• Fine-tuning acoustic analysis.
• LPC in depth.
• LPC analysis/re-synthesis with Praat.
• Automating analysis with Praat scripts.
• Advanced scripting techniques - using variables, loops, conditional logic, etc.
• Advanced synthesis and re-synthesis techniques.
• Producing high quality EPS vector graphics with Praat.
• Building interactive experiments with Praat.
• Web-based data harvesting and metadata encoding - WEG, OLAC, and EMELD.

<< Back to Program

Linguistic tools for reading research
William Labov and Bettina Baker, University of Pennsylvania

This workshop is designed for those who would like to use linguistic data and methods to advance reading levels, with emphasis on African-American and Latino school populations. It will present some of the methods and computational tools developed by the Urban Minorities Reading Project at the U. of Pennsylvania, and address open questions for those who would like to do research in this field. The workshop will consist of five segments:

1. Preconditions for reading research in public schools. Establishing relations between the university and the school system; recruiting university resources for intervention in the schools; dealing with the whole language/phonics controversy; training tutors; linguistically based intervention programs.

2. The RX program for analyzing reading error profiles. Computational approaches to the linguistic analysis of texts and reading errors; distinguishing reading errors from differences in pronunciation.

3. The Individualized Reading Manual. Combining direct instruction with practice; directing instruction by reading error profiles; narratives with controlled vocabulary; addressing the emotional concerns and interests of inner city children.

4. The DX program for the analysis of speech. The study of spontaneous speech in the school context; implanting phonological codes in transcripts; automatic tabulation and production of varbrul files.

5. Current findings on the relation of speech to reading. Effects on English reading of prior reading instruction in Spanish. Relation of home language and dialect to reading errors. Probablistic approaches to the problem of defining reading errors and pedagogical consequences. Semantic shadows cast by decoding errors.

<< Back to Program

Robust sociolinguistic methodology: Tools, data and best practices
Chris Cieri and Stephanie Strassel, University of Pennsylvania

Note: This workshop will take place in the conference room at the Linguistic Data Consortium, 3600 Market St., Suite 810. From JMHH, walk 2 blocks north and 2 blocks east; the entrance to the building is on the south side of Market just west of 36th. You will be asked to sign in at the front desk. One of the workshop leaders will be available to guide participants from JMHH to the LDC; if you're interested, please meet near the NWAVE registration desk at 11:50 am.

The methodology for sociolinguistic research, despite gradual evolution, has not fully exploited recent technological advances. Given the current state of the art in computing, one can envision a completely digital methodology for collecting, coding, analyzing and publishing linguistic data. Although progress toward this end is ongoing, there now exist tools, standards, examples of best practice and other linguistic resources that ease the process of building databases for analysis and sharing them. Shared digital resources promote research that is more robust and repeatable, and allow testing of competing analyses against a stable benchmark.

This workshop describes a rigorous, end-to-end digital methodology covering:

• field data collection and the use of publicly available corpora
• interview transcription, segmentation into meaningful units and alignment of audio and transcript
• token selection and coding that are compatible with multiple analytical packages
• publication and distribution of both the analysis and the underlying data

The format of the workshop will combine discussion of general principles with live demonstration of data and tools. Among the resources made available to workshop participants will be the SLX Corpus of Classic Sociolinguistic Interviews, consisting of 8 sociolinguistic interviews collected by William Labov and his students in the 1960s and 70s. The preparation of this corpus, from digitization, segmentation, transcription, coding and analysis to publication will be discussed as an illustration of the proposed digital sociolinguistic methodology.

<< Back to Program

Methods in the quantitative analysis of historical syntax
Anthony Kroch, Beatrice Santorini, and Laura Whitton, University of Pennsylvania

The aim of this workshop is to introduce researchers to the parsed corpora of historical English that are now or soon will become available and to a computer program, Corpus Search, that allows users to search these corpora for lexical, morphological and syntactic configurations, as well as to code files of examples automatically for statistical analysis. The corpora to be discussed will be the York corpus of Old English and the Penn Corpora of Middle English and of early Modern English. All three of these corpora are based on the Helsinki diachronic corpora of English but include more texts and larger text samples than the Helsinki corpora.

The following topics will be addressed:

1. Motivation: Why parsed corpora are useful. What parsed corpora allow us to find out about a language that simple text corpora would not.

2. Corpus Structure: How corpora are annotated and how these annotations facilitate searches.

3. Searching: How to use CorpusSearch. What the program can do and how to get it to do what we want. Features to be presented will include:

• simple searches
• automatic coding of examples
• the use of definition files to simplify complex searches

4. Analysis: A brief introduction to the quantitative analysis of syntactic data. Examples will be taken from recent work on the history of English that the parsed corpora have made possible.

<< Back to Program