Subtopic Deep Dive

Speech Perception Models
Research Guide

What is Speech Perception Models?

Speech perception models are computational frameworks integrating auditory, phonetic, and lexical processing to simulate human speech recognition, such as TRACE and Shortlist.

These models test predictions via psychophysical experiments, eye-tracking, and neuroimaging. Key examples include Shortlist B, a Bayesian model of continuous speech recognition (Norris and McQueen, 2008, 694 citations), and Merge, arguing feedback is unnecessary (Norris et al., 2000, 769 citations). Over 10 high-citation papers from 1997-2008 define the field.

15
Curated Papers
3
Key Challenges

Why It Matters

Speech perception models explain robustness to noise, accents, and talker variability, as in talker-specific learning (Nygaard and Pisoni, 1998, 593 citations). They inform ASR systems and auditory prosthetics by modeling hierarchical brain processing (Davis and Johnsrude, 2003, 789 citations). Perceptual learning mechanisms from Norris (2003, 876 citations) guide speech therapy for disorders like stuttering (Braun, 1997, 338 citations).

Key Research Challenges

Feedback vs. Modularity

Debate persists on whether top-down lexical feedback aids or hinders recognition. Norris et al. (2000, 769 citations) argue feedback is unnecessary and modular processing suffices. Shortlist B (Norris and McQueen, 2008, 694 citations) implements Bayesian competition without feedback loops.

Talker Normalization

Listeners adapt to speaker-specific acoustics rapidly. Nygaard and Pisoni (1998, 593 citations) show talker-specific learning effects. Eisner and McQueen (2005, 367 citations) detail specificity of perceptual learning in processing.

Lexical Competition Dynamics

Non-native and embedded word recognition involves competition resolution. Weber and Cutler (2003, 499 citations) examine lexical competition in non-natives. Salverda et al. (2003, 415 citations) show prosodic boundaries resolve embedding.

Essential Papers

1.

Perceptual learning in speech

Dennis Norris · 2003 · Cognitive Psychology · 876 citations

2.

Hierarchical Processing in Spoken Language Comprehension

Matthew H. Davis, Ingrid S. Johnsrude · 2003 · Journal of Neuroscience · 789 citations

Understanding spoken language requires a complex series of processing stages to translate speech sounds into meaning. In this study, we use functional magnetic resonance imaging to explore the brai...

3.

Merging information in speech recognition: Feedback is never necessary

Dennis Norris, James M. McQueen, Anne Cutler · 2000 · Behavioral and Brain Sciences · 769 citations

Top-down feedback does not benefit speech recognition; on the contrary, it can hinder it. No experimental data imply that feedback loops are required for speech recognition. Feedback is accordingly...

4.

Shortlist B: A Bayesian model of continuous speech recognition.

Dennis Norris, James M. McQueen · 2008 · Psychological Review · 694 citations

A Bayesian model of continuous speech recognition is presented. It is based on Shortlist (D. Norris, 1994; D. Norris, J. M. McQueen, A. Cutler, & S. Butterfield, 1997) and shares many of its key as...

5.

Talker-specific learning in speech perception

Lynne C. Nygaard, David B. Pisoni · 1998 · Perception & Psychophysics · 593 citations

6.

Lexical competition in non-native spoken-word recognition

Andréa Weber, Anne Cutler · 2003 · Journal of Memory and Language · 499 citations

7.

The role of prosodic boundaries in the resolution of lexical embedding in speech comprehension

Anne Pier Salverda, Delphine Dahan, James M. McQueen · 2003 · Cognition · 415 citations

Reading Guide

Foundational Papers

Start with Norris et al. (2000, 769 citations) for modularity argument without feedback; Norris and McQueen (2008, Shortlist B, 694 citations) for Bayesian continuous recognition; Davis and Johnsrude (2003, 789 citations) for hierarchical brain processing.

Recent Advances

Wilson (2006, 383 citations) on phonology learning bias; Eisner and McQueen (2005, 367 citations) on perceptual learning specificity; Salverda et al. (2003, 415 citations) on prosodic resolution.

Core Methods

Parallel lexical competition (Shortlist); Bayesian inference (Shortlist B); fMRI for stages (Davis and Johnsrude); psychophysics for learning (Norris, Eisner and McQueen).

How PapersFlow Helps You Research Speech Perception Models

Discover & Search

Research Agent uses searchPapers and citationGraph to map core models from Norris et al. (2000) 'Merging information in speech recognition,' revealing 769 citations and connections to Shortlist B (Norris and McQueen, 2008). exaSearch finds psychophysical tests; findSimilarPapers expands to hierarchical processing (Davis and Johnsrude, 2003).

Analyze & Verify

Analysis Agent applies readPaperContent to extract Bayesian assumptions from Shortlist B, then verifyResponse with CoVe checks model predictions against fMRI data in Davis and Johnsrude (2003). runPythonAnalysis simulates lexical competition curves from Norris (2003) perceptual learning data using NumPy; GRADE scores evidence strength for modularity claims.

Synthesize & Write

Synthesis Agent detects gaps in feedback-modularity debate across Norris papers, flagging contradictions. Writing Agent uses latexEditText to draft model comparisons, latexSyncCitations for 10+ papers, and latexCompile for psychophysics figures; exportMermaid visualizes TRACE vs. Shortlist architectures.

Use Cases

"Simulate Shortlist B lexical competition with Python from Norris 2008 paper."

Research Agent → searchPapers('Shortlist B') → Analysis Agent → readPaperContent → runPythonAnalysis (NumPy activation curves) → matplotlib plot of recognition probabilities.

"Write LaTeX review comparing feedback models in Norris 2000 vs Davis 2003."

Synthesis Agent → gap detection → Writing Agent → latexEditText (intro) → latexSyncCitations (10 papers) → latexCompile → PDF with hierarchical processing diagram.

"Find code implementations of speech perception models like TRACE or Shortlist."

Research Agent → paperExtractUrls (Norris 2008) → Code Discovery → paperFindGithubRepo → githubRepoInspect → Python sandbox test of Bayesian recognition scripts.

Automated Workflows

Deep Research workflow conducts systematic review of 50+ speech perception papers, chaining citationGraph from Norris (2003) to generate structured report on perceptual learning. DeepScan applies 7-step analysis with CoVe checkpoints to verify Shortlist B predictions against eye-tracking data (Salverda et al., 2003). Theorizer generates new modular model hypotheses from modularity papers like Norris et al. (2000).

Frequently Asked Questions

What defines speech perception models?

Computational frameworks like TRACE and Shortlist integrate auditory, phonetic, and lexical stages to model recognition (Norris and McQueen, 2008). They predict psychophysical and neuroimaging outcomes.

What are main methods?

Bayesian competition (Shortlist B, Norris and McQueen, 2008), hierarchical fMRI mapping (Davis and Johnsrude, 2003), perceptual learning paradigms (Norris, 2003). No feedback loops needed (Norris et al., 2000).

What are key papers?

Norris (2003, 876 citations) on perceptual learning; Norris et al. (2000, 769 citations) on modularity; Shortlist B (Norris and McQueen, 2008, 694 citations). Nygaard and Pisoni (1998, 593 citations) on talker learning.

What open problems exist?

Resolving talker normalization mechanisms (Nygaard and Pisoni, 1998); integrating prosody in competition (Salverda et al., 2003); scaling Bayesian models to non-native input (Weber and Cutler, 2003).

Research Phonetics and Phonology Research with AI

PapersFlow provides specialized AI tools for Psychology researchers. Here are the most relevant for this topic:

See how researchers in Social Sciences use PapersFlow

Field-specific workflows, example queries, and use cases.

Social Sciences Guide

Start Researching Speech Perception Models with AI

Search 474M+ papers, run AI-powered literature reviews, and write with integrated citations — all in one workspace.

See how PapersFlow works for Psychology researchers