coocc: Generate a contingency table of co-occurrences of sound...

In soundcorrs: Semi-Automatic Analysis of Sound Correspondences

Description

Take all segment-to-segment correspondences in a dataset, and cross-tabulate them with themselves or with metadata taken from a separate column.

Usage

coocc(data, column, count, unit)

Arguments

data	[soundcorrs] The dataset from which to draw frequencies. Only datasets with two languages are supported.
column	[character] Name of the column with metadata. If NULL, sound correspondences are cross-tabulated with themselves. Defaults to NULL.
count	[character] Report the absolute number of times or words, or relative to how many times or in how many words the given segments co-occur in L1 or L2. Accepted values are "a(bs(olute))" and "r(el(ative))". Defaults to "a".
unit	[character] Count how many times a correspondence occurs or in how many words it occurs. Accepted values are "o(cc(ur(ence(s))))" and "w(or(d(s)))". Defaults to "w".

Details

A set of segmented and aligned word pairs/triples/..., such as one held in a soundcorrs object, can be turned into a contingency table in more than one way. This function creates a table which details how often various sound correspondences co-occur in one word. Both rows and columns are named in the same way: L1 phoneme + underscore ("_") + L2 phoneme. The values in the table can be absolute or relative, and they can represent the number of times the given correspondence co-occurs, or the number of words in which it co-occurs. For example, in the pair German koala : French koala, the correspondence G a : Fr a ("a_a") co-occurs twice: the correspondence of the first two a's co-occurs with the correspondence of the second two a's, and vice versa. When the numbers are relative, they add up to 1 in blocks where each block is an intersection of rows and columns whose names begin with the same segment, i.e. those which refer to the correspondences of the same segment. In the relative view, empty cells appear when the given correspondence never co-occurs, and therefore its relative frequency is 0 divided by 0.

Value

[table] The contingency table. The values represent how often the given correspondence co-occurs in the same word with the other correspondence or with the piece of metadata (cf. summary).

See Also

summary.soundcorrs, allCooccs

Examples

dataset <- loadSampleDataset ("data-abc")
coocc (dataset)
coocc (dataset, "DIALECT.L2")
round (coocc(dataset,"DIALECT.L2",count="r"), digits=3)

soundcorrs documentation built on Nov. 16, 2020, 5:09 p.m.