COGS Language test Flashcards
Test 3 content from weeks 9-11
1
Q
What is the special place of language in cognitive science?
A
Language is both a cognitive process to understand and an artifact of the human mind that exists independently.
2
Q
What does the field of linguistics study?
A
Linguistics studies language itself as a cognitive system.
3
Q
What does the field of psycholinguistics study?
A
Psycholinguistics studies how humans process and use language.
4
Q
How does language allow the transmission of thoughts from one mind to another?
A
Language uses a shared symbolic system to convert thoughts into transmittable forms like sounds or written symbols, which are then decoded back into thoughts by the receiver.
5
Q
What are the three main modalities of language?
A
The three main modalities of language are speech, written language, and sign languages.
6
Q
What are the five main levels of linguistic analysis?
A
The five main levels of linguistic analysis are phonetics/phonology (speech sounds), morphology (meaningful units), syntax (grammar), semantics (meaning), and pragmatics (language use in communication).
7
Q
What is the difference between prescriptive and descriptive grammar?
A
Prescriptive grammar refers to defined rules of a language, while descriptive grammar refers to patterns of actual language use.
8
Q
What are Hockett’s four key linguistic universals?
A
Hockett’s linguistic universals are semanticity (meaning), arbitrariness (symbols unrelated to meaning), displacement (referring beyond the present), and productivity/generativity (infinite combinations from finite units).
9
Q
What demonstrates the productivity of language?
A
The productivity of language is demonstrated by the ability to create an infinite number of sentences and narratives from a finite number of sounds and words.
10
Q
What is unique about most sentences longer than 8 words?
A
Most sentences longer than 8 words have likely never been said before, and sentences of 11+ words are almost certainly unique.
11
Q
What are the three main challenges in speech perception?
A
The three main challenges in speech perception are segmentation (pauses within words), coarticulation (sounds altered by surrounding phonemes), and individual differences (gender, age, emotion, accent).
12
Q
What are three solutions that help with speech perception?
A
Three solutions that help with speech perception are language experience (shaping perception), context (disambiguating meaning), and visual cues (showing speech is multi-sensory).
13
Q
How does the human ear respond to different frequencies?
A
Human ears are most sensitive to the frequency range that matches human voices, which is around 1-2 kHz.
14
Q
What information does a spectrogram show?
A
A spectrogram shows the energy at different frequencies over time, with vowels appearing as distinct bands called formants and consonants as bursts of noise.
15
Q
What are the three main challenges in speech perception?
A
The three main challenges in speech perception are segmentation, co-articulation, and individual differences in voices.
16
Q
How does context help in speech perception?
A
The brain uses surrounding words to interpret unclear sounds, which is known as top-down processing.
17
Q
What is the McGurk effect, and what does it demonstrate?
A
The McGurk effect shows that lip movements/ visual cues influence what we hear, demonstrating that speech perception is multisensory.
18
Q
What is categorical perception, and how does it help in speech perception?
A
Categorical perception allows the brain to group similar sounds into distinct phonemes, even when they are acoustically similar.
19
Q
How does language experience affect the brain’s ability to discriminate phonemes?
A
Exposure to a specific language shapes the brain’s ability to discriminate phonemes, as infants lose the ability to distinguish between sounds not present in their native language.
20
Q
What is phonemic restoration, and what does it show about the role of context in speech perception?
A
Phonemic restoration is when the brain “fills in” missing phonemes based on context, showing that context plays a crucial role in speech perception.
21
Q
How do infants use statistical learning to identify word boundaries?
A
Infants use statistical learning to identify word boundaries in continuous speech by detecting patterns and regularities in the input they receive.
22
Q
What did Saffran et al. (1996) demonstrate about infant word segmentation?
A
Saffran et al. (1996) demonstrated that 8-month-old infants could identify novel “words” after just 2 minutes of exposure to a continuous stream of syllables.
23
Q
What are some key aspects of language production in children?
A
Key aspects of language production in children include syntax (sentence structure), productivity (making up new words), over-regularization (applying grammatical rules too broadly), and displacement (talking about things not present).
24
Q
What is Hockett’s Linguistic Universal - Semanticity
A
Symbolic units combine to express meaning
25
What is Hockett's Linguistic Universal - Arbitrariness
Symbols bear no inherent relation to meaning
26
What is Hockett's Linguistic Universal - Displacement
Ability to refer to things beyond the here and now
27
What is Hockett's Linguistic Universal - Productivity/Generativity
Finite units combined in infinite ways
28
What makes speech perception a challenging problem?
Speech perception is hard because pauses are within words, sounds change based on surrounding sounds, and people's voices are different.
29
What is the head-turn preference procedure, and how is it used to study speech perception in infants?
The head-turn preference procedure measures how long infants look towards a sound or visual cue to determine their ability to tell sounds apart or their preference for certain stimuli.
30
How do researchers use the high-amplitude sucking paradigm to study speech perception in infants?
Researchers measure changes in infants' sucking rate on a pacifier, with an increase in sucking indicating a response to a new sound.
31
What did Kuhl et al. (2006) demonstrate about the development of phoneme discrimination in infants?
Kuhl et al. (2006) showed that infants' ability to distinguish /r/ and /l/ sounds changes based on their language experience between 6-12 months of age.
32
What is the typical order of acquisition for different aspects of language in children?
Children typically develop speech sounds first, then word formation, sentence structure, meaning, and finally, language use in context.
33
What is over-regularization, and how does it demonstrate children's language development?
Over-regularization is when children apply grammar rules too broadly, showing they understand language patterns even if not yet perfect.
34
What is displacement, and how does it showcase children's language abilities?
Displacement is the ability to talk about things not present, which is a key aspect of language development in children.
35
What is the speech perception challenge of segmentation?
Pauses don't appear between words but in the middle of words.
36
What is the speech perception challenge of Co-articulation?
Phonemes are altered by surrounding phonemes
37
What is the speech perception challenge of Individual difference?
Voices vary based on gender, age, emotion, and accent
38
How do infants acquire the sounds of their native language?
Infants learn language sounds through exposure. They start by babbling many sounds, then narrow it down to only the phonemes of the language(s) they hear.
39
Can adults learn a new language with native-like proficiency?
It's challenging for adults to fully master a new language, especially the phonology. There may be critical periods in childhood for acquiring language innately.
40
What are the key stages of language production in children?
Babbling (6 months), first words (1 year), vocabulary burst (18 months), word combinations (2 years), sentences (3 years), storytelling (4 years), near-adult speech (5 years).
41
How do deaf children of hearing parents communicate without language exposure?
They may develop "home sign", a system of gestures that has some language-like properties such as symbolism and limited displacement.
42
What happened when deaf children with home signs were brought together in Nicaragua?
They created a new sign language (NSL) by combining their home signs. Successive generations of children developed NSL into a full, complex language.
43
What does the emergence of NSL suggest about language acquisition?
Children have an innate capacity to create language through social interaction, even without a complete language model. Their minds are built for language learning.
44
What were some attempts to teach human language to apes?
Washoe (chimpanzee) learned some sign language. Kanzi (bonobo) used lexigrams and understood some speech. Koko (gorilla) and Nim Chimpsky (chimp) also learned limited signing.
45
How did apes' language acquisition differ from human children?
Apes' signing lacked complex grammar, extensive displacement (talking about absent things), and transmission of the language to others - key features of human language.
46
What are honey bees able to communicate with their "waggle dance"?
Honey bees use the angle and intensity of their dance to tell hive mates the direction and distance of food sources. It shows displacement but limited semanticity.
47
How do zebra finches learn their songs?
Male zebra finches learn a unique courtship song from their fathers. Females prefer more complex songs. It's an innate learning capacity but not open-ended communication.
48
What do studies of animal communication reveal about human language?
Human language stands apart in its complex grammar, open-ended expressivity, and cultural transmission. Animals have innate communication capacities but lack the full power of human language.
49
What is the main benefit of having an arbitrary relationship between words and their meanings?
Arbitrariness allows for flexibility, change, and the ability to combine units to create new words in language.
50
What is onomatopoeia?
Onomatopoeia refers to words that sound like what they mean, such as "bang," "crash," "hiss," and "cough."
51
What is sound symbolism?
Sound symbolism is the resemblance between sound and meaning in language, influenced by the physical shape of articulators or properties of sound.
52
What did Ferdinand de Saussure argue about the nature of language?
Ferdinand de Saussure argued for the arbitrary nature of the sign, which became the orthodox view in linguistics in the early 20th century.
53
What are ideophones?
Ideophones are words in some languages (mainly in Asia, Africa, and the Americas) that have a strong resemblance between sound and meaning.
54
What is iconicity in language?
Iconicity is the resemblance-based mapping between aspects of form and meaning in language.
55
How is sound symbolism related to vowel sounds?
High front vowels (e.g., "ee") are associated with small size, while low back vowels (e.g., "oo," "ah") are associated with large size in many languages.
56
How is sound symbolism used in product names?
Sound symbolism is exploited in product names and can influence perception and preference, such as in beer names and logos.
57
What did a study find about the relationship between US presidential candidate names and election outcomes?
The study found that candidates with names having a higher "comfort index" based on sound symbolism were more likely to win elections.
58
What is an example of iconicity in grammar?
In most languages, the subject comes before the object in the default word order, which matches our perceptual experience and understanding of causality.
59
How do signed languages incorporate iconicity?
Signed languages have a greater degree of visual iconicity, with signs often resembling the things they represent, combined with conventionalization.
60
How does iconicity play a role in language development?
Children are sensitive to sound symbolism, which may help them learn words and generalize concepts.
61
What role might sound symbolism and iconicity have played in language evolution?
Sound symbolism and iconicity may have been important in the evolution of language, starting with role-play, mime, and gesture, and gradually becoming more arbitrary over time.
62
Is iconicity universal across all languages and cultures?
No, while iconic associations are based on physical resemblances, they are not the same across all languages and cultures.
63
How does the relationship between pitch and politeness differ between English and Korean?
In English, high pitch is associated with politeness, while in Korean, low pitch is associated with politeness due to cultural differences in the concept of politeness.
64
What factors influence the implicit associations between voice pitch, body size, and gender?
Current research suggests that listener gender, age, and voice gender influence the strength of these associations, possibly linked to cultural conceptions of gender.
65
What is Mātauranga Māori?
Mātauranga Māori refers to Māori ways of knowing that are derived from their ancestors. It is an evolutionary branch of human philosophy.
66
How are Mātauranga Māori and te reo Māori related?
Mātauranga Māori and te reo Māori are interconnected, similar to how family history, community, and kinship are linked. Tikanga, the code of ethical behaviours, is connected to mātauranga and enacted through te reo Māori in various forms like songs and speech.
67
What is the "language instinct" theory?
The "language instinct" theory, proposed by Pinker, suggests that language is independent of cognition. It holds that powerful abstract thinking can occur without language and that babies have concepts before language.
68
What is the mainstream cognitive science view on language and cognition?
The mainstream cognitive science view is that a common language connects community members into an information-sharing network.
69
What did Berlin and Kay propose about colour terms in languages?
Berlin and Kay proposed that all languages have basic colour terms that split the colour spectrum into categories like reds, greens, and blues. They suggested that languages have between 2-11 basic colour categories introduced in a specific sequence.
70
How can cultural exchange influence colour categorization in languages?
Cultural exchange, such as early European contact with Māori, can influence how a language categorizes colour by introducing new colour terms or concepts.
71
What is the "focal colour" in Berlin and Kay's theory?
The "focal colour" is the most representative shade of each colour category, as identified by Berlin and Kay in their research.
72
How many stages of colour category introduction did Berlin and Kay propose?
Berlin and Kay proposed five stages of colour category introduction: light/warm and dark/cool; reds; greens/yellows; blues; and other colours.
73
What is a criticism of Berlin and Kay's theory?
A criticism of Berlin and Kay's theory is that it prioritizes a Western European view of colour categories and doesn't account for languages that don't have a concept of "colour" as understood in Western cultures.
74
How many basic colour terms did Proto-Eastern Polynesian have in the 13th century?
Proto-Eastern Polynesian, the ancestral language of Polynesian cultures, had five basic colour categories in the 13th century.
75
How did te reo Māori adapt to English colour categories after European contact?
After European contact, te reo Māori adopted English loanwords for colours. By the 21st century, it had expanded to match the 11 English colour categories using compound words.
76
How do children typically learn colour names?
Children learn colour names as part of early language development, often using familiar objects like red apples or blue skies as examples.
77
In which of Piaget's stages of cognitive development does colour naming typically occur?
Colour naming typically develops in the preoperational stage of Piaget's cognitive development theory, which occurs between ages 2-7.
78
What are some aspects of early language acquisition?
Early language acquisition includes "motherese" (simpler speech), imitation, babbling, gestures, 1 and 2-word utterances, and simple sentences.
79
How do context and culture influence children's colour learning?
Children learn colours in their environmental context, with cultural objects playing a key role. Different cultures may categorize colours differently, and Western colour categories are often taught in schools.
80
What are some differences between independent and interdependent cultures?
Independent cultures prioritize internal attributes, personal goals, and a self-concept separate from the group. Interdependent cultures prioritize social roles, group goals, and a self-concept embedded in the group, with relationships being crucial.
81
How do Western and non-Western cultures differ in attributing behaviour?
In Western contexts, behaviour is attributed more to individual dispositions, while in non-Western contexts, behaviour is attributed more to situations.
82
What are some characteristics of Western and Māori thinking styles?
Western thinking tends to be more analytical, focusing on objects and avoiding contradictions. Māori thinking tends to be more holistic and relational, focusing on contexts and connections between situations.
83
How do Americans and Japanese perform on tasks involving context?
Americans perform better on tasks requiring ignoring context, while Japanese do better incorporating contextual information.
84
What is alignment in the context of large language models (LLMs)?
Alignment is the process of fine-tuning LLMs after training to ensure they generate polite, appropriate, and helpful responses. This involves supervised learning and reinforcement learning.
85
How is AI transforming the study of animal languages, specifically dolphin communication?
Researchers like Denise Herzing are using AI and deep learning to categorize and analyse dolphin sounds, which could potentially lead to the development of a "dolphin Rosetta Stone" and two-way communication between humans and dolphins.
86
What is the Sapir-Whorf hypothesis?
The Sapir-Whorf hypothesis proposes that language influences (linguistic relativity) or determines (linguistic determinism) thought.
87
What is the difference between linguistic determinism and linguistic relativity?
Linguistic determinism is the strong form of the Sapir-Whorf hypothesis, stating that language determines thought. Linguistic relativity is the weak form, suggesting that language influences thought.
88
What did the emotion semantics study by Jackson et al. (2019) find?
The study analysed emotion words across 2,474 spoken languages and found cultural variations and universal structures in emotion semantics. For example, Persian uses the same word for grief and regret, while Sirkhi uses the same word for grief and anxiety.
89
How do children acquire language?
Children's brains are adapted to acquire language, and culture provides input for children to learn from.
90
What is the relationship between language models (LLMs) and memory?
LLMs can be understood using a simple model of memory, where the chat window acts as short-term working memory and the LLM itself is like long-term memory. Attention plays a role in encoding and retrieving information.
91
What are the main steps involved in language production?
Language production involves conceptualizing the intended message, selecting appropriate words, arranging them grammatically, and physically producing the sounds or signs.
92
What is the common misconception about Eskimo words for snow?
It's a misconception that Eskimo languages have many more words for snow than English. Both languages have several ways to describe snow.
93
How does the movie "Arrival" explore linguistic determinism?
In "Arrival," learning an alien language changes the protagonist's perception of time and causality, demonstrating the concept of linguistic determinism.
94
What did the World Colour Survey reveal about colour categorization across languages?
The World Colour Survey showed that different languages categorize colours differently.
95
How do Russian and English differ in their categorization of blue?
Russian has two distinct words for blue: "siniy" (darker blue) and "goluboy" (lighter blue), while English does not make this distinction.
96
What did Winawer et al. (2007) find in their study on colour discrimination?
Russian speakers were faster at discriminating between shades of blue that crossed the siniy/goluboy boundary compared to English speakers. This effect disappeared with verbal interference, suggesting a linguistic rather than perceptual difference.
97
What role does culture play in language acquisition and cognition?
Culture provides the input for children and AIs to learn language. It shapes cognition through language.
98
What is the relationship between language and thought?
Language supports and reflects thought. Most modern researchers lean towards linguistic relativity, acknowledging that language influences thought but does not solely determine cognitive processes.
99
What are the basic components of a biological brain, and how do they enable learning?
The brain consists of neurons connected by synapses. Learning occurs by adjusting the strength of these synaptic connections.
100
How do neural networks try to mimic the brain's computational principles?
Neural networks use many small, interconnected processing units (similar to neurons) that communicate with each other to perform computations.
101
How can the output of a neural network be interpreted as a probability distribution?
By scaling the activities of the output units to sum to 1, the output layer can be seen as a probability distribution over all possible categories.
102
What is the purpose of a classifier network, and how is it trained?
A classifier network learns to map inputs to categories or classes. It is trained using supervised learning, where each input is paired with the correct output category.
103
What is the main goal of a language processing network?
A language processing network aims to predict the next word in a sequence of words (the prompt).
104
How can language networks be trained in a self-supervised manner?
Language networks can be trained using large text corpora, where the next word in a sequence is used as the target output, eliminating the need for manual labelling.
105
Why do language models like ChatGPT produce different responses to the same prompt?
Language models sample from a probability distribution over all words in the language when generating text, which can lead to different responses for the same prompt.
106
What is the difference between one-hot encoding and distributed word representations (embeddings)?
One-hot encoding represents each word with its own unit, while distributed representations use dense vectors that capture semantic similarities between words.
107
How do distributed word representations help language networks generalize to unseen word sequences?
By representing words with similar meanings using similar vectors (numerical representations), the network can make educated guesses about the meaning of new word combinations based on their similarity to the word combinations it was trained on. This allows the network to generalize its understanding to new, but related, phrases and sentences.
108
What is the main challenge in processing sequences of words in language tasks?
Mapping between sequences of words, such as in machine translation, while handling long-range dependencies between words.
109
How did early language processing networks handle word sequences?
Early networks processed words one at a time, maintaining a running context representation that was updated after each word.
110
How did early language processing networks handle word sequences?
Early networks processed words one at a time, maintaining a running context representation that was updated after each word.
111
What is the attention mechanism, and how does it improve sequence processing in language networks?
Attention enables the network to learn which input words are most relevant for predicting each output word.
112
What are transformers, and how do they relate to modern large language models (LLMs)?
Transformers, developed by Google, extensively use the attention mechanism and form the basis for modern LLMs like ChatGPT and GPT-4.
113
What can transformer-based language models potentially learn from training data?
Transformers can learn complex patterns and relationships in the training data, potentially capturing the grammar of the language(s) they are trained on.
114
Are transformers a perfect model of how the human brain acquires and processes language?
While transformers have achieved impressive results, it is still an open question whether they accurately model how the human brain acquires and processes language.
