while they listened to no language, were read Japanese sentences aloud and then heard nothing again. There was activation seen in the temporal and frontal lobe during the last period but it was not as strong as in the first, suggesting to the authors that the children had retained a ‘memory’ of the previous sound and these brain regions underpin speech perception and could be activated even in children as young as 3 months old.
  • An infant’s auditory system is well-developed. Wertheimer (1961) found that newborns still in the delivery room can turn their heads towards the source of a sound. Babies 2 or 3 weeks of age can discriminate between the sound of a voice and other sounds. By the age of 2 months, babies can tell an angry voice from a pleasant one; an angry voice produces crying, whereas a pleasant one causes smiling and cooing.
  • One device used to determine what sounds a very young infant can perceive is the pacifier nipple, placed in the baby’s mouth. The nipple is connected by a plastic tube to a pressure-sensitive switch that converts the infant’s sucking movements into electrical signals. These signals can be used to turn on auditory stimuli. Each time the baby sucks, a particular sound is presented. If the auditory stimulus is novel, the baby usually begins to suck at a high rate. If the stimulus remains the same, its novelty wears off (habituation occurs) and the rate of sucking decreases. With another new stimulus, the rate of sucking again suddenly increases, unless the baby cannot discriminate the difference. If the stimuli sound the same to the infant, the rate of sucking remains low after the change.
• Using this technique, Eimas et al (1971) found that 1-month-old infants could tell the difference between the sounds of the consonants ‘b’ and ‘p’. Like Lisker and Abramson (1970) in the study discussed earlier, they presented the sounds ‘ba’ and ‘pa’, synthesised by a computer. The infants, like the adult participants in the earlier study, discriminated between speech sounds having voice-onset times that differed by only 0.02 of a second. Even very early during post-natal development, the human auditory system is ready to make very fine discriminations.

Question 42

Infant communication

Answer 42

• Even before infants learn to talk, they display clear intent to communicate. Most attempts at pre-verbal infant communication fall into three categories: rejection, request (for social interaction, for an object or for an action) and comment (Sachs, 1993). Rejection usually involves pushing the unwanted object away and using facial expression and characteristic vocalisations to indicate displeasure.
  
  • A request for social interaction usually involves the use of gestures and vocalisations to attract the caregiver’s attention. A request for an object usually involves reaching and pointing and particular vocalisations.
  • A request for an action (such as the one described above) similarly involves particular sounds and movements. Finally, a comment usually involves pointing out an object or handing it to the carer, accompanied by some vocalisation.
  • Infants babble before they talk. They often engage in serious ‘conversations’ with their carers, taking turns ‘talking’ with them. Infants’ voices are modulated, and the stream of sounds they make sound as though they are using a secret language (Menn and Stoel-Gammon, 1993).
  • At about 1 year of age, a child begins to produce words. The first sounds children use to produce speech appear to be similar across all languages and cultures: the first vowel is usually the soft ‘a’ sound of ‘father’, and the first consonant is a stop consonant produced with the lips – ‘p’ or ‘b’. So the first word is often ‘papa’ or ‘baba’. The next feature to be added is nasality, which converts the consonants ‘p’ or ‘b’ into ‘m’. Thus, the next word is ‘mama’.
  • Mothers and fathers all over the world recognise these sounds as their children’s attempts to address them. The first sounds of a child’s true speech contain the same phonemes that are found in the babbling sounds that the child is already making; thus, speech emerges from pre-speech sounds. While learning words from their carers and from older children, infants often invent their own protowords, unique strings of phonemes that serve word-like functions.
  • The infants use these protowords consistently in particular situations (Menn and Stoel-Gammon, 1993). At around 6 months, infants will begin connecting the things they see and experience (people, objects, actions) with words (e.g. ‘mama’, ‘foot’, ‘eat’). Bergelson and Swingley (2012) found that when mothers asked 6-month-old infants to look at a particular object on a computer screen, the children would spend more time looking at the object the mother named than another object next to it on screen.
• Some sequences are added very late. For example, the ‘str’ of string and the ‘bl’ of blink are difficult for young children to produce; they usually say ‘tring’ and ‘link’, omitting the first consonant. Most children recognise sounds in adult speech before they can produce them (Dehaene-Lambertz and Spelke, 2015)

Question 43

Pre-speech period and the first words

Answer 43

• The first sound that a baby makes is crying, a useful noise for attracting the attention of its carers. At about 1 month of age, infants start making other sounds, including ‘cooing’ (because of the prevalence of the ‘ooing’ sound). Often during this period, babies also make a series of sounds that resemble a half-hearted attempt to mimic the sound of crying (Kaplan and Kaplan, 1970).
  
  • At around 6 months, a baby’s sounds begin to resemble those that occur in speech. Even though their babbling does not contain words – and does not appear to involve attempts to communicate verbally – the sounds that infants make, and the rhythm in which they are articulated, reflect the adult speech that babies hear. Mehler et al (1988) found that 4-day-old infants preferred to hear a voice speaking French, their parents’ native language. This ability to discriminate the sounds and rhythms of the language spoken around them manifested itself in the infants’ own vocalisations very early on.
  • Boysson-Bardies et al (1984) had adult French speakers listen to recordings of the babbling of children from various cultures. The adults could easily distinguish the babbling of 8-month-old French infants from that of babies with different linguistic backgrounds
  • A study by Kuhl et al (1992) provides further evidence of the effect of children’s linguistic environment on their language development. Native speakers learn not to distinguish between slight variations of sounds present in their language. In fact, they do not even hear the differences between them. For example, Japanese contains a sound that comes midway between /l/ and /r/.
• Different native speakers pronounce the sound differently, but all pronunciations are recognised as examples of the same phoneme. When native speakers of Japanese learn English, they have great difficulty distinguishing the sounds /l/ and /r/; for example, ‘right’ and ‘light’ sound to them like the same word. Presumably, the speech sounds that a child hears alter the brain mechanisms responsible for analysing them so that minor variations are not even perceived.

Question 44

Pre-speech period and the first words - when does this alteration occur

Answer 44

Most researchers have supposed that it happens only after children begin to learn the meanings of words, which occurs at around 10–12 months of age.
- The experimenters presented two different vowel sounds, one found in English but not in Swedish and the other found in Swedish but not in English to 6-month-old infants in the US and Sweden. From time to time, they varied the sound slightly. The reactions of the Swedish infants and the American infants were strikingly different. Swedish infants noticed when the English vowel changed but not when the Swedish vowel changed; and American infants did the opposite. In other words, by the age of 6 months, the infants had learned not to pay attention to slight differences in speech sounds of their own language, but they were still able to distinguish slight differences in speech sounds they had never heard. Even though they were too young to understand the meaning of what they heard, the speech of people around them had affected the development of their perceptual mechanisms.
- These results seem to support the native language recognition hypothesis – that infants have the ability to recognise words which belong to their native language (Moon et al, 1993).

Question 45

the general language discrimination hypothesis

Answer 45

suggests that infants are capable of discriminating sentences from any two languages because they can extract sets of properties that these languages possess.
- The evidence above suggests that there is little support for this hypothesis.
- An alternative to these two hypotheses states that newborns are sensitive to prosody and can discriminate between languages on the basis of intonation and rhythm. This is called the rhythm-based language discrimination hypothesis (Nazzi et al, 1998) and there is some support for this from studies in which infants were able to discriminate between English and Japanese but not English and Dutch.
- Interestingly, there is evidence to suggest that the ability to discriminate between phonetic sounds successfully may change with age. Stager and Werker (1997) have reported that 8-month-old infants are capable of discriminating phonetic detail in a task in which 14-month-old infants cannot. The researchers suggest that this represents a reorganisation in the infant’s language processing capacity: it shifts from the processes needed to learn syllables to the process needed to learn words. This is advantageous to the infant as it grows and has to put names to objects, events and situations. Because these activities are computationally complex and involve a huge increase in input, the amount of detail that needs to be processed is, therefore, limited.

Question 46

Two-word stage

Answer 46

By 12 months, children understand around 50 words. At around 18–20 months of age, children start learning language by putting two words together, and their linguistic development takes a leap forward. It is at this stage that linguistic creativity begins.
- As for first sounds, children’s two-word utterances are remarkably consistent across all cultures. Children use words in the same way, regardless of the language their parents speak. Even deaf children who learn sign language from their parents put two words together in the same way as children who can hear (Bellugi and Klima, 1972). And deaf children whose parents do not know sign language invent their own signs and use them in orderly, ‘rule-governed’ ways (Goldin-Meadow and Feldman, 1977). Thus, the grammar of children’s language at the two-word stage appears to be universal.
- For many years, investigators described the speech of young children in terms of adult grammar, but researchers now recognise that children’s speech simply follows different rules. Young children are incapable of forming complex sentences, partly because their vocabulary is small, partly because their short-term ‘working’ memory is limited (they cannot yet encode a long string of words), and partly because their cognitive development has not yet reached a stage at which they can learn complex rules of syntax (Locke, 1993)

Question 47

Acquisition of adult rules of grammar

Answer 47

• The first words that children use tend to be content words: these words are emphasised in adult speech and refer to objects and actions that children can directly observe. As children develop past the two-word stage, they begin to learn and use more and more of the grammatical rules that adults use. The first form of sentence lengthening appears to be the expansion of object nouns into noun phrases (Bloom, 1970). For example, ‘that ball’ becomes ‘that big ball’. Next, verbs are used more frequently, articles are added, prepositional phrases are mastered and sentences become more complex.
  
  • These results involve the use of inflections and function words. It is more difficult for children to add an inflection or function word to their vocabulary than to add a new content word because the rules that govern the use of inflections or function words are more complex than those that govern the use of most content words. In addition, content words usually refer to concrete objects or activities.
• The rules that govern the use of inflections or function words are rarely made explicit. A parent seldom says, ‘When you want to use the past tense, add ‘-ed’ to the verb’, nor would a young child understand such a pronouncement. Instead, children must listen to speech and figure out how to express such concepts as the past tense.

Question 48

Acquisition of adult rules of grammar - - Languages seem to differ significantly in terms of inflection

Answer 48

Of the 6,912 languages spoken in the world, each with its median number of 7,000 speakers, what is common to the languages spoken by the largest groups? Are there particular morphological, geographic, or even social features these languages share that makes them so popular? Lupyan and Dale (2011) examined the structural properties of over 2,000 languages in an attempt to find an answer.
- Their research uncovered universal features of the most widely spoken languages, across the greatest geographical area: these had the simplest inflectional morphology and were the ones which used syntax, rather than modality, to indicate possession and provide evidence. The who-did-what-to-whom structure in these languages relied less on inflection/morphology and more on word order and the architecture of the language. One reason for the popularity of these languages, the authors argue, is that less complex morphology is easier to learn and, therefore, more economic to pass on to the next generation.

Question 49

Acquisition of adult rules of grammar - The most frequently used verbs in most languages are irregular.

Answer 49

. Forming the past tense of such verbs in English does not involve adding ‘-ed’ (e.g. go/went, throw/threw, buy/bought, see/saw). The past tense of such verbs must be learned individually. Because irregular verbs get more use than do regular ones, children learn them first, producing the past tense easily in sentences such as ‘I came’, ‘I fell down’, and ‘she hit me’. Shortly after this period, they discover the regular past tense inflection and expand their vocabulary, producing sentences such as ‘he dropped the ball’. But they also begin to say ‘I comed’, ‘I falled down’, and ‘she hitted me’. Having learned a rule, they apply it to all verbs, including the irregular ones that they were previously using correctly. It takes children several years to learn to use the irregular past tense correctly again
- Children’s rudimentary understanding, or at least recognition, of language and parts of speech seems to begin in the first few months of life. Children learn to assign meaning to words – decide whether they are nouns, verbs, and so on – and use these words to form semi-structured sentences. That is, the child begins to follow the rules of grammar. Grammatical words tend to be phonetically and structurally smaller than lexical words – nouns, verbs, and so on – and the commonest of them (‘in’, ‘a’, ‘and’) are used more frequently in conversation than are the most common lexical words. Does the child, therefore, show a preference for spoken grammatical or lexical words?
- One study exposed 6-month-old infants to spoken lexical and grammatical words and measured their preference for each type of stimulus (Shi and Werker, 2001). The researchers found that the infants showed a preference for the lexical words. The authors suggest that although grammatical words are the most used, lexical words may be more striking and acoustically interesting. Lexical words tend to be longer and have a more complex structure; mothers also tend to use lexical words in isolation (i.e. without the accompanying grammar).
- The preference for lexical words may help the child to give meaning to its world and act as a first essential step to developing more complex communication. It may be that children prefer and use lexical words first and then clamp them on to grammatical structures later

Question 50

Acquisition of meaning

Answer 50

• The simplest explanation of why children use and understand language is that they hear a word spoken at the same time that they see (or hear, or touch) the object to which the word refers. After several such pairings, they add a word to their vocabulary. In fact, children first learn the names of things with which they interact, or things that change (and thus attract their attention). For example, they are quick to learn words like ‘biscuit’ or ‘blanket’ but are slow to learn words like ‘wall’ or ‘window’ (Ross et al, 1986; Pease et al, 1993).

Question 51

Fast mapping

Answer 51

• This quick learning of new, content words has been called fast mapping (Carey and Bartlett, 1978; Markman, 1989). There is some debate over whether fast mapping is specific to language or whether it is generated by other, cognitive processes. For example, if fast mapping is seen only for words, then this would suggest that the process is language based; if fast mapping can extend to other domains, this suggests that the process is underpinned by general cognitive abilities (such as the ability to memorise).
  
  • In two experiments, Markson and Bloom (1997) taught 3–4-year-old children and a group of university undergraduates to learn a word referring to an object (‘kobi’) and a fact about this object. In one experiment, participants were told that this was an object given to the experimenter by her uncle. The participants’ ability to remember and identify the object was tested immediately after learning, one week after or one month after. Although the adults were better at remembering the object and object name than were the children, all children performed comparably well when asked to retrieve the word, identify the object about which facts were presented, and to identify the object given to the experimenter by her uncle. The study suggests that fast mapping may not necessarily be specific to language processing but is made possible by learning and memory mechanisms that are not specific to the language domain.
  • Waxman and Booth (2000) replicated Markson and Bloom’s original finding but suggested that there is a crucial difference between the principles underpinning noun learning and fact learning. They introduced pre-school children to an unfamiliar object and required them to associate it with a noun (‘This is a koba’) or a fact (‘My uncle gave this to me’). The researchers then investigated whether (1) the children were able to map the word or fact correctly by choosing the ‘koba’ or ‘the object the uncle gave to the experimenter’, from a series of 10 familiar objects and (2) the children were able to extend their knowledge of the object by identifying the object from unfamiliar ones. In the second condition, the children were asked, ‘Is this one a koba?’ (word condition), or ‘Is this the one my uncle gave me?’ (fact condition). The children were able to map successfully using the word or the fact. However, there was a difference between the two conditions when children had to extend their knowledge; the children extended the noun to other, similar objects but did not extend the fact.
• In another study, 2–4-year-old children were taught a novel name for an object (‘My cat stepped on this agnew’) and given an arbitrary fact for a second, unfamiliar object (such as metal shelving brackets and Allen keys) (Behrend et al, 2001). The children extended the novel name to more exemplars than they did facts, suggesting that some of the principles underpinning the learning of words and facts are different.

Question 52

Overextension and underextension

Answer 52

• Often a child may commit what are called errors of overextension or underextension. If a child has learned to identify a ball but says ‘ball’ when they see an apple or an orange, or even the moon, we must conclude that they do not know the meaning of ‘ball’. This error is called overextension: the use of a word to denote a larger class of items than is appropriate. If the child uses the word to refer only to the small red plastic ball, the error is called an underextension: the use of a word to denote a smaller class of items than is appropriate.
  
  • Both overextensions and underextensions are normal; a single pairing of a word with the object does not provide enough information for accurate generalisation.
• Carers often correct children’s overextensions. The most effective type of instruction occurs when an adult provides the correct label and points out the features that distinguish the object from the one with which the child has confused it (Chapman et al, 1986). For example, if a child calls a yo-yo a ball, the carer might say, ‘That’s a yo-yo. See? It goes up and down’ (Pease et al, 1993).

Question 53

Is there a language acquisition device?

Answer 53

• There is controversy regarding why children learn to speak and, especially, why they learn to speak grammatically. Chomsky (1965) observed that the recorded speech of adults is not as correct as the dialogue we read in a novel or hear in a play; often it is ungrammatical, hesitating and full of unfinished sentences. In fact, he characterised everyday adult speech as ‘defective’ and ‘degenerate’. If this speech is really what children hear when they learn to speak, it is amazing that they manage to acquire the rules of grammar.
  
  • The view that children learn regular rules from apparently haphazard samples of speech has led many linguists to conclude that the ability to learn language is innate. All a child has to do is to be in the company of speakers of a language. Linguists have proposed that a child’s brain contains a language acquisition device which embodies rules of ‘universal grammar’; because each language expresses these rules in slightly different ways, the child must learn the details, but the basics are already there in the brain (Chomsky, 1965; Lenneberg, 1967; McNeill, 1970).
  • The assertion that an innate language acquisition device guides children’s acquisition of a language is part of a general theory about the cognitive structures responsible for language and its acquisitions (Pinker, 1990). The most important components are as follows:
    *Children who are learning a language make hypotheses about the grammatical rules they need to follow. These hypotheses are confirmed or disconfirmed by the speech that they hear.
    *An innate language acquisition device guides children’s hypothesis formation. Because they have this device, there are certain types of hypothetical rule that they will never entertain and certain types of sentences that they will never utter.
    *The language acquisition device makes reinforcement unnecessary; the device provides the motivation for the child to learn a language.
    *There is a critical period for learning a language. The language acquisition device works best during childhood; after childhood, languages are difficult to learn and almost impossible to master

Question 54

Evaluation of the evidence for a language acquisition device

Answer 54

No investigator regards the first assertion – that children make and test hypotheses about grammatical rules – as tenable. Thus, we cannot simply ask children why they say what they do. Children’s hypothesis testing is a convenient metaphor for the fact that their speech sometimes follows one rule or another.
A more important – and testable – assertion is that the hypothesis testing is guided by the language acquisition device. The most important piece of evidence in favour of this assertion is the discovery of language universals: characteristics that can be found in all languages that linguists have studied. Some of the more important language universals include the existence of noun phrases (‘the quick brown fox . . . ’); verb phrases (‘. . . ate the chicken’); grammatical categories of words such as nouns and adjectives; and syntactical rules that permit the expression of subject–verb–object relations (‘John hit Andy’), plurality (‘two birds’) and possession (‘Rachel’s pen’).
- However, the fact that all languages share certain characteristics does not mean that they are the products of innate brain mechanisms. For example, Hebb et al (1973) observed that language universals may simply reflect realities of the world. When people deal with each other and with nature, their interactions often take the form of an agent acting on an object. Thus, the fact that all languages have ways of expressing these interactions is not surprising.
- Similarly, objects come in slightly different shapes, sizes and colours, so we can expect the need for ways (such as adjectives) to distinguish among them. It is not unreasonable to suppose that the same kinds of linguistic device have been independently invented at different times and in different places by different cultures. After all, archaeologists tell us that similar tools have been invented by different cultures all around the world. People need to cut, hammer, chisel, scrape and wedge things apart, and different cultures have invented similar devices to perform these tasks. We need not conclude that these inventions are products of a ‘tool-making device’ located in the brain. But even if some language universals are dictated by reality, others could indeed be the result of a language acquisition device.
- For example, consider the following sentences, adapted from Pinker (1990):
A1. Bill drove the car into the garage.
A2. Bill drove the car.
B1. Bill put the car into the garage.
B2. Bill put the car

- Someone (such as a child learning a language) who heard sentences A1 and A2 could reasonably infer that sentence B1 could be transformed into sentence B2. But the inference obviously is false; sentence B2 is ungrammatical. The linguistic rules say that sentence A2 is acceptable but that sentence B2 is not very complex; and their complexity is taken as evidence that they must be innate, not learned. Pinker (1990, p. 206) concludes: ‘The solution to the problem [that children do not utter sentence B2] must be that children’s learning mechanisms ultimately do not allow them to make the generalisation.’
- This conclusion rests on the assumption that children use rules similar to the ones that linguists use. How, the reasoning goes, could a child master such complicated rules at such an early stage of cognitive development unless the rules were already wired into the brain? But perhaps the children are not following such complex rules. Perhaps they learn that when you say ‘put’ (something) you must always go on to say where you put something. 
- Linguists do not like rules that deal with particular words, such as put (something) (somewhere); they prefer abstract and general rules that deal with categories: clauses, prepositions, noun phrases and the like. But children learn particular words and their meanings; why should they not also learn that certain words must be followed (or must never be followed) by certain others? Doing so is certainly simpler than learning the complex and subtle rules that linguists have devised. It would seem that both complex and simple rules (or innate or learned ones) could explain the fact that children do not utter sentence B2
- The third assertion is that language acquisition occurs without the need of reinforcement, or even of correction. Brown and Hanlon (1970) recorded dialogue between children and parents and found that adults generally did not show disapproval when the children’s utterances were ungrammatical and approval when they were grammatical. Instead, approval appeared to be contingent on the truth or accuracy of the children’s statements. If there is no differential reinforcement, how can we explain the fact that children eventually learn to speak grammatically? It is undoubtedly true that adults rarely say, ‘Good, you said that correctly’, or ‘No, you said that wrongly’. However, adults do distinguish between grammatical and ungrammatical speech of children. 
- A study by Bohannon and Stanowicz (1988) found that adults are likely to repeat children’s grammatically correct sentences verbatim but to correct ungrammatical sentences. For example, if a child says, ‘That be monkey’, an adult would say, ‘That is a monkey’. Adults were also more likely to ask for clarifications of ungrammatical sentences. Thus, adults do tend to provide the information children need to correct their faulty speech
- Chomsky’s assertion about the defectiveness and degeneracy of adult speech is not strictly true, at least as far as it applies to what children hear. 
- In fact, according to Newport et al (1977), almost all the speech that a young child hears (at least, in industrialised English-speaking societies) is grammatically correct. If that is so, why should we hypothesise that a language acquisition device exists? Because, say some researchers, not all children are exposed to child-directed speech (i.e. speech which adults use specifically when communicating with children). 
- ‘In some societies people tacitly assume that children aren’t worth speaking to and don’t have anything to say that is worth listening to. Such children learn to speak by overhearing streams of adult-to-adult speech’ (Pinker, 1990, p. 218).
- Pinker’s statement is very strong; it says that children in some cultures have no speech directed towards them until they have mastered the language. It implies that the children’s mothers do not talk to them and ignores the fact that older children may not be quite so choosy about their conversational partners. To conclude that such an extreme statement is true would require extensive observation and documentation of child-rearing practices in other cultures.
- In fact, children do not learn a language that they simply overhear. Bonvillian et al (1976) studied children of deaf parents whose only exposure to spoken language was through television or radio. This exposure was not enough; although the children could hear and did watch television and listen to the radio, they did not learn to speak English. It takes more than ‘overhearing streams of adult-to-adult speech’ to learn a language.
- The final assertion – that the language acquisition device works best during childhood – has received the most experimental support. For example, Newport and Supalla (1987) studied the ability of people who were deaf from birth to use sign language. They found that the earlier the training began, the better the person was able to communicate.

Question 55

bilingualism

Answer 55

• If an individual can meet the communication demands of the self or the individual’s culture in two or more languages, they are considered bilingual.
• Bilingualism is described as ‘simultaneous’ when two or more languages develop in childhood simultaneously, spontaneously and naturally, and ‘successive’ when a second (and third) language is learned after the first, such as learning a second language during puberty (Romaine, 1989). That said, there is no generally accepted operational definition of, or an agreed set of measures which determine, bilingualism, although a recent study has attempted just this (De Cat et al, 2022)
• Until relatively recently, it was thought that bilingualism was detrimental to cognitive performance such as lexical decision time: bilinguals were slower, committed more errors when naming pictures and had more tip-of-the tongue experiences. None of these is very much evident in conversation, however (Bialystok and Craik, 2010).
• Executive functions are thought to be much better performed by bilingual speakers; for example, bilinguals sort cards by colour and shape and complete the Stroop task better (Bialystok and Martin, 2004; Bialystok et al, 2008) although the evidence is not completely consistent (Paap, 2014). Paap, for example, notes that these studies are characterised by small samples and that direct replications are rare
• Bilinguals appear to perceive events and people less egocentrically than do monolinguals, a finding that is correlated with better performance on a theory of mind task (Rubio-Fernández and Glucksberg, 2012).
• The brain regions responsible for control of executive functions also overlap with those areas responsible for the control of language in bilingual speakers (Pliatsikas and Luk, 2016). One reason why better executive function might be observed in bilinguals is that conflict resolution between two languages is a feature of executive function, and that the constant exercise of this conflict resolution enhances general executive function (Bialystok and Craik, 2010).
• Bilingualism may even be a cognitive reserve, protecting against cognitive decline. One study found that the age of onset of dementia is four years later for bilingual than monolingual speakers (Bialystok et al, 2007). Monolingual English and bilingual speakers of Spanish-English – whether adults or children – appear to recruit different brain regions when they listen to English syllables (Archila-Suerte et al, 2013). The former recruits the superior temporal gyrus in early and late childhood when listening to native speech whereas the bilingual group recruits the same area in early childhood and executive areas (bilateral middle frontal gyrus) in late childhood when listening to non-native speech.
• Although there is considerable evidence of a bilingual cognitive advantage, that is, speaking two (or more) languages confers cognitive benefits (de Bruin et al, 2021), meta-analyses suggest that the advantage is weak, or the data are inconsistent. The reasoning behind superior executive function in bilinguals, for example, proposed that because bilinguals had to suppress one language while speaking another and could switch between languages easily, they had better executive skill as a result. Although studies have found evidence of bilingual advantage – even in meta-analyses – when publication bias is removed from these analyses, the advantage becomes weak or inconsistent (Lehtonen et al, 2018).
• Part of the problem is one of definition. A person who only speaks one language is usually described as monolingual, but it is likely that they will have knowledge of other languages, and some have learned a language but may not be completely proficient in it.
  There are differences reported between bilinguals and monolinguals in terms of fMRI activation and cortical volume and thickness (de Bruin et al, 2021).

Question 56

Can other primates acquire language?

Answer 56

• In the 1960s, Beatrice and Roger Gardner of the University of Nevada began Project Washoe (Gardner and Gardner, 1969, 1978), a remarkably successful attempt to teach sign language to a female chimpanzee named Washoe. Previous attempts to teach chimps to learn and use human language focused on speech (Hayes, 1952). These attempts failed because, as we noted above, chimps lack the control of tongue, lips, palate and vocal cords that humans have and thus cannot produce the variety of complex sounds that characterise human speech.
• Gardner and Gardner realised this limitation and decided to attempt to teach Washoe a manual language, that is, one that makes use of hand movements. Chimps’ hand and finger dexterity is excellent, so the only limitations in their ability would be cognitive ones. The manual language the Gardner’s chose was based on ASL, the American sign language used by deaf people. This is a true language, containing function words and content words and having regular grammatical rules.
• Washoe was 1-year-old when she began learning sign language; by the time she was 4, she had a vocabulary of over 130 signs. Like children, she used single signs at first; then, she began to produce two-word sentences such as ‘Washoe sorry’, ‘gimme flower’, ‘more fruit’ and ‘Roger tickle’. Sometimes, she strung three or more words together, using the concept of agent and object: ‘You tickle me’. She asked and answered questions, apologised, made assertions; in short, she did the kinds of things that children would do while learning to talk. She showed overextensions and underextensions, just as human children do. Occasionally, she even made correct generalisations by herself.
• After learning the sign for the verb ‘open’ (as in open box, open cupboard), she used it to say open faucet, when requesting a drink. She made signs to herself when she was alone and used them to ‘talk’ to cats and dogs, just as children will do. Although it is difficult to compare her progress with that of human children (the fairest comparison would be with that of deaf children learning to sign), humans clearly learn language much more readily than Washoe did.
  Inspired by Project Washoe’s success (Washoe died in 2007), several other investigators have taught primate species to use sign language. For example, Patterson began to teach a gorilla (Patterson and Linden, 1981) and Miles (1983) began to teach an orangutan. Washoe’s training started relatively late in her life, and her trainers were not, at the beginning of the project, fluent in sign language. Other chimpanzees, raised from birth by humans who are native speakers of ASL, have begun to use signs when they are 3 months old (Gardner and Gardner, 1975)

Question 57

Many psychologists and linguists have questioned whether the behaviour of these animals can really be classified as verbal behaviour

Answer 57

For example, Terrace et al (1979) argue that the apes simply learned to imitate the gestures made by their trainers and that sequences of signs such as, ‘please milk please me like drink apple bottle’ (produced by a young gorilla) are nothing like the sequences that human children produce. Others have challenged these criticisms (Fouts, 1983; Miles, 1983; Stokoe, 1983), blaming much of the controversy on the method that Terrace and his colleagues used to train their chimpanzee.
- Chimpanzees can, apparently, use symbols to represent real objects and can manipulate these symbols logically (Premack, 1976). These abilities are two of the most powerful features of language. For Premack’s chimpanzees, a blue plastic triangle means ‘apple’. If the chimpanzees are given a blue plastic triangle and asked to choose the appropriate symbols denoting its colour and shape, they choose the ones that signify ‘red’ and ‘round’, not ‘blue’ and ‘triangular’. Thus, the blue triangle is not simply a token the animals can use to obtain apples; it represents an apple for them, just as the word apple represents it for us.
- Even though humans are the only primates who can pronounce words, several other species can recognise them. Savage-Rumbaugh (1990) and Savage-Rumbaugh et al (1998) taught Kanzi, a pygmy chimpanzee, to communicate with humans by pressing buttons that contained symbols for words.
Kanzi’s human companions talked with him, and he learned to understand them. Although the structure of his vocal apparatus prevented him from responding vocally, he often tried to do so. During a three-month period, Savage-Rumbaugh and her colleagues tested Kanzi with 310 sentences, such as ‘Put a toothbrush in the lemonade’. Three hundred and two of these had never been heard by the chimpanzee before. Only situations in which Kanzi could not have been guided by non-verbal cues from the human companions were counted; often, Kanzi’s back was to the speaker. He responded correctly 298 times.
- The most successful attempts at teaching a language to other primates are those in which the animal and the trainer have established a close relationship in which they can successfully communicate non-verbally by means of facial expressions, movements and gestures. Such interactions naturally lead to attempts at communication; and if signs (or spoken words) serve to make communication easier and more effective, they will most readily be learned.