Aggregate Analysis of Vowel Pronunciation Introduction The Goal in - - PowerPoint PPT Presentation

aggregate analysis of vowel pronunciation
SMART_READER_LITE
LIVE PREVIEW

Aggregate Analysis of Vowel Pronunciation Introduction The Goal in - - PowerPoint PPT Presentation

Aggregate Analysis of Vowel Pronunciation in Swedish Dialects Mei-Shin Wu Aggregate Analysis of Vowel Pronunciation Introduction The Goal in Swedish Dialects Language leveling The aggregating methods Swedish Dialects Leinonen, Therese,


slide-1
SLIDE 1

Aggregate Analysis

  • f Vowel

Pronunciation in Swedish Dialects Mei-Shin Wu Introduction

The Goal Language leveling The aggregating methods Swedish Dialects

Data and Experimental Design

Data Acoustic analysis Analysis of dialectal variation

Result

Geographic dialect continuum Dialect leveling

Discussion And Conclusions

The observation The pros and cons of aggregation analysis

Reference

Aggregate Analysis of Vowel Pronunciation in Swedish Dialects

Leinonen, Therese, Vol 3, No 2 (2011) Mei-Shin Wu1

1Computational Linguistics

University of Tbingen

27th of January, 2016

1 / 32

slide-2
SLIDE 2

Aggregate Analysis

  • f Vowel

Pronunciation in Swedish Dialects Mei-Shin Wu Introduction

The Goal Language leveling The aggregating methods Swedish Dialects

Data and Experimental Design

Data Acoustic analysis Analysis of dialectal variation

Result

Geographic dialect continuum Dialect leveling

Discussion And Conclusions

The observation The pros and cons of aggregation analysis

Reference

Outline

Introduction The Goal Language leveling The aggregating methods Swedish Dialects Data and Experimental Design Data Acoustic analysis Analysis of dialectal variation Result Geographic dialect continuum Dialect leveling Discussion And Conclusions The observation The pros and cons of aggregation analysis

2 / 32

slide-3
SLIDE 3

Aggregate Analysis

  • f Vowel

Pronunciation in Swedish Dialects Mei-Shin Wu Introduction

The Goal Language leveling The aggregating methods Swedish Dialects

Data and Experimental Design

Data Acoustic analysis Analysis of dialectal variation

Result

Geographic dialect continuum Dialect leveling

Discussion And Conclusions

The observation The pros and cons of aggregation analysis

Reference

Outline

Introduction The Goal Language leveling The aggregating methods Swedish Dialects Data and Experimental Design Data Acoustic analysis Analysis of dialectal variation Result Geographic dialect continuum Dialect leveling Discussion And Conclusions The observation The pros and cons of aggregation analysis

3 / 32

slide-4
SLIDE 4

Aggregate Analysis

  • f Vowel

Pronunciation in Swedish Dialects Mei-Shin Wu Introduction

The Goal Language leveling The aggregating methods Swedish Dialects

Data and Experimental Design

Data Acoustic analysis Analysis of dialectal variation

Result

Geographic dialect continuum Dialect leveling

Discussion And Conclusions

The observation The pros and cons of aggregation analysis

Reference

The Goal

◮ Swedish dialects haven gone through massive leveling

in the latter half of the 20th century.

◮ The aggregating methods are used for identifying

geographic dialect areas and studying ongoing changes in Swedish dialects.

4 / 32

slide-5
SLIDE 5

Aggregate Analysis

  • f Vowel

Pronunciation in Swedish Dialects Mei-Shin Wu Introduction

The Goal Language leveling The aggregating methods Swedish Dialects

Data and Experimental Design

Data Acoustic analysis Analysis of dialectal variation

Result

Geographic dialect continuum Dialect leveling

Discussion And Conclusions

The observation The pros and cons of aggregation analysis

Reference

Outline

Introduction The Goal Language leveling The aggregating methods Swedish Dialects Data and Experimental Design Data Acoustic analysis Analysis of dialectal variation Result Geographic dialect continuum Dialect leveling Discussion And Conclusions The observation The pros and cons of aggregation analysis

5 / 32

slide-6
SLIDE 6

Aggregate Analysis

  • f Vowel

Pronunciation in Swedish Dialects Mei-Shin Wu Introduction

The Goal Language leveling The aggregating methods Swedish Dialects

Data and Experimental Design

Data Acoustic analysis Analysis of dialectal variation

Result

Geographic dialect continuum Dialect leveling

Discussion And Conclusions

The observation The pros and cons of aggregation analysis

Reference

Language leveling

◮ Dialect levelling is defined as the

reduction of structural variation

◮ Dialect levelling makes (a)

individual dialects more homogeneous, and (b) different dialects more similar and ,consequently, diasystems more homogeneous Standard DialectA

  • DialectZ

6 / 32

slide-7
SLIDE 7

Aggregate Analysis

  • f Vowel

Pronunciation in Swedish Dialects Mei-Shin Wu Introduction

The Goal Language leveling The aggregating methods Swedish Dialects

Data and Experimental Design

Data Acoustic analysis Analysis of dialectal variation

Result

Geographic dialect continuum Dialect leveling

Discussion And Conclusions

The observation The pros and cons of aggregation analysis

Reference

Outline

Introduction The Goal Language leveling The aggregating methods Swedish Dialects Data and Experimental Design Data Acoustic analysis Analysis of dialectal variation Result Geographic dialect continuum Dialect leveling Discussion And Conclusions The observation The pros and cons of aggregation analysis

7 / 32

slide-8
SLIDE 8

Aggregate Analysis

  • f Vowel

Pronunciation in Swedish Dialects Mei-Shin Wu Introduction

The Goal Language leveling The aggregating methods Swedish Dialects

Data and Experimental Design

Data Acoustic analysis Analysis of dialectal variation

Result

Geographic dialect continuum Dialect leveling

Discussion And Conclusions

The observation The pros and cons of aggregation analysis

Reference

The aggregating methods

◮ Feature-Based Variation Studies analyze variation in

  • ne element. While Aggregating methods consider

more than single feature.

◮ Both rural Swedish dialects and regional varieties of

Standard Swedish vary a lot when it comes to vowel pronunciation, and vowels have been important for characterizing varieties of Swedish and classifying dialects.

◮ are used for identifying geographic dialect areas and

studying ongoing change.

8 / 32

slide-9
SLIDE 9

Aggregate Analysis

  • f Vowel

Pronunciation in Swedish Dialects Mei-Shin Wu Introduction

The Goal Language leveling The aggregating methods Swedish Dialects

Data and Experimental Design

Data Acoustic analysis Analysis of dialectal variation

Result

Geographic dialect continuum Dialect leveling

Discussion And Conclusions

The observation The pros and cons of aggregation analysis

Reference

Outline

Introduction The Goal Language leveling The aggregating methods Swedish Dialects Data and Experimental Design Data Acoustic analysis Analysis of dialectal variation Result Geographic dialect continuum Dialect leveling Discussion And Conclusions The observation The pros and cons of aggregation analysis

9 / 32

slide-10
SLIDE 10

Aggregate Analysis

  • f Vowel

Pronunciation in Swedish Dialects Mei-Shin Wu Introduction

The Goal Language leveling The aggregating methods Swedish Dialects

Data and Experimental Design

Data Acoustic analysis Analysis of dialectal variation

Result

Geographic dialect continuum Dialect leveling

Discussion And Conclusions

The observation The pros and cons of aggregation analysis

Reference

Swedish Dialects

◮ South Swedish dialects ◮ Gotaland dialects ◮ Svealand dialects ◮ Norrland dialects ◮ Gotland dialects ◮ Finland-Swedish dialects

Figure: The traditional Swedish

dialect area according to Wessn

10 / 32

slide-11
SLIDE 11

Aggregate Analysis

  • f Vowel

Pronunciation in Swedish Dialects Mei-Shin Wu Introduction

The Goal Language leveling The aggregating methods Swedish Dialects

Data and Experimental Design

Data Acoustic analysis Analysis of dialectal variation

Result

Geographic dialect continuum Dialect leveling

Discussion And Conclusions

The observation The pros and cons of aggregation analysis

Reference

Outline

Introduction The Goal Language leveling The aggregating methods Swedish Dialects Data and Experimental Design Data Acoustic analysis Analysis of dialectal variation Result Geographic dialect continuum Dialect leveling Discussion And Conclusions The observation The pros and cons of aggregation analysis

11 / 32

slide-12
SLIDE 12

Aggregate Analysis

  • f Vowel

Pronunciation in Swedish Dialects Mei-Shin Wu Introduction

The Goal Language leveling The aggregating methods Swedish Dialects

Data and Experimental Design

Data Acoustic analysis Analysis of dialectal variation

Result

Geographic dialect continuum Dialect leveling

Discussion And Conclusions

The observation The pros and cons of aggregation analysis

Reference

Phonology data resource

◮ 98 rural sites ◮ from 1998 to 2001 ◮ 1170 speakers ◮ 12 people for each

location

◮ older group from 55 to

75 years

◮ younger group from 20

to 35 years

◮ 48 kHz sample rate

and 16-bit amplitude resolution (recording)

◮ downsampled to 16

kHz/16 bit for analysis

◮ swedia.ling.gu.se

12 / 32

slide-13
SLIDE 13

Aggregate Analysis

  • f Vowel

Pronunciation in Swedish Dialects Mei-Shin Wu Introduction

The Goal Language leveling The aggregating methods Swedish Dialects

Data and Experimental Design

Data Acoustic analysis Analysis of dialectal variation

Result

Geographic dialect continuum Dialect leveling

Discussion And Conclusions

The observation The pros and cons of aggregation analysis

Reference

Phonology data resource-2

◮ Vowel segments were elicited with mono- or bi-syllabic

words

◮ target vowels were surrounded by coronal consonants ◮ vowel recruitment criteria : long vowels, allophonic

variants of /E:/, /ø:/, vowels reflect the Historical development

◮ 19 words were used : dis[i:], disk[I], d¨

  • r[œ:], d¨
  • rr[œ],

flytta[Y], lass[a], lat[a:], leta[e:], lett[e fl], lott[O], lus[(0)], l˚ as/l˚ at[o:], l¨ ar[æ:], l¨

  • s[ø], n¨

at[E:], sot[u:], s¨ ark[æ], s¨

  • t[ø:], typ [y:]

13 / 32

slide-14
SLIDE 14

Aggregate Analysis

  • f Vowel

Pronunciation in Swedish Dialects Mei-Shin Wu Introduction

The Goal Language leveling The aggregating methods Swedish Dialects

Data and Experimental Design

Data Acoustic analysis Analysis of dialectal variation

Result

Geographic dialect continuum Dialect leveling

Discussion And Conclusions

The observation The pros and cons of aggregation analysis

Reference

Outline

Introduction The Goal Language leveling The aggregating methods Swedish Dialects Data and Experimental Design Data Acoustic analysis Analysis of dialectal variation Result Geographic dialect continuum Dialect leveling Discussion And Conclusions The observation The pros and cons of aggregation analysis

14 / 32

slide-15
SLIDE 15

Aggregate Analysis

  • f Vowel

Pronunciation in Swedish Dialects Mei-Shin Wu Introduction

The Goal Language leveling The aggregating methods Swedish Dialects

Data and Experimental Design

Data Acoustic analysis Analysis of dialectal variation

Result

Geographic dialect continuum Dialect leveling

Discussion And Conclusions

The observation The pros and cons of aggregation analysis

Reference

Acoustic analysis

◮ All vowel segments were filtered with Bark filters. ◮ principal component analysis(PCA) : PC1 is related

to vowel height and PC2 is related to tongue advancement

◮ males and females vocal tracts were normalized for by

applying PCA separately to data from male and females speakers.

15 / 32

slide-16
SLIDE 16

Aggregate Analysis

  • f Vowel

Pronunciation in Swedish Dialects Mei-Shin Wu Introduction

The Goal Language leveling The aggregating methods Swedish Dialects

Data and Experimental Design

Data Acoustic analysis Analysis of dialectal variation

Result

Geographic dialect continuum Dialect leveling

Discussion And Conclusions

The observation The pros and cons of aggregation analysis

Reference

Outline

Introduction The Goal Language leveling The aggregating methods Swedish Dialects Data and Experimental Design Data Acoustic analysis Analysis of dialectal variation Result Geographic dialect continuum Dialect leveling Discussion And Conclusions The observation The pros and cons of aggregation analysis

16 / 32

slide-17
SLIDE 17

Aggregate Analysis

  • f Vowel

Pronunciation in Swedish Dialects Mei-Shin Wu Introduction

The Goal Language leveling The aggregating methods Swedish Dialects

Data and Experimental Design

Data Acoustic analysis Analysis of dialectal variation

Result

Geographic dialect continuum Dialect leveling

Discussion And Conclusions

The observation The pros and cons of aggregation analysis

Reference

Analysis of dialectal variation

◮ multidimensional scaling (MDS) is a method for

reducing complex distance data. Gap statistic is used for estimating the number of significant clusters.

◮ MDS is suitable for visualizing dialect continua.And

cluster analysis detects dialect groups.

◮ This paper used Kruskal’s non-metric MDS ◮ Euclidean distance represents the vowel quality. ◮ x, y are either two different sites or two different

speaker groups.

17 / 32

slide-18
SLIDE 18

Aggregate Analysis

  • f Vowel

Pronunciation in Swedish Dialects Mei-Shin Wu Introduction

The Goal Language leveling The aggregating methods Swedish Dialects

Data and Experimental Design

Data Acoustic analysis Analysis of dialectal variation

Result

Geographic dialect continuum Dialect leveling

Discussion And Conclusions

The observation The pros and cons of aggregation analysis

Reference

Outline

Introduction The Goal Language leveling The aggregating methods Swedish Dialects Data and Experimental Design Data Acoustic analysis Analysis of dialectal variation Result Geographic dialect continuum Dialect leveling Discussion And Conclusions The observation The pros and cons of aggregation analysis

18 / 32

slide-19
SLIDE 19

Aggregate Analysis

  • f Vowel

Pronunciation in Swedish Dialects Mei-Shin Wu Introduction

The Goal Language leveling The aggregating methods Swedish Dialects

Data and Experimental Design

Data Acoustic analysis Analysis of dialectal variation

Result

Geographic dialect continuum Dialect leveling

Discussion And Conclusions

The observation The pros and cons of aggregation analysis

Reference

Geographic dialect continuum

  • ne dimension explains 81.4%, two dimension explains 93.6% and three dimension explains

96.3%

the dimension cannot be interpreted as PC1 and PC2 fully. 19 / 32

slide-20
SLIDE 20

Aggregate Analysis

  • f Vowel

Pronunciation in Swedish Dialects Mei-Shin Wu Introduction

The Goal Language leveling The aggregating methods Swedish Dialects

Data and Experimental Design

Data Acoustic analysis Analysis of dialectal variation

Result

Geographic dialect continuum Dialect leveling

Discussion And Conclusions

The observation The pros and cons of aggregation analysis

Reference

Geographic dialect continuum-2

coherent area : Sk˚ ane, Gotland, Uppland

incoherent area : Norrland, G¨

  • taland

20 / 32

slide-21
SLIDE 21

Aggregate Analysis

  • f Vowel

Pronunciation in Swedish Dialects Mei-Shin Wu Introduction

The Goal Language leveling The aggregating methods Swedish Dialects

Data and Experimental Design

Data Acoustic analysis Analysis of dialectal variation

Result

Geographic dialect continuum Dialect leveling

Discussion And Conclusions

The observation The pros and cons of aggregation analysis

Reference

Outline

Introduction The Goal Language leveling The aggregating methods Swedish Dialects Data and Experimental Design Data Acoustic analysis Analysis of dialectal variation Result Geographic dialect continuum Dialect leveling Discussion And Conclusions The observation The pros and cons of aggregation analysis

21 / 32

slide-22
SLIDE 22

Aggregate Analysis

  • f Vowel

Pronunciation in Swedish Dialects Mei-Shin Wu Introduction

The Goal Language leveling The aggregating methods Swedish Dialects

Data and Experimental Design

Data Acoustic analysis Analysis of dialectal variation

Result

Geographic dialect continuum Dialect leveling

Discussion And Conclusions

The observation The pros and cons of aggregation analysis

Reference

Dialect leveling

The extreme outlier is younger speakers of L¨

  • derup

22 / 32

slide-23
SLIDE 23

Aggregate Analysis

  • f Vowel

Pronunciation in Swedish Dialects Mei-Shin Wu Introduction

The Goal Language leveling The aggregating methods Swedish Dialects

Data and Experimental Design

Data Acoustic analysis Analysis of dialectal variation

Result

Geographic dialect continuum Dialect leveling

Discussion And Conclusions

The observation The pros and cons of aggregation analysis

Reference

Dialect leveling-2

A broad spectrum of colors is found in older group. While the younger group shows mainly green color

A large scale of dialect leveling can be confirmed.

The difference between younger speakers do exist! But the difference is so much smaller than between older speakers. Therefore, the color cannot reflect the differences precisely 23 / 32

slide-24
SLIDE 24

Aggregate Analysis

  • f Vowel

Pronunciation in Swedish Dialects Mei-Shin Wu Introduction

The Goal Language leveling The aggregating methods Swedish Dialects

Data and Experimental Design

Data Acoustic analysis Analysis of dialectal variation

Result

Geographic dialect continuum Dialect leveling

Discussion And Conclusions

The observation The pros and cons of aggregation analysis

Reference

Dialect leveling-3

the map shows the similarity distance.

the sites close to biggest cities, such as Stockhom and G¨

  • teborg shows a large ongoing changes

24 / 32

slide-25
SLIDE 25

Aggregate Analysis

  • f Vowel

Pronunciation in Swedish Dialects Mei-Shin Wu Introduction

The Goal Language leveling The aggregating methods Swedish Dialects

Data and Experimental Design

Data Acoustic analysis Analysis of dialectal variation

Result

Geographic dialect continuum Dialect leveling

Discussion And Conclusions

The observation The pros and cons of aggregation analysis

Reference

Dialect leveling-4

In order to present the difference between younger groups, MDS were applied to two age groups seperatly.

Two maps are independent.

A more uniform spoken variety of Norrland is emerging 25 / 32

slide-26
SLIDE 26

Aggregate Analysis

  • f Vowel

Pronunciation in Swedish Dialects Mei-Shin Wu Introduction

The Goal Language leveling The aggregating methods Swedish Dialects

Data and Experimental Design

Data Acoustic analysis Analysis of dialectal variation

Result

Geographic dialect continuum Dialect leveling

Discussion And Conclusions

The observation The pros and cons of aggregation analysis

Reference

Outline

Introduction The Goal Language leveling The aggregating methods Swedish Dialects Data and Experimental Design Data Acoustic analysis Analysis of dialectal variation Result Geographic dialect continuum Dialect leveling Discussion And Conclusions The observation The pros and cons of aggregation analysis

26 / 32

slide-27
SLIDE 27

Aggregate Analysis

  • f Vowel

Pronunciation in Swedish Dialects Mei-Shin Wu Introduction

The Goal Language leveling The aggregating methods Swedish Dialects

Data and Experimental Design

Data Acoustic analysis Analysis of dialectal variation

Result

Geographic dialect continuum Dialect leveling

Discussion And Conclusions

The observation The pros and cons of aggregation analysis

Reference

The observation

◮ The paper showed that the variation in vowel

pronunciation across Swedish dialects is continuous and abrupt dialect borders exist. And MDS result agree with the traditional division of Swedish dialects by Wess´ en.

◮ The dialect leveling has been confirmed statistically

by the aggregate analysis.

27 / 32

slide-28
SLIDE 28

Aggregate Analysis

  • f Vowel

Pronunciation in Swedish Dialects Mei-Shin Wu Introduction

The Goal Language leveling The aggregating methods Swedish Dialects

Data and Experimental Design

Data Acoustic analysis Analysis of dialectal variation

Result

Geographic dialect continuum Dialect leveling

Discussion And Conclusions

The observation The pros and cons of aggregation analysis

Reference

The observation-2

◮ A general lowering of front

vowels going on

◮ The areas where the vowel

shift is strongest are areas close to south-west of Stockholm and G¨

  • teborg

◮ The chain shift can be

regarded as simplification of the vowel system : the vowel inventory becomes smaller.

28 / 32

slide-29
SLIDE 29

Aggregate Analysis

  • f Vowel

Pronunciation in Swedish Dialects Mei-Shin Wu Introduction

The Goal Language leveling The aggregating methods Swedish Dialects

Data and Experimental Design

Data Acoustic analysis Analysis of dialectal variation

Result

Geographic dialect continuum Dialect leveling

Discussion And Conclusions

The observation The pros and cons of aggregation analysis

Reference

The observation-3

Figure: http://goo.gl/hbRDxa

◮ Not much change in vowel pronounciation : South Sweden

(province Sk˚ ane), Gotland and the Swedish dialect area in Finland.

◮ Aggregation of linguistic data reveals global tendencies and overall

relationships between varieties.

29 / 32

slide-30
SLIDE 30

Aggregate Analysis

  • f Vowel

Pronunciation in Swedish Dialects Mei-Shin Wu Introduction

The Goal Language leveling The aggregating methods Swedish Dialects

Data and Experimental Design

Data Acoustic analysis Analysis of dialectal variation

Result

Geographic dialect continuum Dialect leveling

Discussion And Conclusions

The observation The pros and cons of aggregation analysis

Reference

Outline

Introduction The Goal Language leveling The aggregating methods Swedish Dialects Data and Experimental Design Data Acoustic analysis Analysis of dialectal variation Result Geographic dialect continuum Dialect leveling Discussion And Conclusions The observation The pros and cons of aggregation analysis

30 / 32

slide-31
SLIDE 31

Aggregate Analysis

  • f Vowel

Pronunciation in Swedish Dialects Mei-Shin Wu Introduction

The Goal Language leveling The aggregating methods Swedish Dialects

Data and Experimental Design

Data Acoustic analysis Analysis of dialectal variation

Result

Geographic dialect continuum Dialect leveling

Discussion And Conclusions

The observation The pros and cons of aggregation analysis

Reference

The pros and cons of aggregation analysis

The advantage

◮ Aggregation gives a more reliable signal of provenance

than single features do.

◮ It allows us to examine the question whether dialectal

variation can best be described in terms of areas of continua. The limitation

◮ The linguist structure of the variation is not revealed ◮ The aggregate analysis does not reveal which linguistic

features characterize the dialect areas.

31 / 32

slide-32
SLIDE 32

Aggregate Analysis

  • f Vowel

Pronunciation in Swedish Dialects Mei-Shin Wu Introduction

The Goal Language leveling The aggregating methods Swedish Dialects

Data and Experimental Design

Data Acoustic analysis Analysis of dialectal variation

Result

Geographic dialect continuum Dialect leveling

Discussion And Conclusions

The observation The pros and cons of aggregation analysis

Reference

Reference

◮ Leinonen, Therese. ”Aggregate analysis of vowel

pronunciation in Swedish dialects.” Oslo Studies in Language 3.2 (2011).

◮ Hinskens, Frans. ”Dialect levelling: A two-dimensional

process.” Folia linguistica 32.1-2 (1998): 35-52.

◮ Auer, Peter. ”Dialect levelling and the standard varieties

in Europe.” Folia Linguistica 32.1-2 (1998): 1-9.

◮ Nerbonne, John. ”DataDriven Dialectology.” Language

and Linguistics Compass 3.1 (2009): 175-198.

◮ multidimensional scaling

http://www.bristol.ac.uk/media- library/sites/cmm/migrated/documents/chapter3.pdf

32 / 32