resources

Everything a new team needs

The shared instruments of the network. Registered teams receive the full pack by email after applying on the Join page; the summaries below let you assess fit before applying.

instruments

The three core instruments

Researcher workshop pack

18 printable worksheets: project identity, RQ0–RQ6, H1–H6 with observables, the theoretical framework, the linguistic-prediction module, corpus and annotation protocols, the AI pipeline, the reader study, six weekly day-by-day plans with Friday checkpoints, deliverables checklist, starter bibliography.

COGNILANG Analyzer

A Streamlit app for AI-assisted exploration: textual-index extraction, similarity/divergence comparison, sentence alignment, TRA shift detection, DIS classification, PRD marker identification, and a cross-analysis tab linking features × shifts × prediction to cognitive hypotheses. JSON and PDF export with full Unicode (RTL Arabic included).

Code-book with examples

The frozen LIN/DIS/TRA/PRD grid with one positive and one negative example per code per language, the κ procedure, and the deviation-log template every replication fills in.

tooling

Recommended tools per component

Model substitution across languages is expected; the validation logic is what stays constant. Versions of every tool used must be logged.

ComponentReference tools (pilot)
EN/FR processingspaCy (en_core_web_*, fr_core_news_*); stanza as cross-check.
Arabic processingCAMeL Tools, Farasa, or stanza (ar) — normalisation settings documented.
Other languagesstanza covers 70+ languages; for CJK add jieba / fugashi; report segmentation agreement on a sample.
Embeddings / alignmentsentence-transformers (LaBSE), LASER; faiss for nearest-neighbour pairing.
Surprisal (PRD-1)transformers + minicons with one monolingual causal LM per language; within-language comparisons or normalised profiles only.
LLM assistanceAny capable LLM API for frame labelling and shift pre-classification — always benchmarked against the gold subset; prompts versioned.
Statisticspandas, scikit-learn, statsmodels; κ on doubled annotations; notebooks under version control.
reference plan

The six-week cycle

WeekFocusKey deliverables
Week 1Framing and architectureTitle, RQ0–RQ6, H1–H6 with observables, definitions (incl. linguistic prediction), outline, abstracts, ethics application submitted.
Week 2Literature reviewThree-block review (cognition · translation · AI) + state of the art on translation as cognition; frozen DIS-1 label set; formatted bibliography.
Week 3Methodology and corpusCorpus collected and documented; grid piloted and frozen (κ ≥ 0.70); AI pipeline (incl. surprisal module) validated on a sample.
Week 4AnalysisComparative LIN/DIS tables; TRA typology; predictability profiles; shift × surprisal cross-table; results and discussion drafts.
Week 5FinalisationIntroduction and conclusion; harmonised text; verified citations and bibliography; near-final version.
Week 6Revision and deliveryFinal document (DOCX + PDF); annexes; oral presentation pack; deviation log to the network.
bibliography

Starter bibliography, by axis

A — Cognition & comprehension

Kintsch, Comprehension: A Paradigm for Cognition · Sweller et al. on cognitive load theory · interdisciplinary work on cognition and language contact.

B — Cognitive translation studies

Pym on cognitive translation studies · the CTIS programmatic literature ("an evolving research area and a thriving community") · systematic reviews of cognitive approaches to translation · translation-meets- cognitive-science syntheses.

C — News discourse & framing

Entman on framing as a fractured paradigm · corpus-based discourse analysis of translated news (e.g. intensifiers in bilingual news translation) · Bielsa & Bassnett, Translation in Global News.

D — AI / NLP for multilingual news

Multilingual multifaceted understanding of online news · scaling up multilingual news framing analysis · tool papers for the components actually used (cite per team).

E — Linguistic prediction

Kutas & Federmeier on the N400 and anticipatory semantic processing · Pickering & Garrod's integrated theory of production and comprehension · Huettig's four central questions about prediction · Hale and Levy on surprisal and expectation-based comprehension · Chernov, Inference and Anticipation in Simultaneous Interpreting · Amos & Pickering on prediction in simultaneous interpreting · Seeber on cognitive load in SI.

Citation rule. Every reference used must be read at least strategically, summarised in a reading fiche, and traceable to a specific claim. No decorative citations.