Functional Load

Roughly speaking, Functional Load is a measure of the amount of use a language makes of a phonological contrast or rule.

Papers introducing the concept

• Quantifying the functional load of phonemic oppositions, distinctive features, and suprasegmentals by DS and Partha Niyogi : draft of book chapter to appear in Nedergaard Thomsen, Ole (ed. fc.) Current trends in the theory of linguistic change. In commemoration of Eugenio Coseriu (1921-2002). Amsterdam & Philadelphia: Benjamins.
• Quantifying the usefulness (functional load) of phonological contrasts (tech report) by DS and Partha Niyogi
• The Quantification of Functional Load (Masters Thesis) by DS.

Papers using the concept

• The functional load of Tone in Mandarin is as high as that of Vowels, DS and Gina-Anne Levow, Proceedings of the International Conference on Speech Prosody, Nara, Japan, 1:99-102, 2004.

Code

The C++ and Matlab code I used for the calculations (to call the documentation crap would be an understatement - I still have to change this).

The FL of several distinctive features and groups of phonemes in five languages

All these values are with word unigram models.

All values in percentages. So, for example, if one cant tell between consonants in German, the fraction of information lost (in a word unigram model) is 0.15. If one cant tell between consonants and between vowels and between tone in Mandarin, the fraction of information lost is 0.48, i.e. nearly half of it.

The higher the FL, the more important the contrast is.

	         Cantonese     Mandarin     Dutch       English        German
Consonants       12             8           19          18             15
Vowels            6             2            5           5              4
C + V            33            23           51          40             45
C + V + Tone     62            48     
C + V + Stress                              53          46             48

Obstruents        4.8           4.8          4.9         7.0            5.9
Sonorants        14             6           14          12              7
Obs + Son        29            26           36          32             29
Obs + Son + Tone 57            53      
Obs + Son + Stress                          37          40             31

Stress                                        .07         .01            .02
Tone              7             2.1          

Voicing                                       .31         .45            .11
Aspiration         .31           .27
Nasality           .12           .31          .15         .33            .18
Place             2.1           1.4          1.1         2.0            1.3
Manner            1.6            .6           .5         1.1             .8

Stops              .9            .6          1.1         1.1             .5
Affricates         .2            .5          0            .01            .00
Fricatives         .3            .5           .8         1.8            1.4
Nasals             .4            .3           .4          .2             .3
Labials            .3            .2           .9          .6             .4
Alveolars         2.5            .8          3.8         4.2            2.3

Nasality refers to being able to tell nasalized stops from the corresponding stops, e.g. ng from g. Nasals refers to being able to distinguish one nasal consonant from another nasal consonant.

Data for Dutch, English and German came from CELEX. Data for Mandarin from VOA Broadcasts (TDT version 2). Data for Cantonese from CANCORP, and as this is child-adult speech and much smaller, comparisons of its results with those of the other corpora should be made with additional salt.

FL of binary oppositions in English, based on syllable unigrams.