Bilingualism and the Cognitive Science of Writing

Message Encoding: The cycle begins with conceptualization, where a speaker encodes a preverbal message from a nonlinguistic notion (Konopka & Brown-Schmidt, 2014).
Formulation: Includes grammatical and morphophonological encoding. Lexicalization is a two-step process: accessing a lemma (semantic and syntactic information) and then a lexeme (phonological form) (Thompson & Kielar, 2014).
Lexicalization Evidence:
- Tip-of-the-tongue (TOT) states: Accessing meaning/gender but failing to retrieve sound form (lexeme).
- Speech Errors: Semantic substitutions (e.g., "dog" for "cat") at the lemma stage versus phonological errors (e.g., "cap" for "cat") at the lexeme stage.
- Picture-word interference: Semantic distractors interfere earlier than phonological ones.
Articulation and Monitoring: The phonetic plan drives the motor system, overseen by a self-monitoring system that repairs errors (Hartsuiker, 2014).

Definition: Writing is a visual system for representing language that allows for the permanent accumulation of knowledge across time and geography (Olson, 2012).
Distinction from Speech: Speech is biologically prepared and transient; writing is a cultural invention that requires explicit instruction and allows for recursive refinement (Eysenck & Keane, 2010).
Hayes (2012) Writing Architecture:
- Control Level: Goal setting and motivation.
- Process Level: Proposer (planning), Translator (basic processes), Transcriber (motor processes), and Evaluator (reviewing).
- Resource Level: Working memory, long-term memory, and reading skills.

Complex Processes:
- Planning: Generating/organizing ideas and retrieving knowledge from long-term memory.
- Translation: Converting non-linguistic ideas into linear sentence sequences.
- Reviewing: Evaluating and editing text qualities (Hayes & Flower, 1986).
Basic Processes (Spelling):
- Lexical-semantic route: Retrieves stored forms from the orthographic lexicon for familiar/irregular words like "yacht."
- Sublexical (Non-lexical) route: Uses phoneme-to-grapheme conversion ( $PGC$ ) rules for unfamiliar words or nonwords.
- Graphemic Buffer: A short-term store maintaining letter order during motor preparation.
Motor Processes (Transcription): Transitions from abstract grapheme to allograph (shape/style) to graph (motor program of strokes) (Ellis, 1982).

Handwriting Fluency: Automaticity reduces cognitive load on working memory, freeing resources for higher-level composition. Slow speed predicts lower quality in exam essays (Connelly, Dockrell, & Barnett, 2005).
Literacy Benefits: Explicit handwriting instruction improves letter recognition and helps overcome mirror-image confusions (e.g., distinguishing ‘b’ from ‘d’) (Dehaene, 2014).
Note-taking Functions:
- Encoding Function: Improved retention from the process itself, driven by sustained attention and summarization.
- External Storage Function: Benefits derived from subsequent review of recorded information (Kiewra et al., 1991).
Medium Effects: Handwriting notes encourages condensing and paraphrasing, leading to deeper processing than verbatim typing (Mueller & Oppenheimer, 2014).

Expert vs. Novice: Experts spend up to two-thirds of their time on planning and use a knowledge-transforming strategy (thinking through writing). Novices use knowledge-telling (simply listing known ideas) (Bereiter & Scardamalia, 1987).
Large Language Models (GPT):
- GPT simulates the products of writing (structural outlines, syntactically correct sentences) through probabilistic next-token prediction.
- GPT lacks communicative intent, genuine metacognition, and a "theory of mind" to model reader needs (Olson, 2014).

Shifting Definitions:
- Fractional view: Treating a bilingual as two monolinguals (Bloomfield’s 1935 definition required "native-like control").
- Holistic/Functional view: Defining bilingualism by regular use in daily life, even with minimal fluency (Grosjean, 2013).
Key Dimensions:
- Age of Acquisition (AoA): Simultaneous (from birth/before age $3$ ) vs. Sequential (after age $3$ ).
- Complementarity Principle: Bilinguals often use different languages for specific domains (e.g., home vs. work).
Bilingualism as a Process: Both languages remain active simultaneously. Evidence from Thierry and Wu (2007) shows unconscious $L1$ (Chinese) activation via the $N400$ brainwave during an exclusively $L2$ (English) task.

Phonology: The sensitive period for native-like accent closure is early, possibly around age $6$ (Flege, 1999).
Grammar: Historically thought to end at puberty, but Hartshorne et al. (2018) analyzed data from $670,000$ participants, suggesting the window for native-like grammar acquisition remains open until age $17$ or $18$ .
Brain Activation: Less proficient bilinguals show more diffuse activation in the prefrontal cortex; as proficiency increases, $L2$ neural representation converges with $L1$ .

Lexical Disadvantage: Bilinguals may have smaller vocabularies per language, slower picture-naming, and more TOT states (Baus et al., 2020).
- Frequency-Lag Hypothesis: Lower frequency of use for each word leads to weaker mental links.
- Lexical Competition: Parallel activation causes interference from the non-target language.
Executive Advantage: Constant conflict management may enhance executive control. In the Attentional Network Task ( $ANT$ ), bilinguals often show smaller conflict and switching costs (Costa et al., 2008).
Neuroplasticity and Reserve: Bilingualism is linked to greater grey matter density in the inferior parietal lobule. It may contribute to cognitive reserve, delaying dementia symptoms by $4$ to $5$ years (Alladi et al., 2013).
The UBET Framework: DeLuca et al. (2020) propose the Unifying the Bilingual Experience Trajectories model, focusing on intensity, switching, proficiency, and duration as drivers of neurocognitive adaptation.