Naomi Nagy

Linguistics at U of T

A Quantitative Categorization of Phonemic Dialect Features in Context

Naomi Nagy, Xiaoli Zhang, George Nagy, and Edgar W. Schneider

Addendum to article to appear in the Proceedings of CONTEXT '05, The Fifth International and Interdisciplinary Conference on Modeling and Using Context. Paris, July 2005

Addendum to: Section 5. Methods: Clustering and Mutual Information

As a simple example, consider the five varieties (1-5) shown in Table 2. The dissimilarity matrix is symmetric, so only the elements above the diagonal are shown. The formation of clusters can be represented by a dendrogram, as in Fig 1.

Table 2. Dissimilarity matrix for 5 varieties

Resulting clusters at 5 different thresholds
(including all words)

Varieties	1	2	3	4	5
1	0	0.36	0.40	0.46	0.50
2		0	0.47	0.44	0.49
3			0	0.50	0.55
4				0	0.55
5					0

Threshold (θ)	Clusters of varieties
0.55	{ 1 2 3 4 5 }
0.47	{ 1 2 3 }, { 4 5 }
0.40	{ 1 2 }, { 3 }, { 4 5 }
0.36	{ 1 2 }, { 3 }, { 4 }, { 5 }
0.00	{ 1 }, { 2 }, { 3 }, { 4 }, { 5 }

Fig. 1: Dendrogram for clusters in Table 2