Correspondence Analysis

N.B.: The pictures shown in this section have been obtained by using a previous version of T-LAB. These pictures look slightly different in T-LAB 10. Moreover: a) there is a new button (TREE MAP PREVIEW) which allows the user to create dynamic charts in HTML format; b) by right clicking on the keyword tables, additional options become available; c) two new buttons allows us to check the specificities of each variable values either by using the chi-square test or the test value; d) a new button allows the user to carry out a cluster analysis that uses the coordinates of the objects (i.e. either lexical units or context units) on the first factorial axes (up to a maximum of 10); e) a quick access gallery of pictures which works as an additional menu allows one to switch between various outputs with a single click. .
Some of these new features are highlighted in the below image.

This T-LAB tool highlights the similarities and the differences between context units.

More precisely, in T-LAB, correspondence analysis can be applied to three kinds of tables:

(A) tables of words by variables with occurrence values;
(B) tables of elementary contexts by words with co-occurrence values;
(C) tables of documents by words with co-occurrence values;

In order to analyse occurrence tables (A), the corpus should be made up of a minimum of three texts or should be codified with some variables (not less than three categories).
The variables are listed in an appropriate box and can be used one at time.
After every selection - in sequence - the contingency table is dispalyed and T-LAB asks us to click on the analyse button (see below).

The analysis results allow the drawing of graphs in which the relationships between both the corpus subsets and the lexical units that make them up are represented.
More precisely, depending on the case, the types of graphs available show the relationships between active variables, between illustrative variables, between lemmas and between lemmas and variables (see below).

Moreover, when analysing a document by word table, it is possible to visualize the points (Max 3,000) corresponding to each document (see below).

All the graphs can be maximized and customized by using the appropriate dialog box (just right click on the chart). Moreover, when variable categories are 3 or more, their relationships can be explored through 3d moving (see below).

In order to explore the various combinations of the factorial axes it is sufficient to select them in the appropriate boxes ("X Axis", "Y Axis").

In T-LAB the characteristics of each factorial pole (i.e. the opposites on the horizontal and vertical axes) are shown using the Absolute Contributions, the threshold value of which is 1/N (in this case, N = rows of contingence tables), and the Test Values, the threshold value of which is +/- 1.96.

The eigenvalue chart enables the evaluation of the relative weight of each factor, that is the percentage of variance explained by each one of them.

A click on the button "Complete Results" enables the user to visualize and export a file that contains all the results of the analysis: eigenvalues, coordinates, absolute and relative contributions, and test values.

All contingency tables can be easily exported and allow us to create various charts.
Moreover, by clicking on specific cells of the table (see below), it is possible to create a HTML file including all elementary contexts where the word in row is present in the corresponding subset.

In the case of tables (B) and (C) (see above), they consist of as many rows as there are context units (max. 10,000) and as many columns as there are selected key words (max. 3,000).

The calculation algorithm and the outputs are similar to those of the analysis of lexical unit by variable tables, except that - in this case - in order to cut down processing time, In T-LAB the limits itself to the extraction of the first 10 factors, which is a more than sufficient number in order to summarize the variability of the data.

Moreover,subsequently it is possible to carry out a Cluster Analysis.