Infant Language Acquisition: Key Stages and Processes

Posted on Apr 1, 2025 in English Studies. Language, Literature, and Culture

UNIT 2

U2B1 Questions

1. Usage-Based Accounts: Definition and Proposals

Usage-Based Accounts are theories proposed by Tomasello. According to this account, language is an inventory of constructions, each with a specific function in communication. One key concept in usage-based accounts is that of construction learning. These theories propose that language learning begins with the recognition of patterns in linguistic input.

Instead of focusing solely on individual words or grammatical rules, usage-based approaches emphasize the learning of abstract constructions, which are patterns of linguistic form and meaning that recur across different contexts.

Unlike traditional theories such as Chomsky’s Universal Grammar, which suggests that humans possess innate linguistic knowledge, usage-based approaches emphasize the importance of experience and interaction in language learning, stating that learning language construction and their meaning does not require innate knowledge of linguistic principles and rules.

More in accordance with Skinner’s behaviorism, infants learn from the environment; it is their experience which teaches them language. As infants are exposed to language, they gradually build up a repertoire of linguistic constructions and their meanings based on the frequency and salience of these patterns. According to usage-based accounts, this process is facilitated by cognitive mechanisms such as pattern recognition, categorization, and generalization and social abilities.

Furthermore, usage-based theories emphasize the role of social interaction in language learning. Through interactions, for instance with caregivers and other speakers, infants receive feedback on their language use and are provided with opportunities to practice and refine their linguistic skills.

2. Cooing and Babbling: Onset and Characteristics

Cooing and babbling are two important preverbal stages that precede the start of verbal communication in infants. These stages mark significant milestones in speech production and provide insights into the emergence of speech sounds and language production.

Cooing typically occurs between 1 to 6 months and involves the production of vowel-like sounds such as “ooh” and “aah.” These sounds are often characterized by their melodious and repetitive nature.

Babbling emerges around 7-8 months of age and is characterized by the production of repetitive syllables composed of consonant-vowel combinations, such as “ba-ba” or “da-da.” Unlike cooing, babbling involves more complex vocalizations that resemble the phonetic structures of the language(s) to which the infant is exposed. Babbling is a crucial stage in language development, allowing infants to experiment with different speech sounds and patterns.

The beginning of the verbal stage, where infants begin to produce meaningful words, typically occurs around 12 months of age onwards. This transition from prelinguistic vocalizations to meaningful speech marks a significant milestone in language development. It is influenced by various factors, including the infant’s exposure to language input, cognitive development, and social interactions.

3. Onset of Exposure to Linguistic Input

Exposure to linguistic input begins even before birth, as fetuses are capable of hearing sounds from the external environment while still in the womb. Our exposure to linguistic input begins during the third trimester of gestation, which is the moment in which the peripheral auditory system becomes operative. However, this exposure is very limited as inside the womb (due to the liquid around) fetuses are only able to hear low frequency sounds which are below 300 Hz (for instance, vowels). Additionally, even if a big part of information is filtered out, a good part of prosody is preserved. The fetus is able to hear rhythm and intonation; however, that output is kind of muffled speech. The only prosodic information that is preserved is that of the mother.

4. Newborns’ Bias for Speech: Elaboration

Newborn infants demonstrate a clear preference for speech sounds over non-speech sounds, suggesting an innate bias for language stimuli from birth. The study of Vouloumanos & Werker (2007a) compared neonates’ attention/responsiveness to speech stimuli and sine-wave analogs stimuli via High Amplitude Sucking (HAS). The study showed that neonates sucked significantly more when listening to speech than to non-speech in the second block, which demonstrated that neonates have a preference for speech over sine-wave or other types of synthetic sound such as white noise (Butterfield & Siperstein,1970), filtered speech (Spence & DeCasper, 1987), and backward speech ((Peña et al., 2003).

5. Specificity of Newborns’ Bias for Speech

The bias for speech in newborns is highly specific, as infants demonstrate a preference for the sounds and patterns of their species and native language(s) over unfamiliar or non-native speech stimuli. This specificity suggests that infants are not simply attracted to any speech-like sounds but are instead tuned to the phonetic properties of the languages to which they are constantly exposed.

A study by Vouloumanos , Hauser, Werker, & Martin (2010) investigated how specific this bias was. For that purpose they presented neonates and 3-month-old infants with nonsense speech and rhesus monkey vocalizations (as it is the most similar to human speech). The results showed that neonates showed no preference for speech over rhesus vocalizations but showed a preference for both these sounds over synthetic sounds. However, 3-month-old children did show a preference for human speech over rhesus vocalizations. These results suggest that even though initial biases minimally include speech and monkey vocalizations over synthetic sound, these listening preferences are sharpened over 3 months, showing a species-specific preference for speech.

6. High Amplitude Sucking (HAS) and NIRS Techniques

The High Amplitude Sucking Technique is a method used in infant research to investigate early auditory discrimination and learning. This method was developed by Eimas (1971). In this technique, infants are provided with a pacifier connected to a pressure transducer, which measures the amplitude and frequency of their sucking behavior. Firstly, a baseline needs to be established to determine the threshold for high-amplitude sucks; for that purpose, the infants are presented with 1-2 minutes of silence. Infants are then presented with auditory stimuli, such as speech sounds or tones, and their sucking responses are recorded. Changes in sucking rate or intensity can indicate infants’ discrimination of different auditory stimuli or their preference for specific sounds. The way to know which condition do infants prefer is by comparing the mariners of HAS in both types of blocks, the one with a greater number of HAS is preferred.

Near Infrared Spectroscopy is a neuroimaging technique used to measure changes in the oxidation of the blood in a region of the brain (metabolic response). NIRS relies on the principle that oxygenated and deoxygenated hemoglobin exhibit different absorption spectra in the near-infrared range of light. NIRS systems can estimate changes in cerebral blood flow and oxygenation, which are indicative of neural activation in underlying brain regions. With this technique newborns can be tested even when they are asleep.

U2B2 Questions

1. Perceptual Reorganization or Narrowing

Perceptual reorganization refers to the phenomenon in which infants initially demonstrate the ability to discriminate between a wide range of sounds from different languages, but over time, as they grow, with experience, they become more attuned to the specific sounds present in their native languages. This process of diminishing abilities to discriminate usually happens during the 10 month onwards, when infants become more specialized in processing the phonetic distinctions that are relevant to their environment.

2. Discrimination After Perceptual Reorganization

This statement is true. The study of Werker & Tees (1984) compared two groups of infants, 8-10 month infants with 10-12 months. This study showed that babies at 10-12, after perceptual reorganization, were not able to discriminate non-native phonetic contrast anymore. However, the ability to discriminate is maintained when the non-native contrast is very different from the native one (as shown in the study by Best, McRoberts & Sithole (1987)).

Another study investigated this but with native phonetic contrast. This study showed that in their native language the discrimination between phonetic contrast was enhanced.

3. How Babies Perceive Sounds

When we are born we are able to hear strings of words, but not as a continuum; we are able to discriminate between sounds. Infants perceive sounds as members of categories.

(I was not sure about what we needed to answer to this question)

4. Mechanisms Employed in Sound Perception

Babies employ two complementary mechanisms in sound perception: distributional learning and protolexicon.

Infants are sensitive to the distributional properties of speech sounds, such as their frequency of occurrence and co-occurrence patterns within words and syllables. By tracking the statistical regularities present in their linguistic environment, infants can infer underlying phonological structures and phonetic categories. Distributional learning allows infants to develop sensitivity to the phonetic contrasts and syllable structures relevant to their language(s). This mechanism works better in consonants, to locate boundaries between them.

Regarding vowels, the protolexicon mechanism helps mostly. At 6 months, babies already seem to have a protolexicon. At this age babies do not know what a word is (meaning, they do not have a lexical representation of it), but they are able to detect the sounds the word contains. Infants do not perceive the vowels in isolation (as opposed with consonants), they use this phonological information to cluster all words that share that same vowel sound.

U2B3 Questions

1. Onset of Word Segmentation from Fluent Speech

By the time they are 7.5 months infants are able to the detect the acoustic form of the words (this does not necessarily mean that they know the meaning of the words). This was concluded in the study by Jusczyk & Aslin (1995). In this study they use the Headturn Preference Procedure, in which a light is blinking to get the infants attention, this way it can also be observed the looking time of infants (how long does the baby spend staring at the light). The first step in the study was to have the babies familiarize themselves with some target words (dog and feet), and after that, they listened to 4 passages (2 of them containing the target words, and the other 2 containing different words). The results show that 7.5 month infants were able to segment the words they have been previously familiarized with in the passages.

2. Cues Employed to Segment Words

One of the cues that infants employ to segment words is statistical learning. This is related to Transitional Probabilities (TPs). This means the chance there is for a particular syllable to occur with another particular syllable. For TPs we need two computations, the relative frequency of a given syllable, and the frequency of co-occurrence of that syllable with another one). The study by Saffran, Aslin & Newport (1996) demonstrated that 8 month babies can extract information by tracking the TPs of the syllables in the input, and that they use this information to locate words. This was shown by the fact that babies were able to differentiate words from part-words. So, those statistical learning cues are there from the beginning, and babies are able to extract regularities on TPs.

Another study by Johnson & Tyler’s (2010) investigated statistical learning in more complex contexts with infants of 5.5 and 8 months. The results showed that 5.5 and 8 month infants could recognize part words in natural stimuli when the word length was uniform. However, these infants could not track down TPs when the word length is mixed.

Another cue used by babies to segment words is prosodic information. Different languages have different rhythm classes: stressed-timed languages, syllable-timed languages, and mora-timed languages. Different studies showed that babies used rhythmic classes to extract words from speech. The study by Cutler & Carter (1987) showed that 6-7.5 month old babies learning stressed-timed languages used this rhythm information to segment words, as they expected strong syllables to be the beginnings of words, rather than weak ones. Jucszyk conducted a series of 15 experiments to test if they use this cue to segment words. One of his studies showed that these infants were able to segment words with a strong+weak syllable sequence (they segment words with the predominant rhythmic class in their native language).

3. Integration of Multiple Cues for Segmenting Words

Mattys, White & Melhorn (2005) suggested that babies integrated the multiple cues in a hierarchical way. The suggested hierarchy for babies was as follows: statistical information, word-stress cues, phonotactic cues, familiar words cues, sentential context cues. Also, the order of these cues changes over the span of our lifetime, as adults we tend to use sentential context cues more, using all the information available we have. Nowadays, however, it is believed that the segmentation cues are interactive with each other all the time.

U2B4 Questions

1. Onset of Grammar Acquisition

The acquisition of grammar begins earlier than traditionally believed, as by the time infants start combining words, when they are 18-24 months old, their productions tend to already follow the word-order from their native languages. Babies are able to produce two word phrases such as ‘the house’, following the determiner+noun word order.

A study by Benavides-Varela &Gervain (2017) showed that infants can detect changes in word-order as shown by the NIRS; they show higher activation in blocks with change in word-order than in blocks with the normal word-order of their native language. Also, the study by Gervian, Macagno, Cogoi, Peña & Mehler (2008) shows that neonates are able to detect a simple rule of language (reduplication, when it is adjacent).

2. Role of Functors in Grammar Acquisition

Functors help babies in their acquisition of word-order; they seem to work as anchoring points. Babies track functors to learn grammar, while an artificial language with functors is learnable, an artificial language without them is unlearnable. Infants are sensitive to functors since birth, even if they cannot produce them.

3. Pivotal Role of Functors in Grammar Acquisition

We know that functors play a pivotal role in acquisition of grammar as they help infants to acquire the word-order of their native languages. An experiment by Gervain, Newport, Mazuka, Horne & Mehler with Italian (SVO, functor initial language) and Japanese (SOV, functor final language) infants of 8 month old. The results showed that Italian infants showed a preference for word-initial, while Japanese infants showed a preference for word faunal. This showed a tendency that infants have a strong preference towards their native language word order.

4. Frequency Based Bootstrapping Hypothesis

The Frequency Based Bootstrapping Hypothesis claims that the differing properties of functors and content words, and specifically their frequency of occurrence in the input, could allow infants to segment speech and discover the word order of the native language early in development. Babies tend to rely on the different properties of functors to acquire grammar. This was backed by the experiment mentioned in the previous question by Gervain, Newport, Mazuka, Horne & Mehler.

U2B5 Questions

1. Speech as an Audiovisual Process

Speech is an audiovisual process, as not only the auditory system is the only system contributing to our perception of language, but also other systems such as the visual one. When we communicate, we not only rely on the sounds we hear, but also on the visual information provided by the speaker’s facial movements, gestures, and lip movements. According to the Motor Theory of Language, the understanding of spoken language is not solely based on auditory perception but also involves the visual perception of articulatory gestures. This theory posits that when we observe the movements of the articulators involved in speech production, our brain simulates the corresponding motor actions required to produce those sounds, aiding in our understanding of spoken language. The information obtained from these two systems, visual and auditory, will be automatically integrated into our perception of language, as such, that sometimes our visual and auditory information may conflict with each other.

2. The McGurk Effect

The McGurk effect is when the visual information and the auditory information we obtain from communication conflict with each other, as a consequence our brain perceives an intermediate sound, as an illusion. For instance, if the speaker producer the consonant /b/, but he moves the lips as he was pronouncing a /g/, the information we receive will conflict with each other, and as a result we will perceive that he is producing the sound /d/. This effect shows the interplay between seeing and hearing.

3. Speech Perception Sensitivities in Young Infants

Yes, speech perception sensitivities in young infants partly rely on the productive system, particularly in the context of speech sound discrimination and segmentation. One study investigated whether sensorimotor influences speech perception in infancy. According to the study’s findings, a 6-month-old’s ability to distinguish between nonnative speech sounds is selectively compromised when the tongue movement necessary to produce that contrast is momentarily disabled. This showed that 6 month infants also focused a lot in the movement of the speaker, and not just on what they hear.

UNIT 3

1. Variables Affecting Word Processing

WORD PROCESSING: Word processing is the set of processes involved in the comprehension and production of words.

• During a normal conversation – 150 words per minute
• Master debaters and auctioneers can reach speeds of 400 to 500 words per minute!
• Therefore, we have less than half a second to match up the sounds with a voluminous vocabulary
• Written language is fairly recent human innovation (the oldest known writing system dates back 5,000 years)
• We can read about 200 to 400 words per minute
• When we want to say a word, several information is activated:
• We activate: Its meaning, Its phonological information, Its orthographic information, Its morphosyntactic information, i.e. we know that dog is a noun, it is masculin, and singular.
• Characteristics of word processing: Automatic, effortless, fast, efficient/precise, robust.
• The mental lexicon is the storage or deposit where speakers keep all the words that they know. We can picture it as a warehouse, but above all else it is a cognitive function.

Characteristics:

• Speech shadowing
• The participant is given auditory sentences that they listen to, and what they have to do is repeat the sentences that they hear (like a parrot) as fast and as accurately as they can.
• What we measure is the latency (i.e. the difference in time) between the presentation of each word in the sentence and the repetition of that word by the participant.
• Speakers can identify and repeat words before they are done listening to them.
• Average response latency is 250 ms.
• 18-month-old babies can already do it!

CHAT GPT

Word processing, the cognitive process of recognizing and understanding words, is influenced by various factors that can impact how efficiently and accurately words are perceived and comprehended. Some of the key variables that affect word processing include:

• Word Frequency: The frequency with which a word occurs in language can influence how quickly it is recognized and understood. High-frequency words, which are encountered more frequently in everyday language, are typically processed more quickly than low-frequency words. More frequent words are easier to recognize than less frequent words. Very robust effect.
• Word Length: The length of a word, measured in the number of letters or syllables, can affect processing speed. Generally, shorter words are processed more quickly than longer words due to their simpler and more efficient visual or phonological encoding.
• Orthographic Regularity: Regular words follow predictable spelling patterns and sound-letter correspondences, making them easier to process than irregular words that do not adhere to typical orthographic rules.
• Ambiguity.
• Semantic Context (similarity): The meaning of a word can be influenced by the context in which it appears. Semantic context facilitates word recognition by providing additional information that aids in disambiguating between potential word meanings.
• Phonological Neighborhood Density: Words that have many similar-sounding neighbors in the mental lexicon may be more difficult to process due to competition among phonologically similar words during lexical access.
• Uniqueness point: The point of the word in which the word is unique: there is no other word that starts with that phonemes. We activate all words that start with the same phonemes. If we hear /a/ all these words will be activated: animal, aim, ant, angry, ape, etc. The uniqueness points vary across words: /an/ can activate still animal, ant and angry.
• Age of acquisition: At what age do we learn each word. Frequency and age of acquisition are very related: normally, children start learning very frequent words. This variable is more difficult to quantify.
• Lexicality: In a decision task, no-words that follow phonological and orthographic rules of the language. These no-words are called pseudowords. Pseudowords take more time to be considered as no-words in a specific language.
• Other variables:

• Context and predictability: words that are predicted by context have less processing cost than unexpected words.
• Imagination: words that are easily associated with an imagined meaning are processed faster than words that are harder to conjure up.
• Ambiguity: cross-modal priming task examples.
• Emotional content: emotionally-charged words are associated with a target processing cost than emotionally-neutral words

An experiment that provides evidence for the influence of word frequency on word processing is the lexical decision task. In this experiment, participants are presented with a series of letter strings and are asked to indicate whether each string is a real word or a non-word by pressing a button as quickly and accurately as possible.

For example, participants might be presented with letter strings such as “table,” “zorch,” “cat,” and “blug.” High-frequency words like “table” are typically responded to more quickly and accurately compared to low-frequency words like “zorch” because high-frequency words have stronger and more accessible lexical representations in the mental lexicon. This pattern of results supports the notion that word frequency influences the speed and accuracy of word processing.

2. Models of Word Processing

1: Autonomous serial search model / Forster’s model

It explains:

1.Frequency effect: high frequent words are higher in the access files than low frequent words.

2.Semantic priming effect – semantic similarity: words that are semantically related, are connected in the main file (but it does not include the decay function)

3.Nonwords: to identify a stimulus as a nonword, we have to look into the whole file; it takes more time

• Serial model: Similar to searching a word in a dictionary. (Two types of files: Orthographic file, Phonological file, Semantic / syntactic file).
• More frequent words are placed in higher bins than less frequent words – it explains the frequency effect.
• Searching involves two steps: First step: access files, Second step: main file
• Access files are connected to the main file.
• Main file: once we have found the word in the access file, we access to the main file and we can get all the available information about the word.
• Inside the main file, there are connections to other semantically related words.
• This model considers that top-down processes do not effect word recognition; they are the last step. The context effect here is the last step.

Limitations:

• Uniqueness effect
• Inhibition due to phonologically related words
• Neighborhood density effect
• Ambiguity
• Context effect
• Nonwords / pseudowords

2: Mixed Models/Cohort Model

It explains:

1.Uniqueness point.

2.Top-down processes: the context information facilitates the elimination of words.

• It is a intermediate solution between direct access models (logogen model) and serial models (Forster model).
• There are some steps, but several lexical elements are activated simultaneously.
• It is just conceived to speaking words.
• When someone starts listening a word, all words that have the same beginning will be activated. All of them are possible candidates. They are competitors.
• When the uniqueness point arrives, there is only one possible candidate.
• Uniqueness point is different for each word.

Limitations:

• Inhibition due to phonologically related words
• Neighborhood density effect
• Ambiguity
• Frequency effect
• Semantic priming effect
• Nonwords / pseudoword
• Activation of words phonologically related (not the beginning): beaker/speaker; hemlock/lock

UNIT 4

1. Garden-Path Sentences

A grammatically correct sentence that starts in such a way that the readers’ most likely interpretation will be incorrect. They can be also defined as sentences that are difficult to understand because they contain a temporary ambiguity. The tendency is for hearers or readers to initially interpret the ambiguous structure incorrectly, and then experience confusion when that initial interpretation turns out to be grammatically incompatible with later material in the sentence.

Garden path theory: it claims that in its first pass, the parser only cares about identifying coarse grammatical categories, leaving aside these fancy details about verbs’ syntactic frames.

Constraint-based approach: the parser has access very early to information about the syntactic frames that are linked with specific verbs. Access to such information would lead to much “smarter” parsing decisions.

2. Variables Affecting Sentence Processing

Thematic roles: A parser that has access to this semantic information (as predicted by the constraint-based approach) could make much smarter guesses about the likely structure of ambiguous word strings in at least some cases.

The syntactic frames of verbs:Understanding the syntactic frames of verbs is essential for comprehending how sentences are structured and processed in language.

• Intransitive verbs – one argument / participant
• Transitive verbs – two arguments / participants
• Ditransitive verbs – three arguments / participants
• Sentential complement verbs – they introduce a clause rather than a direct object

Frequency-based information: Parsing seems to be sensitive to frequency-based information related to specific verbs and the structural frames in which they commonly appear. Frequency-based information refers to the statistical distribution of linguistic elements within a language based on their occurrence or usage frequency.

The context: Some studies have demonstrated that the right or the correct context can reduce garden path effects, eliminating the ambiguity including complex structures such as a reduced relative clause.

UNIT 5

QUESTIONS U5B1

1. Structural Differences Between Grey and White Matter

Grey matter, which is composed of neuronal cell bodies and dendrites, is found in the outermost layer of the cortex. Their structure helps them to handle information processing and computation. On the other hand, white matter, made of myelinated nerve axons, is located beneath the cortex. The structure of the white matter facilitates communication between different brain regions, for instance, the corpus callosum is made up of white matter and connects the two hemispheres of the brain and facilitates communication between them allowing for coordinated functioning.

2. Split-Brain Experiments and Functional Specialization

Split-brain people are those who do not have the corpus callosum, so their brain hemispheres are not connected, and cooperation and connection (information transmission) between them is disabled. As revealed in the video on split-brain studied by Roger Sperry and Michael Gazzaniga (1960s), people with this demonstrated to have hemispheric specialization; this is, each hemisphere could do some specific task. The left hemisphere handled language, while the right hemisphere handled visual tasks (drawing).

3. The Penfield Homunculus

The Penfield Homunculus illustrates the representation of various

body parts and the organisation of the motor and sensory cortices. It facilitates comprehension of the relationship between particular brain regions and both motor control and sensory experience.

4. How does neuroplasticity manifest in deaf or blind individuals, and what does this tell us about the brain’s capacity to adapt sensory processing areas for other uses?

Neuroplasticity refers to the brain’s ability to modify its neuron connections and adapt its behaviour based on its experiences, environment, or trauma. The definition includes both structural changes to the brain and the movement of particular functions from one region of the brain to another. In the case of deaf people, the auditory cortex is repurposed for tactile or visual processing (cross-modal plasticity), while in blind people, the brain regions normally responsible for vision may change to accommodate enhanced touch or auditory processing. These two examples of readjustments highlight the brain’s plasticity and potential for functional reassignment, demonstrating the brain’s amazing capacity to reorganise in response to a lack of sensory input.

QUESTIONS U5B2

1. What is the perisylvian region and how does it relate it to language? Give examples

The perisylvian region is a network in the brain that includes Broca’s area (BA44 and BA45), Wernicke’s area (BA22), and the inferior parietal lobe, also known as Geschwind’s Territory (BA39 and BA49). Although several regions participate in language production and comprehension, the perisylvian region is favoured as the key area that contributes to production and comprehension (Broca and Wernicke, respectively).

2. What are the key brain areas involved in Friederici’s model of language processing?

Friederici’s model of language processing involves the areas of Broca’s area, Wernicke’s area, the superior temporal gyrus, the inferior frontal gyrus, and the angular gyrus. These regions are interconnected and contribute to various aspects of language comprehension and production.

3. How does Hickok and Poeppel’s dual-stream model of speech processing differentiate the roles of the dorsal and ventral streams in language comprehension and production?

Hickok and Poeppel’s dual-stream model proposes that language processing involves two main pathways:

1. The dorsal stream (The “How” Pathway), which runs from the temporal lobe, through the parietal lobe, and into the frontal regions (including Broca’s area). This stream is considered critical for speech production as it’s specialized in mapping sound onto motor representations, this is on translating what we hear into the motor commands needed for speech output.
2. The ventral stream (The “What” Pathway), which extends from the temporal lobe directly to the ventral frontal cortex. This stream involves both sides of the brain, so it is bilateral. This pathway focuses more on what words mean (compared to the dorsal stream which focuses more on how words are formed). It processes the content of language, allowing us to comprehend spoken and written messages. So, this pathway is specialized in language comprehension.

4. What is the neural basis of sign language?

The neural basis of sign language involves the activation of the same brain areas that are activated with spoken language in case of production. This shows that the temporal lobe is not only activated by acoustic stimuli but by the production of both, signs and spoken language. Regarding comprehension, a similar pattern of activation is observed compared with spoken language; however, the stimuli that activate those areas are different. While in spoken language, those areas are activated by auditory stimuli, in deaf individuals, they are activated by visual stimuli. This indicates that language processing is not tied exclusively to auditory input.

UNIT 6

1. Compare and contrast the word association and the concept mediation models of bilingual language organization. Compare each of them to the revised hierarchical model. Why have language scientists largely abandoned the word association and concept mediation models in favour of the revised hierarchical model?

The word association model posited that bilinguals store translations of words in each language separately, and when one language is activated, it automatically activates translations in the other language, without necessarily activating the concept of the words in question. According to this model, a picture naming task would involve more cognitive effort than a translation task. In contrast, the concept mediation model suggests words in each language are linked to shared concepts, so translating from the L1 to the L2 involves accessing the concept that goes with the L1 label, and then following the link from the concept to the L2 label. And, according to this model, both the picture naming task and the translation task would take the same cognitive effort.

To contrast these models, a series of experiments were carried out which compared highly proficient Chinese-English bilinguals (who named pictures in their L2 (English) and translated matching Chinese words for the same concepts ) and novice English-French bilinguals (who performed the same two tasks in their L2 (French)). The results showed that both tasks took the same time (= the same cognitive effort); it took them the same time translating from the L1 to the L2 and naming the pictures in their L2. These results went against the word association model, which predicted that the picture naming task would take more time than the translation task. Instead, they appeared to activate concepts (meanings) associated with the L1 labels and used connections between the activated concepts and the L2 labels to complete the translation task. However, what the concept mediation model suggests is also not entirely correct. Further research from forward translation (from the L1 to the L2) and backward translation (from the L2 to the L1) showed that participants took longer to process from the L1 to the L2 than from the L2 to the L1. These results went against the concept mediated model, which proposed that both languages’ labels were linked to the same concept, thus it should take the same time to translate from the L1 to the L2 and from the L2 to the L1.

To account for the results of the previously mentioned experiments, Kroll and Steward (1994) proposed a revised hierarchical model. This model suggests that bilingual language organization involves interconnected nodes. According to this model, L1 labels connect directly to L2 labels, but those connections are weaker in the L1 to L2 direction than in the L2 to L1 direction. This model predicts that it should be faster to translate from the L2 to the L1 as it would be possible to translate from the L2 to the L1 without passing through the conceptual representations. However, translating from the L1 to the L2 should still work like what the concept mediated model predicts.

2. What has the study of cognates contributed to the understanding of bilingualism? How about interlingual homographs?

 A cognate is a word that has a similar form and meaning across bilingual’s two languages (e.g., piano). Cognates tend to facilitate processing on word recognition and word production tasks. Homographs are words that have a similar lexical form but conflicting meanings across the bilingual’s two languages (e.g., pie). They typically produce interference in processing because we have to inhibit the unnecessary meaning. These types of words help to understand one of the important discoveries of the last twenty years of research: the parallel activation of the bilingual’s two languages. Research with cognates and homographs showed that bilinguals also demonstrate parallel activation of the two languages while listening to speech in either language. Additionally, both languages appear to be active when bilinguals plan to speak in one language alone. The cognate facilitation shows that both of the bilingual speaker’s languages are active at the same time if bilinguals could completely switch off the task-irrelevant language, cognate effects would not appear when the task requires only one of the two languages.

3. What is a language-switching task, and how is it used to measure language control in bilinguals?

A language-switching task is a cognitive task where bilingual individuals are required to switch between languages according to certain cues or instructions. It’s used to measure language control in bilinguals by assessing their ability to inhibit one language and activate the other based on demands.

Participants might be asked to name pictures or read words in one language and then switch to the other language based on specific prompts. For instance, participants can be shown some pictures, and when the pictures appear surrounded by a red colour, they need to name them in Spanish, and when they appear surrounded by a blue colour, they need to name the word in English. In this type of task, we can observe the switching cost by observing the time needed to switch from one language to another. This time reflects the time the brain needs to suppress the language, and the shorter or longer time frames indicate lower or higher processing costs. If these switching costs are asymmetrical, it means that the processing costs are different when switching from the L1 to the L2 and when switching from the L2 to the L1 (asymmetrical switching costs can be due to different reasons: disinhibition of the L1, parallel activation, lexical competition, increased cognitive efforts, the context of the language use, etc.). On the other hand, when processing costs are symmetrical, it means that the processing costs of switching from the L1 to the L2 and switching from the L2 to the L1 are the same.

4. Why might switching from a second language (L2) to a first language (L1) incur greater cognitive costs than switching from L1 to L2?

Switching from an L2 to an L1 might incur greater cognitive costs for several reasons. Firstly, for many bilinguals, when they use L2, the more dominant L1 (as it is more rooted) is actively suppressed (inhibited) to prevent interference, which requires cognitive effort. Secondly, switching back to the L1 involves releasing the suppression, this is disinhibition. This reactivation of L1 is not merely turning off the suppression but re-engaging language processes that were temporarily inactive or less active. This requires cognitive recalibration, making the switch back to L1 potentially slower or more cognitively demanding. Thirdly, both languages are typically activated to some extent regardless of which language is being used, due to the interconnected nature of bilingual lexical access. This parallel activation of lexical representations from both languages complicates the task of switching, especially when moving back to L1. Additionally, when switching from L2 to L1, lexical items from L2 might still be activated and can compete with the reactivation of L1 items. This competition requires additional cognitive resources to resolve, which can heighten the cognitive load and slow down the switching process. Also, the switch from L2 to L1 often involves a transition from using a language that may be less automatic and less fluent to one that is more automatic. This shift can temporarily increase cognitive load as more automatic processes (L1) need to be re-engaged after being suppressed. Lastly, the context in which each language is used can also affect switching costs. For instance, if L2 is used primarily in complex cognitive environments (e.g., professional settings), switching back to L1 used in more familiar or less demanding settings may involve adjusting to different cognitive demands.

5. What does it mean when we say the bilingual language system is adaptive? Give examples

Saying that the bilingual language system is adaptive means that the language system is permeable in both directions. This means that the L1 influences the L2 and the other way around. L2 learning affects the L1, especially when the learner achieves L2 proficiency. Therefore, evidence suggesting that the purportedly stable L1 syntactic system is subjective to influence from the L2 provides strong evidence for a linguistic system that is far more open and dynamic than previously thought. Dussias & Sagarra’s (2017) study tried to test the exposure hypothesis. They conducted their experiment with 2 groups whose L1 was Spanish, however, 1 group was immersed in an L1 Spanish context, while the other group was immersed in an L1 English context. They tested these participants in anaphora resolution in embedded relative clauses with ambiguous antecedents (e.g., Alguien disparó al hijo de la actriz que estaba en el balcón). The results of the experiment showed that the bilinguals immersed in an L1 Spanish context showed that they processed the ambiguity using a high attachment preference, the expected of Spanish speakers. On the other hand, the ones immersed in an L1 English context showed a low attachment preference  (the common processing of English speakers) when reading in Spanish. The results suggested that the NL of the participants immersed in an L1 English context changes in response to the L2 use. Another study by Link et al. (2009) tested comprehension and production in native English students studying in Sp (L2) environments, in an immersed context. These participants were matched with classroom-only learners counterparts (native English speakers learning Spanish in classes in the US). The results showed that immersed learners had more difficulties speaking English than classroom learners, suggesting that immersed learners suppress the NL. This difficulty in comprehension persisted for 6 months after returning to the L1 environment. These studies illustrate how the L2 influenced the L1, supporting the claim that the bilingual language system is permeable.

6. What are the three main components of EC defined by Miyake et al., (2000)? How can we measure them? Define a linguistic and a non-linguistic task for each of them.

Miyake et al. (2000) proposed a framework of executive control (EC) consisting of three distinct processes, each serving unique cognitive functions and measurable through diverse tasks. The first component, inhibition, involves the ability to suppress automatic or dominant responses. This can be assessed linguistically through the Stroop task or non-linguistically with the Simon task or the flanker task.

The second component, mental-set shifting, refers to the capability to switch tasks or mental sets flexibly. Linguistically, this is measured through a verbal fluency task or a switching task. Non-linguistically, the Trail Making Test serves this purpose.

Lastly, information updating and monitoring deals with the ability to revise and replace outdated information in the working memory with new, relevant data. This is evaluated linguistically through the keep track task or non-linguistically with the N-back task.

As aforementioned, one linguistic task is verbal fluency (used to measure mental-set shifting). In this kind of task, participants have to name things. In the first condition, participants need to name all the animals they recall within a minute. During the second condition, they need to name all the fruit they recall in one minute. Lastly, in the alternating condition, participants need to alternate between naming different categories (fruit animal fruit animal) without repenting them. The aim of this task is to measure the switching cost and number of switches (compared to the blocked condition). And with the results you obtain an index of cognitive flexibility.

On the other, a non-linguistic task is the Simon task (used to measure inhibition). In this case, participants look at a screen and are given instructions to press, for instance, the left button if what appears on the screen is a green circle and to press the right button if what it appears is a blue circle. During the compatible condition, the circles appear in the appropriate site of the button that needs to be pressed, while during the incompatible condition, the circles appear on the opposite side, and participants need to suppress the impulse to press the incorrect one, this is the contradictory stimuli they are given.

7. Why does the bilingual advantage in executive control functions remain controversial in the field? Give examples.

The debate over the bilingual advantage in executive control (EC) functions is controversial, mainly due to inconsistencies in research findings and methodological variations across studies. The relationship between bilingualism and enhanced EC is not fully established, as various factors contribute to the different results obtained.

Firstly, the characteristics of the bilingual individuals participating across studies can greatly influence the results. As argued by Hernández et al. (2013), the age at which participants acquired their second language, their proficiency levels, and the frequency with which they use both languages are some of the critical variables that can determine whether an advantage is observed. Studies with heterogeneous bilingual groups may yield different results compared to those with more homogeneous samples.

Also, as noted by Costa et al. (2009), the variability and complexity of the tasks used to measure executive functions are another factor that needs to be considered. Different tasks, even when they measure the same component of EC, can vary significantly in their demands on cognitive resources. This variation can lead to conflicting results, as some tasks may be more sensitive to the nuances of bilingual processing than others. Some have pointed out, for instance, that the Simon task was not a good task to measure inhibition.

Lastly, as Hernández et al. (2013) pointed out, the socio-linguistic context in which bilinguals are immersed also plays a significant role. For example, bilinguals living in environments where both languages are equally valued and used might exhibit stronger executive control compared to those in a community where one language predominates. Even though the same experiment is tried to be reduplicated, if the participants are from different socio-linguistic environments, the results will not be the same.

In essence, the bilingual advantage in executive functions remains controversial because of these diverse and interacting factors, making it challenging to draw definitive conclusions from the available research. Also, as the replication of previous studies has not shown the same results, it may also suggest that the effect may not be as robust as thought. Or, maybe, that the sample or the measures used to measure bilinguals are not the appropriate ones.

8. “Having two language systems not only reorganizes the bilingual mind, it also shapes the bilingual brain.” Explain what this statement means and give examples.

The statement “Having two language systems not only reorganizes the bilingual mind, it also shapes the bilingual brain” emphasizes the profound impact that bilingualism has on both cognitive functions and brain structures. This effect is an excellent example of cognitive reserve, a concept defined by Pettigrew and Soldan (2019) as the brain’s adaptability that enables individuals to maintain cognitive function despite challenges such as ageing, brain disease, or trauma. Bilingualism, as a component of cognitive reserve, contributes to this resilience by enhancing the efficiency, capacity, and flexibility of brain networks. It is important to note that this does not mean that it prevents it, just that it helps to delay it for some time.

Research has demonstrated several ways in which bilingualism beneficially restructures the brain and supports cognitive function. For instance, bilingual individuals often show a delayed onset of Alzheimer’s disease symptoms by approximately 4-5 years compared to monolinguals, according to studies by Bialystok et al. (2016) (it mitigates the effects, this is, delays the onset because bilinguals possessing more cognitive reserve helps compensate from the loss they are suffering, but once it starts it moves very fast because the brain is already damaged). Additionally, neuroimaging studies have further supported these findings, revealing that bilinguals typically have increased grey matter density and greater integrity of white matter (Mechelli et al., 2004; Cummine & Boliek, 2013). These changes are particularly notable in areas of the brain associated with executive control (EC) and language processing, highlighting the direct influence of bilingualism on brain structure and function.

Moreover, the benefits of bilingualism extend to recovery from neurological damage. For example, Alladi et al. (2016) found that bilinguals have a facilitated recovery from cognitive impairments following a stroke, and Papiklar et al. (2018) reported that bilingual stroke patients tend to experience less severe aphasia after a stroke.

These findings illustrate that bilingualism not only modifies the brain’s anatomical features but also its operational capabilities, enhancing its ability to cope with and adapt to cognitive challenges across the lifespan.