T-LAB Home
T-LAB 7.3 ON-LINE HELP Prev Page Prev Page
T-LAB
Introduction
What T-LAB does and what it enables us to do
Requirements and Performances
Corpus Preparation
Corpus Preparation
Structural Criteria
Formal Criteria
File
New Corpus
Gather your Texts
Open Corpus
Settings
Automatic Settings
Customized Settings
Co-occurrence Analysis
Word Associations
Co-Word Analysis and Concept Mapping
Comparison between Word pairs
Sequence Analysis
Concordances
Thematic Analysis
Thematic Analysis of Elementary Contexts
Modeling of Emerging Themes
Sequences of Themes
Key Contexts of Thematic Words
Thematic Document Classification
Comparative Analysis
Specificity Analysis
Correspondence Analysis
Multiple Correspondence Analysis
Cluster Analysis
Contingency Tables
Lexical Tools
Stop-Word List
Multi-Word List
Corpus Vocabulary
Disambiguation
Dictionary Building
Utilities
Editor
Memo
Variable Manager
Create a Sub-Corpus
Glossary
Analysis Unit
Association Indexes
Chi-Square
Cluster Analysis
Coding
Context Unit
Corpus and Subsets
Correspondence Analysis
Data Table
Disambiguation
Dictionary
Elementary Context
Frequency Threshold
GraphML
Homograph
IDnumber
Isotopy
Key-Word (Key-Term)
Lemmatization
Lexical Unit
Lexie and Lexicalization
Markov Chain
MDS
Multiwords
Naïve Bayes
Normalization
Occurrences and Co-occurrences
Poles of Factors
Primary Document
Profile
Specificity
Stop Word List
Test Value
Thematic Nucleus
TF-IDF
Variables and Categories
Words and Lemmas
Bibliografia

Structural Criteria



There are two structural criteria which must be observed: the corpus size and its subdivision into parts.

As for the size, T-LAB has been tested with a 30Mb corpus, approximately equivalent to 18,000 pages in ASCII/ANSI format.

Limits for the minimum size require different evaluation criteria, because, under a certain threshold, the corpus size can prejudice the reliability of many statistical analyses. Just follow these simple instructions: use corpora with at least 5,000 occurrences (approximately 30 Kb); otherwise, in the case of open-ended questions, a minimum of 50 answers.

In order to be processed, a corpus can be made up of: a single text without further partitions; a single text subdivided according to criteria established by the user (for example, a book divided into chapters); a number of texts (for example, different interviews or documents) classified through the use of labels linked to as many variables or IDnumber. In any case, the corpus is subdivided into parts that must be defined by precise formal criteria.