Date of Award

4-2016

Degree Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Linguistics

First Advisor

April Ginther

Committee Chair

April Ginther

Committee Member 1

Elaine J. Francis

Committee Member 2

Atsushi Fukada

Committee Member 3

Mary K. Niepokuj

Abstract

This paper examines temporal variables and pausing patterns in L2 English speech to investigate fluency as a measurable component of oral proficiency. Fluency can be defined as ‘speed and smoothness of oral delivery’. We can measure the speed of oral delivery through calculating temporal variables such as speech rate and mean syllables per run where ‘run’ is the vocal chunk between silent pauses. The smoothness of oral delivery can be measured through examination of pausing patterns by classifying the placement of pauses. Pauses may be placed in expected positions such as clause/phrase boundaries or in unexpected positions. Pause placement in unexpected positions may reduce the smoothness of oral delivery. The data sets are speech samples from the Oral English Proficiency Test (OEPT) but include the responses from two items (RAL: read aloud; NP: news passage). A total of 325 speakers across four different language groups (native speakers of Korean, Chinese, Hindi, and English) are represented across 6 proficiency levels (rated by holistic scoring based on the OEPT scale from 35 to 60). The speech samples were transcribed manually using a computer-assisted annotation tool that allowed capture of information about syllables, pausing boundaries, and types of pausing positions. Development of the annotation tool became a central concern of this study as establishing reliable and efficient methods in fluency research. Speech rate, mean syllables per run, and number of pauses per second were selected to examine temporal variables; number of unexpected pauses per second and expected pausing ratio were selected to compare pausing patterns across proficiency levels and language backgrounds. The results show that there are some linear relationships in temporal and pausing variables. High proficiency level speakers spoke at higher rates with expected pausing patterns compared to low proficiency level speakers who spoke at slower rates with almost no identifiable pausing patterns.

Download

Included in

Linguistics Commons

Share

COinS