Yue Chen bio photo

Psycholinguistics

Email

My CV

Github

Google Scholar

🫪 Era

Latest Update: June 29 2026
Note: These tutorials are based on materials I developed for my own projects and/or for training students/research assistants.

On this page:

Vocabulary Assessment – PPVT

What is the PPVT?

The Peabody Picture Vocabulary Test (PPVT) is a standardized test of receptive vocabulary. On each trial, the examiner says a word, and the participant selects one picture out of four that best matches the meaning of that word. Because the response only requires pointing (rather than speaking or reading), the PPVT is widely used with children, multilingual speakers, heritage speakers, and populations where productive language may be limited.

Scores are norm-referenced by age and can be used to estimate vocabulary level, compare groups, or include receptive vocabulary as a covariate in experimental work.

Download PPVT Tutorial (PDF)

This PDF provides a brief introduction to the Peabody Picture Vocabulary Test (PPVT), including background, basal/ceiling rules, and example items for administration. It is adapted from materials I developed for one of my previous research projects.

Transcription & Corpus Analysis – CLAN

What is CLAN?

CLAN (Computerized Language ANalysis) is the main software used with the CHAT transcription format and the TalkBank/CHILDES corpora. It supports:

  • Creating and editing CHAT-formatted transcripts
  • Running automatic morphological analysis via MOR or batchalign2
  • Computing frequencies, MLU, TTR, Word Type, Word Tokens, and many other measures from transcripts
  • Working with multiple languages (e.g., eng, zho, yue, etc.)

It’s widely used in child language, clinical, and corpus-based research, and is basically the standard toolchain if you work with TalkBank data.

Download CLAN Tutorial (PDF)

This PDF walks through installing CLAN on macOS, adding MOR grammars, locating CLAN on your laptop, setting up new scripts, using header and ID tiers, choosing language codes, and applying common shortcuts and conventions for gestures, code-switching, and other CHAT symbols. It is adapted from materials I developed while working as a research assistant under the supervision of Maggie Ziyin Mai and for training students and research assistants.

Download CLAN Transcription Tutorial (PDF)

This PDF walks through how to perform transcription in CLAN. It is adapted from materials I developed while working as a research assistant under the supervision of Maggie Ziyin Mai and for training students and research assistants.

Syntactic & Morphological Parsing with Batchalign2

Batchalign2 is a command-line pipeline developed in the TalkBank project for automatic morphological tagging and Universal Dependencies (UD) syntactic parsing.

  • Takes CHAT .cha transcripts as input
  • Produces %mor tiers (lemmas, POS tags, morphological features)
  • Produces %gra tiers (UD-style dependency parses)
  • Integrates with CLAN, so you can immediately use kwal, freq, etc. on the annotated corpus

Typical workflow:

  • Use R or other scripts to format the raw text into valid .cha files (including headers such as @Begin, @Languages:, @ID, *ADU:, etc.).
  • Use Batchalign2 in the local terminal to run batchalign morphotag ba_input ba_output.
  • Inspect the %mor and %gra tiers in the output .cha files, and then use CLAN for further analyses (e.g., kwal +t%mor, kwal +t%gra).

Download Batchalign2 Parsing Tutorial (PDF)

This PDF provides step-by-step instructions for installing Batchalign2 with uv, preparing .cha files, running batchalign morphotag on input/output folders, and interpreting %mor and %gra tiers, with concrete examples from Italian data and suggestions for CLAN-based post-processing, which is adapted from materials I developed for previous research projects.

Forced Alignment – Montreal Forced Aligner (MFA)

What is MFA?

Montreal Forced Aligner (MFA) is an open-source toolkit for forced alignment: automatically aligning audio recordings with their transcripts to obtain word- and phone-level timestamps. It is widely used in phonetics, corpus linguistics, and clinical/experimental work where you need time-aligned segments but don’t want to hand-label everything in Praat.

MFA uses:

  • Pre-trained acoustic models for many languages
  • Corresponding pronunciation dictionaries
  • Command-line workflows that can be integrated with other tools (e.g., Python, Praat, R)

Typical use cases include preparing data for formant analysis, VOT measurements, prosody studies, and large-scale speech corpora.

Download Montreal Forced Aligner Tutorial (PDF)

This PDF provides a step-by-step guide to installing MFA with Miniconda, setting up a dedicated environment, installing MFA, downloading acoustic models and pronunciation dictionaries (e.g., english_us_arpa, mandarin_mfa), and running alignment commands from the terminal with concrete path examples and sample output. It is adapted from materials I developed for a project testing Whisper-based automatic transcription of naturalistic speech production.

Factor analysis – in preparation

What is Factor Analysis?

Factor analysis is a family of statistical methods used to identify latent variables (factors) that explain patterns of correlations among observed variables. It is widely used in psychology, linguistics, education, and other social sciences when researchers want to understand whether a larger set of measured variables can be represented by a smaller number of underlying constructs.

Two major approaches are:

  • Exploratory Factor Analysis (EFA): used when the underlying factor structure is not known in advance. EFA helps explore how many latent factors may exist and which observed variables tend to cluster together.
  • In EFA, the relationships between observed variables and latent factors are largely estimated from the data. Researchers typically examine factor loadings, eigenvalues, scree plots, factor rotations, and model interpretability when deciding on an appropriate factor solution.
  • EFA asks: What latent structure might explain these variables?
  • Confirmatory Factor Analysis (CFA): used when researchers have a theoretically motivated factor structure and want to test how well that structure fits the observed data.
  • In CFA, the researcher specifies the expected relationships between observed variables and latent factors a priori. Model fit can then be evaluated using indices such as the χ² test, CFI, TLI, RMSEA, and SRMR, together with parameter estimates such as factor loadings and factor correlations.
  • CFA asks: Does my hypothesized latent structure adequately explain these variables?

Factor analysis can be used to:

  • Examine the dimensional structure of questionnaires, rating scales, or behavioral measures
  • Identify groups of correlated variables that may reflect a common underlying construct
  • Evaluate whether survey items measure the intended latent constructs

Psych Verb Corpus (English–Chinese) – in preparation

This is a bilingual corpus of English psychological verbs and their Chinese translations, linked to large-scale norming data from three studies on implicit causality/consequentiality and related semantic dimensions.

The resource is designed to support work on:

  • Pronoun resolution and discourse coherence
  • Cross-linguistic comparisons of psych verbs
  • Experimental stimulus construction in English and Mandarin

Planned public release: after the main papers based on these data are published.
A subset of the corpus and documentation will be posted here when available.

Narrow Phonetic Transcription Project (Mandarin Chinese) — in preparation

This ongoing project provides narrow phonetic transcriptions for the entire Mandarin CHILDES collection. Our goal is to systematically annotate speech errors—including tone errors, segmental (consonant/vowel) errors, and prosodic irregularities—across naturalistic adult–child interactions in Mandarin.

Designed to support work on:

  • Error patterns in child and caregiver speech from naturalistic corpora
  • Tone acquisition and tone-specific error typology
  • Segmental development and phonotactic constraints in early speech
  • Links between disfluency/prosody and segmental or tonal errors

Planned public release: after the main papers based on these data are published.
A subset of the corpus and documentation will be posted here when available.