Disfluency and Proficiency in Second

Language Speech Production 1st

Edition Simon Williams
Visit to download the full and correct content document:
https://ebookmass.com/product/disfluency-and-proficiency-in-second-language-speec
h-production-1st-edition-simon-williams/
Simon Williams

Disfluency and Proficiency in Second

Language Speech Production
Simon Williams
School of Media, Arts and Humanities, University of Sussex, Falmer, UK

ISBN 978-3-031-12487-7 e-ISBN 978-3-031-12488-4

https://doi.org/10.1007/978-3-031-12488-4

© The Editor(s) (if applicable) and The Author(s), under exclusive

licence to Springer Nature Switzerland AG 2022

This work is subject to copyright. All rights are solely and exclusively
licensed by the Publisher, whether the whole or part of the material is
concerned, specifically the rights of translation, reprinting, reuse of
illustrations, recitation, broadcasting, reproduction on microfilms or in
any other physical way, and transmission or information storage and
retrieval, electronic adaptation, computer software, or by similar or
dissimilar methodology now known or hereafter developed.

The use of general descriptive names, registered names, trademarks,

service marks, etc. in this publication does not imply, even in the
absence of a specific statement, that such names are exempt from the
relevant protective laws and regulations and therefore free for general
use.

The publisher, the authors, and the editors are safe to assume that the
advice and information in this book are believed to be true and accurate
at the date of publication. Neither the publisher nor the authors or the
editors give a warranty, expressed or implied, with respect to the
material contained herein or for any errors or omissions that may have
been made. The publisher remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.

Cover Illustration: Marina Lohrbach_shutterstock.com

This Palgrave Macmillan imprint is published by the registered
company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham,
Switzerland
For Cynthia, Janet and Maureen
Transcription Symbols
Below are the main symbols used within the book. Additional symbols
are described in examples in certain chapters.

= Short (untimed) pause

(0.5) Silent pauses are represented in seconds and tenths of a second
within round brackets
: A colon indicates a prolongation of the preceding sound: the more
colons, the longer the prolongation.
? A question mark indicates a rising inflection.
| An upward arrow indicates a marked rise in intonation.
£ the pound sterling sign indicates a smiley voice.
>example< words that occur between the greater than sign and the
less than sign indicate talk delivered at a faster speed than the
surrounding talk.
<example> words that occur between the less than sign and the
greater than sign indicate talk delivered at a slower speed than the
surrounding talk.
( ) single round brackets containing no transcription indicate talk
that is too indistinct to transcribe
(( )) talk between double round brackets indicates some sort of
activity within the interaction.
Acknowledgements
I owe a huge debt of gratitude to Cathy Scott, Senior Editor at Palgrave
Macmillan, for suggesting I write this monograph and for reading
through first drafts. I’m also grateful to Asma Azeezullah, Project
Coordinator, and the production team for their care and efficiency. I
would like to thank Simon Cassar, Anne-Meike Fechter, Sebastian Loew,
Roberta Piazza, Kevin Porthouse, and Robin Williams for their support
and encouragement; Nina Studer and Charlotte Taylor for invaluable
practical advice; and Janet Lacey, Cynthia Leverton, and Maureen
Porthouse for their kindness throughout.
Acronyms
A2 (CEFR) post-beginner language level
ASU syntax-based Analysis of Speech Unit
B1 (CEFR) pre-intermediate language level
B2 (CEFR) post-intermediate language level
C1 (CEFR) advanced language level
CEFR Common European Framework of Reference for Languages
CSR Continuous Speech Recogniser – software capable of detecting
and processing meaningful natural language, e.g. based on vocabulary
items, that allows a suitable response to be automatically generated
and enables human-computer interaction to take place based on this
principle
DELE Diploma de Españ ol como Lengua Extranjera
EAIS Equal Appearing Interval Scale - a form of attitude
measurement that allows quantitative findings to be reported on
qualitative variables such as attitude
EEG Electrical Geodesics here refers to a non-invasive proprietary
head net containing electrodes sensitive to small changes in voltage
fields generated within the brain that is widely used in neuroscience
studies
EIT Elicited imitation test—a test in which listeners repeat samples
of language, e.g. sentences, as accurately as possible
ERP Event-related potential is a measurable electrical response in the
brain to a cognitive, motor, or sensory stimulus
ETS Educational Testing Service—a non-profit company specialising
in educational testing and assessment
FP Filled pause
FS False start
GECO Ghent Eye-Tracking Corpus
IELTS International English Language Testing System
L1 First language
L2 Second language
LFP Lengthening, Fragment, Pause (filled)
MLS Mean length of syllable
MOS Mean opinion score
N400 an electrical signal detectable in the brain in response to
stimuli that include signs in the form of words, sounds and images. It is
a form of ERP that peaks at 400 ms after onset.
NS Native speaker, i.e. speaker of an L1
NNS Non-native speaker, i.e. speaker of an L2
OET Occupational English Test
OISR Other-initiated self-repair
PRL Prolongation
PTE Pearson English Language Tests
RPT Repetition
SC Self-correction
SISR Self-initiated self-repair
SITAF Spécificités des Interactions verbales dans le cadre de Tandems
linguistiques Anglais-Français—verbal and non-verbal corpus of
conversational French
SLA Second language acquisition
SP Silent pause
SPEAK Test the Speaking Proficiency English Assessment Kit
developed by the Educational Testing Service (ETS)
TOEFL iBT Test of English as a Foreign Language Internet-based Test
TOEIC Test of English for International Communication
TRP Transition relevance place in conversation
UCLES University of Cambridge Local Examinations Syndicate
WISP What is Speaking Proficiency corpus
Contents
1 Introduction
1.​1 Disfluency as Fluency
1.​2 Introduction to the Six Disfluencies:​Formal Descriptions
1.​3 Effects on Listeners
1.​4 Main Types of Disfluency Data
1.​5 Disfluency Types and Proficiency Level
1.​6 Disfluencies in Public Language Test Descriptors
1.​7 Comment
1.​8 Overview
References
2 Silent Pauses
2.​1 Introduction
2.​2 Formal Description of Silent Pauses
2.​3 Effect of Silent Pauses on Listeners
2.​4 Silent Pauses in Classroom Interaction
2.​5 Silent Pauses and Proficiency Level
2.​6 Silent Pauses in Public Language Tests
2.​7 Comment
References
3 Filled Pauses
3.​1 Introduction
3.​2 Formal Description of Filled Pauses
3.​2.​1 Location/​Position
3.​2.​2 Frequency
3.​2.​3 Length/​Duration
 3.​2.​4 Pitch
3.​3 Effect of Filled Pauses on Listeners
3.3.1 uh and um Studies
3.​3.​2 Second Language Rater Studies
3.​4 Speech Environment of Filled Pauses
3.​5 Filled Pauses and Speaker Proficiency
3.​6 Filled Pauses in Public Language Tests
3.​7 Comment
References
4 Prolongations
4.​1 Introduction
4.​2 Formal Description of Prolongations
4.​3 Effect of Prolongations on Listeners
4.​4 Main Types of Prolongation Data
4.​4.​1 Corpora Studies
4.​4.​2 Natural Language Processing
4.​4.​3 Prolongations in Classroom Interaction
4.​4.​4 Prolongations in Elicited Data
4.​5 Prolongations and Proficiency Level
4.​6 Prolongations in Public Language Tests
4.​7 Comment
References
5 Repetitions
5.​1 Introduction
5.​2 Formal Description of Repetitions
5.​3 Effect of Repetitions on Listeners
 5.​4 Repetitions Outside the Classroom
5.​5 Repetitions in Elicited Data
5.​6 Task Effect on Repetitions
5.​7 Repetitions and Proficiency Level
5.​8 Repetitions in Public Language Tests
5.​9 Comment
References
6 Self-Corrections
6.​1 Introduction
6.​2 Formal Description of Self-Corrections
6.​3 Effect of Self-Corrections on Listeners
6.​4 Main Types of Self-Correction Data
6.​4.​1 Self-Corrections Outside the Classroom
6.​4.​2 Self-Corrections in Classroom Interaction
6.​4.​3 Task Effect on Self-Corrections in Elicited Data
6.​5 Self-Corrections and Proficiency Level
6.​6 Self-Corrections in Public Language Tests
6.​7 Comment
References
7 False Starts
7.​1 Introduction
7.​2 Formal Description of False Starts
7.​3 Effect of False Starts on Listeners
7.​4 Main Types of False Starts Data
7.​4.​1 False Starts Outside the Classroom
7.​4.​2 False Starts in Classroom Interaction
 7.​4.​3 Task Effect on False Starts in Elicited Data
7.​5 False Starts and Proficiency Level
7.​6 False Starts in Public Language Tests
7.​7 Comment
References
8 Conclusion
8.​1 Introduction
8.​2 Formal Description of Disfluencies
8.​3 Effects of Disfluencies on Listeners
8.​4 Disfluencies in Classroom Interaction
8.​5 Disfluencies and Proficiency Level
8.​6 Disfluencies in Public Language Tests
8.​7 Comment
Appendix
A
B
C
References
Index
List of Figures
Fig.​1.​1 Framework for L2 (dis)fluency and proficiency (based on
Segalowitz, 2016)

Fig.​2.​1 Rating scale for Experiment 2 (Bosker et al.​, 2013, p.​175)

Fig.​3.​1 The nine-level Equal Appearing Interval Scale completed by

participants in Experiment 1 (Bosker et al.​, 2013, p.​175)

Fig.​5.​1 Incidence of repetition forms (based on Maclay and Osgood,

1959)

Fig.​6.​1 Overview of the speech-processing monitor and repair as

comprehensible output (based on Levelt, 1989, p.​9)
List of Tables
Table 2.​1 References to SPs in rating scale for individuals (based on
Sato 2014, p.​85)

Table 2.​2 References to SPs in rating scale for interaction (based on Sato
2014, p.​86)

Table 2.​3 L2 studies reporting effect of SPs on listeners’ judgements of

fluency (studies inviting raters’ additional qualitative comments on the
speech samples are italicised)

Table 2.​4 Number of SPs per 60 seconds and total silent pausing time
(based on Iwashita et al.​, 2008, p.​40)

Table 2.​5 Proficiency test results (based on García-Amaya, 2015, p.​24)

Table 2.​6 SPs by gender and level in Mandarin Chinese (L2) speakers
(based on Yuan et al.​, 2016, n.​p)

Table 2.​7 Utterance and clause boundaries with SPs, language, and
English (L2) proficiency (percentages) (adapted from Rose, 2017, p.​50)

Table 3.​1 Duration of FPs vs silent pauses within speaker

reformulations (based on Williams &​Korko, 2019)

Table 4.​1 Ten most frequent disfluent word prolongations in Betz et al.​
(2016)
Table 4.​2 Participants’ gender and age in Bosker et al.​(2013)

Table 4.3 Percentage of pauses and fillers preceding and following thee
and thi:y (Fox Tree & Clark, 1997, p. 158)

Table 4.​4 Filled pause tokens in four corpora in Clark &​Tree (2002)

Table 4.​5 Key differences in filled pauses vs prolongations in Moniz et

al.​(2007)

Table 5.​1 Three subclasses of disfluent repetition in Plauché &​Shriberg

(1999)

Table 5.​2 Repetitions in post-semester interview home and abroad

participants (extract adapted from Table 8.​Comparisons of repair
phenomena for at home and abroad groups, Freed, 1995, p.​141)

Table 6.​1 CEFR level comparability with OET and Aptis tests

Table 7.​1 False starts (T-test Mean totals) in four picture narratives
(Tavakoli &​Foster, 2011; Foster &​Tavakoli, 2009)

Table 7.​2 Repairs in Van Hest (1996)

Table 8.​1 Disfluencies and errors in three semi-spontaneous
monologues

Table 8.​2 Measures of speaker speed and breakdown fluency to achieve

consistency (but not statistical significance) across proficiency levels in
the Aptis speaking test
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022
S. Williams, Disfluency and Proficiency in Second Language Speech Production
https://doi.org/10.1007/978-3-031-12488-4_1

1. Introduction
Simon Williams1
(1) School of Media, Arts and Humanities, University of Sussex, Falmer,
UK

Simon Williams
Email: s.a.williams@sussex.ac.uk

1.1 Disfluency as Fluency

Disfluency is a relatively recent construct. Not until the last century did
fluency with the specific meaning of ease of speaking become a subject
of widespread interest; and only in the mid-twentieth century did its
corollary, the absence of such a facility, become the focus of systematic
empirical study. Although the concepts of fluency and disfluency have
been applied in a metaphorical sense to a number of human, and
occasionally non-human, activities, the locus of much investigation has
been the linguistically productive areas of speech and writing. While
disciplines such as medicine and neuroscience have taken an interest in
the pathology of dysfluent speech in physiological conditions such as
aphasia, apraxia, and dysarthria, linguists and psycholinguists have
sought to understand and analyze the mundane occurrences of the non-
pathological disorders of articulation. Six of those resulting from social-
psychological conditions have become standard psycholinguistic
typologies of speech disfluency: repetitions, filled and silent pauses,
prolongations, self-corrections and false starts.
Disfluency remains a relatively specialised area of investigation.
Perhaps this is not surprising when recognising that the study of
effortless behaviour has always been of more interest than its effortful
equivalent. Nevertheless, an examination of the components signalling
breakdown can reveal a great deal about the economy and detail of
elements that are in more optimal relation to each other. In human
behaviour, those associations may offer insights into the mechanics of a
competence like language, into the workings of the brain, and into
human relationships.
If fluency has its own markers, such as fast speech, long turns,
accurate syntax, and syntactic complexity, it also demands a repertory
of strategies for achieving them, and for dealing with their
consequences: the complex syntax that loses its way, the speed that
produces breakdowns in articulation, the accuracy that becomes
pedantic, and the turn that never ends. As social interactants, speakers
have to consider the effect on the listener and beware of interruptions,
non-sequiturs, mistaken inferences, and other hazards, and intervene in
the stream of speech to change course when necessary. Although
(dis)fluency studies frequently cite Lennon’s (1990) distinction
between the ‘broad’ and ‘narrow’ senses of the construct (p. 388),
broad meaning speaking proficiency in general and narrow meaning
aspects such as ‘correctness, idiomaticity, relevance, appropriateness,
pronunciation, lexical range’ (p. 389), few are careful to define where
speaking proficiency ends and more general language proficiency
begins. In other words, while the narrow sense of speaking fluency
might refer to (lack of) silent and filled pauses, repetitions, self-
corrections and so on, and the broader sense to these absences plus
pronunciation, grammatical accuracy, and vocabulary range, the
proficiency here still refers to speech and not to facility in other
language activities such as listening, reading, and writing. So while
‘fluent in English’ (p. 389) might mean proficient in speaking the
language, it could also be generalised in one direction to mean
proficiency in all uses of the language, or in another as speaking the
language in a particularly flawless, non-disfluent way.
Research in this area has followed a similar course to other objects
of linguistic attention, from early identification of disfluency as a novel
candidate for investigation to the increasing discrimination of
exemplars, an appreciation of their mutual relationship and functions,
and the extension of empirical studies to new populations. An initial
focus on first language speakers expanded to second language speakers
and thence to natural language processing for machine learning. Early
work on one kind of disfluency (Goldman-Eisler, 1968: pauses) moved
to investigation of other kinds (Fox Tree, 1995: false starts) and the
development of taxonomies (Levelt, 1983); and this approach was
extended to research on second language disfluency (Kormos, 1998) in
speakers of first languages other than English (Crible & Pascual, 2020)
and to comparative studies of the disfluencies of bilingual speakers in
both languages (Riazantseva, 2001). In addition, the influence of
contextual variables has been studied, including task type (Ahmadian et
al., 2012; Gilabert, 2007; Foster & Skehan, 1996), cognitive load
(Skehan, 2009), gender (Bortfeld et al., 2001), culture (Swerts, 1998),
and listener (Brennan & Schober, 2001; MacGregor et al., 2009).
The general direction of study has been from highly controlled
elicitation of data within laboratory settings with first language
participants being taken as a given to more naturalistic, less controlled
data and settings, and thence to human-machine interfaces. Within L2
research, the classroom has served as the equivalent to the laboratory
and study has not generally gravitated beyond. The focus reflects the
overriding concerns of L2 researchers with pedagogic environments to
the extent of the relative neglect of naturalistic settings with their
inferred dangers for L2 learners such as fossilization (e.g. Schmidt,
1983).
Disfluency research began by studying monologic data—elicited by
read-aloud prompts (Goldman-Eisler, various) and static visual
networks (Levelt, 1983). Only later did investigators turn their
attention to the analysis of spoken interaction and acknowledge the
role of the addressee in the production of disfluent speech. For
example, Clark (1994), following Schegloff et al. (1977), noted that
disfluencies were joint problems for interactants and had three
purposes: prevention, warning and repair (Clark, 1994). Forms of
hesitation have also been found helpful to conversationalists by alerting
the listener to expect a next word or utterance from the speaker that is
either unusually complex or marked in some other way. Certain forms
of hesitation such as the filled pause, may be culturally defined, e.g. the
function of the filled pause has been found to differ according to the
speakers’ first language (Schmid & Fä gersten, 2010; Crible & Pascual,
2020); and to vary with gender, e.g. men make more use of the filled
pause than women when they are searching for a word and want to
hold the floor (Shriberg, 2001). The two types of interactional
approach, i.e. the language production approach and the social
interaction approach (Faerch & Kasper 1983, 1984; and see Chiang &
Mi, 2011 for a discussion) represent a long-standing, cognitive-social
debate on second-language acquisition (SLA) (cf. Firth & Wagner, 1997,
2007; Gass, 2004).
Perhaps because of the high demands of research time and, apart
from test providers and certain government agencies, the paucity of
funding for most spoken language research, questions regarding
individual differences, noted by Engelhardt et al. (2019), and the
speech style and stock of disfluencies produced by the same speakers in
different conditions and in response to social judgements, remain
under-researched. The range of fluencies demanded and codified by
cultures was early recognised in first language studies: the fluency of
the DJ is different from that of the diplomat, which is different again
from the after-dinner speaker, though all three would be acknowledged
as notably fluent in their professional lives. Influenced in this way by a
mix of speech style and cultural norms, disfluencies as conversational
markers become rhetorical strategies, associated both with speech
genres and with personal identity.
The six kinds of disfluency discussed in these chapters have all been
classed as hesitation (Maclay & Osgood, 1959), with silent or unfilled
pauses occurring most frequently.
Perhaps the most recognizable feature associated with the absence
of fluency is pausing. It is present to a greater or lesser extent on its
own and as part of other forms of dysfluency. Many speakers interrupt
the flow of speech, creating a temporary hiatus and a moment of
silence. Silent breaks occur often in conversation and if less than 0.25
milliseconds are not normally noticed by an interlocutor. Pauses of
longer duration, however, are marked and are likely to cause the
interlocutor to search for an explanation and make a judgement about
the quality of the exchange. For example, speaker pauses to conduct a
word search have been shown to alert the listener to the less ordinary
nature of the following word, aiding mutual understanding and
contributing to the success of the discourse.
 As another form of codified description of (dis)fluency, the level
descriptors of well-known English language tests are useful in
illustrating this point. With first and second language development,
disfluencies never wholly disappear but rather become invisible
through cultural standardisation. Thus, the remarkable mid-juncture
silences of the L2 beginner migrate towards end-juncture and become
longer though less noticeable. Filled pauses become lexicalised, e.g. like
and yeah, and function words may be imperceptibly prolonged as the
speaker searches for a low frequency content word. Function words
and phrases may also be repeated in hardly noticeable close succession
for the same reason. The L2 learner becomes adept at self-correcting
from the previously articulated function word, so contextualising the
repair and minimising the mental work of the listener. The same
principle applies to false starts, with the added advantage to the
discourse that the repair is information more obviously intended for
the benefit of the listener at a here-and-now level of interaction rather
than an attempt to conform to more general social norms of language
accuracy. Eventually, the L2 (dis)fluency patterns are indistinguishable
from those of an L1 speaker, save that wide individual variation in the
use of filled and silent pauses and repetitions, all delaying tactics, is
well-recognised and accepted.

1.2 Introduction to the Six Disfluencies: Formal

Descriptions
The nature of disfluent production and its relation to general
proficiency in speaking is the subject of the next six chapters. The
object of enquiry is first one of definition, description, and discovery,
and then its application to the wider spoken situation. The survey that
occupies most of these pages collects a range of studies from sub-areas
of spoken fluency, each with its own agenda, terminology, and concerns,
and responds to the ‘So what?’ question by understanding (dis)fluency
as a social project in which speakers are constantly adapting to the
norms that surround them, and in their turn contributing to those
norms. An illustration might be the previously mentioned lexical filler
like, a disfluency that functions to give the speaker more time in the
planning process and is also a discourse marker conveying a particular
speaker identity and contributing to fluent speech by maintaining the
flow of talk.
If a central notion associated with fluency is keeping the talk going,
then disfluency is the opposite, a temporary interruption to the flow of
speech. The speaker is faced with an ongoing tension between planning
and production: keep going and hesitate from time to time to plan the
next stretch of talk or carry on regardless and repair the inevitable
errors and breakdowns. Thus, fluency has become synonymous with
speed, and disfluency with forms of hesitation and repair. Four
recognised forms of hesitation are silent pauses, filled pauses,
prolongations, and repetitions. Two forms of repair are self-corrections
and false starts. These six disfluencies are the subject matter of the
central chapters. The modern history of their investigation and the
nature and amount of interest in each varies considerably. Common to
each chapter is a general introduction to the disfluency, its formal
description, effect on listeners, relation to speaker proficiency,
significance in international language tests, and a final comment.
Interest in silent and filled pauses goes back to Goldman-Eisler’s
work in the early 1950s; and analysis of what they define as hesitations
types: silent and filled pauses, repetitions and false starts, to the
seminal study of Maclay and Osgood (1959). Prolongations in non-
pathological speech are the most youthful of the research areas in
disfluencies (e.g. Eklund, 2001), although in relation to L1 stutterers
prolongations have a much longer history. Similarly, studies in repair
usually cite Schegloff, Jefferson and Sacks’ (1977) paper, and disfluency
studies in particular cite Levelt (1983), who distinguishes five subtypes
in Dutch (L1) data that include self-corrections and other forms of
repair recognisable as false starts. Notable so far is the focus on L1
speech. It was van Hest (1996) in her empirical study who applied
Levelt’s (1983) taxonomy to the L1 and L2 speech of her participants.
Common to all these disfluency specialisms is the interest of the
enquirers in solving linguistic problems in a specialist area. For
example, the interest in non-pathological prolongations as disfluencies
is shared by a group of specialists in human-computer communication
such as Eklund (human-computer travel booking systems) and Betz
(spoken dialogue systems). The purpose of their enquiries is to
improve the increasingly common human-machine communication
experience by introducing disfluencies like fillers and prolongations
into machine-simulated natural language to make it sound more
natural and increase the time available for processing responses. If any
evidence were needed that the status of disfluencies has been revised,
one example is their inclusion in speech synthesis, e.g. filled pauses
(Adell et al., 2010), in order to simulate spontaneous rather than read
speech.
The interest in second language self-corrections is concentrated
particularly in language teaching practitioners and specialists keen to
maximise the acquisition of the L2 by learners. Evidence exists that at
an early stage, language learners prioritise acquiring the formal rules of
the target language and devote considerable energy to eliciting models
from expert others, gradually internalising them.
Silent pauses (SPs) as stretches of silence within the speech stream
occur naturally at clause and sentence boundaries but are heard as
disfluencies when they interrupt a word, phrase or clause. Pause
duration is less significant in this regard, though longer boundary
clauses are associated with longer following sentences (Goldman-Eisler,
1972). In this way, pauses signal an impending information increase,
and words following pauses are found to be less predictable. Words
that are not preceded by pauses are therefore more predictable, and
fluency as absence of mid-clause pausing is associated with
predictability or word frequency, providing some justification for
encouraging learners to include formulaic sequences in their speech as
a means to increase fluency. Pauses can be lengthy at boundaries
without sounding marked. Thus, the distribution of pauses is more
significant than their duration.
Filled pauses (FPs) can be defined as voiced hesitation such as er
and erm in English and include lexical fillers such as yeah, like and you
know that carry no additional information. Speakers may produce filled
pauses for at least three reasons, some of which overlap, depending on
the situation. Of interest to psycholinguists is the filled pause as marker
for a word search. Interactionists are concerned with the filled pause as
back-channelling. And discourse analysts will note the turn-taking
function of filled pauses. As with silent pauses, work has centred on the
distribution, duration, and frequency of filled pauses. And like silent
pauses, the unmarked location of filled pauses is at major boundaries,
where they might indicate either a desire to continue speaking or an
invitation to the listener to take the floor, depending on pitch and
intonation. Again, like silent pauses, filled pauses are more likely to
occur before a low-frequency content word and before longer clauses;
therefore, like silent pauses in this regard, they indicate syntactic
uncertainty. The duration of FPs is markedly shorter than SPs and
produced at lower pitch than the surrounding words.
Prolongation is the continuation of the immediately prior phoneme
and is classed as disfluent when it continues for longer than usual
relative to similar phones produced by the same speaker. The
prolonged sound and the usual length of production can be measured
digitally for comparison. Prolongations are one way of the speaker’s
achieving extra time for planning and information processing without
resorting to a more noticeable repetition, filled pause, or silent pause.
In this sense, prolongations may be the first option for speakers who
need extra time. Three-quarters of disfluent prolongations occur in
function words rather than content words. Speakers’ preferred location
for prolongations is the long vowel in the nuclei of syllables but the
location is capable of migrating, e.g. to a sonorant in the coda, when
options are reduced.
Repetitions are contiguous repeats of the same sound, syllable,
word, or phrase that convey no additional meaning for the listener and
are more likely to be heard as disfluent than discontinuous repeats, i.e.
repeats separated by some intervening language. Like the location of
prolongations, repetitions are more likely to be function words that
occur before content words, presumably allowing the speaker time to
search for the lexical item. Repetitions are most often of single words,
occasionally more, and rarely less. Like filled pauses, repetitions are
often found near to, or at the start of, an utterance (Schegloff, 1987),
possibly to repair speaker overlaps. By analysing intonation and pause
patterns surrounding repetitions, Plauché and Shriberg (1999) offer a
model of three repetition types: canonical, covert self-repair, and
stalling, illustrated with the definite article representing a function
word. As the term suggests, canonical repetition is the most numerous
of the three subtypes, and the repetition, retrospective in nature,
provides some continuity for the listener. Covert repetition is rather
prospective, its rising intonation expressive of the speaker’s planning
uncertainty. The function of stalling repetitions may be to keep the floor
during planning problems and they seem to be simultaneously
prospective and retrospective.
A self-correction is the speaker’s replacement of non-standard
output with an alternative supposed to be standard. The concern with
standardisation makes self-corrections different from false starts, the
other type of self-repair. Answering Schegloff et al.’s call for ‘an account
of the organization of repair’ (Schegloff et al., 1977, p. 381), Levelt
(1983) describes a three-part model, which applies equally to self-
corrections and false starts. It consists of a problem source, an editing
phase characterised by a silent pause and which may include an editing
term such as a filler, and a revision or repair of the trouble source. To be
well-formed and not impose an additional burden on the listener, the
speaker needs to repeat in the revision any words that followed the
problem source that were articulated before the interruption of the
editing phase, and optionally any words immediately in front of the
problem source, especially if they are function words and the problem
is a content word. In contrast to false starts, most self-corrections
repair only the problem element. The repetition of language around the
problem source, mandated where words follow the problem source,
provides evidence that the syntactic structure is held in the speaker’s
working memory and is available for the revision.
False starts may be defined as the speaker’s abandonment of an
apparently standard utterance to start afresh or continue in a different
direction. The three-part structure of false starts is as described for
self-corrections, consisting of problem source, editing phase, and
revision. An important difference is that if the speaker had not
commenced the editing phase, the listener would not have recognised
any need for a revision. This is particularly the case with so-called
‘covert repairs’ (Levelt, 1983, p. 45), where a word is simply repeated
after an intervening editing term such as a silent or filled pause, or the
utterance continues following an editing phase, apparently in the same
direction. Because no standard form exists for the speaker’s false start
to orient to, the editing phase and revision are likely to be longer and
more complex than a self-correction (Kormos, 2000). On the other
hand, while this complexity and the interruption to the speech flow it
causes is what makes false starts disfluent, the speaker’s orientation
may be wider than finding the optimum expression for a thought. It
may signal an adjustment to the discourse for the mutual benefit of
speaker and listener. Certain other factors, such as the well-formedness
of the repair, are associated with proficiency.

1.3 Effects on Listeners

SPs are invariably included in quantitative acoustic measures in rater
studies and their location at non-clause boundaries is significant. SP
mean length is less important to raters’ fluency assessments than their
frequency (Cucchiarini et al., 2002). Their importance in fluency
judgements is underscored by their accounting for the largest
percentage of any temporal variable in listeners’ negative impressions
of fluency (Derwing et al., 2004). In monologic speech production,
raters seem to interpret SPs as L2 speaker processing problems. In
dialogic speech production, in contrast, raters seem to interpret SPs as
a turn-taking phenomenon rather than a disfluency. Speakers who
produce fewer SPs have been found to pause for longer, and longer SPs
may imply content planning. Shorter SPs, on the other hand, may imply
breakdowns (Préfontaine et al., 2016). Slow speech may be judged
fluent in itself, but the addition of SPs results in its being perceived as
unduly slow and being awarded a lower fluency rating (van Os et al.,
2020). In a meta-analysis of rater studies, Suzuki et al. (2021) find
found no association between length of pause and perceived disfluency.
The negative effect of clause-internal pauses and pause frequency,
which may signal syntax building and word retrieval problems, on the
other hand, was strong.
FPs have been shown to influence listeners’ expectations of a
following speaker referent, and to be heard as disfluencies when not
conforming to target language placement, i.e. end clause, and
lexicalisation, e.g. well, like, you know (English L1). Discussion on the
signalling effect of FPs concerning the identification of approaching
referents as low- rather than high-frequency centres on speaker
intentionality and listener inference. Experiments by Bosker et al.
(2014), Fraundorf and Watson (2011), Watanabe et al. (2008) and
others indicate that FPs are signals rather than signs, and listeners
must learn to recognise the associations of speaker difficulty with
cognitive or environmental challenges. In this respect, Bosker et al.
(2014) report that when listeners heard disfluent L2 instructions, their
anticipatory effect was neutralised and they no longer inferred the
approach of a low-frequency referent. Rather, they reassessed the
speaker’s overall competence instead of assuming a local and
temporary challenge. The results of rater studies indicate the
significance to listeners of FPs. Together with SPs, FPs were responsible
for 59% of listener judgements of temporal disfluency in Rossiter
(2009); but Cucchiarini et al. (2002) conclude from their findings that
FPs are a poor indicator of fluency and may instead be a strategy
reflecting sentence planning. Perhaps FPs rather reflect individual
differences or broader cultural norms.
Prolongations in the form of longer syllables have been found to
occur before the vast majority of syntactic junctures and are also
typical of grammatical junctures, where length of syllable is generally
longer, as it is before rather than after pauses (Martin, 1970). Listeners
identify prolongations as disfluencies yet tolerate them well. In rater
studies, listeners prefer prolongations to compound disfluencies (Betz
et al., 2015) and, consistent with their effect at grammatical junctures,
rate prolongations that occur within conjunctions highly. Prolongations
are thus a means of preserving fluent speech while planning more
complex language (Moniz et al., 2007). Disfluent prolongations, i.e.
certain extended marked forms, especially in function words such as
thee for the, have been shown to prime listeners for a following referent
that is new or unfamiliar (Arnold et al., 2003). Listeners’ response
seems to be inferentially learned rather than associative because when
prompts are given by an L2-accented speaker, the attenuation effect
disappears, i.e. listeners show no preference for old or new referents
following a prolongation. It may be that the wide variation in length of
prolongations is what makes them notably acceptable to listeners.
Furthermore, because speakers seem to plan ahead during longer
prolongations, overall speed is not reduced (Betz et al., 2017).
Like FPs, repetitions may serve as planning devices for speakers
(Lennon, 1990). Similarly, when speakers repeat a function word such
as an article, they may be conducting a word search for a following
noun (Temple, 1992). The fact that repetitions have been reported to
increase in number in study abroad groups, presumably as a result of
pressure on speakers to keep pace in interaction with L1 speakers
(Freed, 1995), reveals the consequences of the additional real-time
planning necessary in the new speech environment. Although listeners
are often oblivious to repetitions in interaction and repetitions impose
no additional burden on listener comprehension, their Event Related
Potential (ERP) effect, recorded by Electrical Geodesics (EEG), is similar
to prolongations and consistent with semantically incongruous input
(McAllister et al., 2001). The incongruity may explain why listeners in
rater studies see them as a key indicator of disfluency (Rossiter, 2009).
In general, self-corrections seem not to influence listeners’
judgements of speaker fluency: although listeners are sensitive to
repairs, their judgement is barely affected (Bosker et al., 2013). An
exception is reported in McRobie (1993), whose raters awarded higher
fluency scores to non-corrected vs corrected speech. Expert L1 and L2
raters’ written comments as they audited L2 speech showed that they
consistently reacted more negatively to self-corrections than false
starts (Rossiter, 2009). To sum up, where self-correction is a named
fluency measure in listener rating studies, it hardly figures in raters’
perceptions.
False starts are another disfluency that holds meaning for listeners,
in this case by recycling material to create ‘discourse coherence’
(Levelt, 1983, p. 42). ERP data shows that false starts also involve a
processing cost (McAllister et al., 2001) and slow monitoring times
(Hindle, 1983), which becomes evident from the more numerous false
starts of beginner and intermediate-level speakers compared to
advanced-level speakers. Lower level speaker FSs also tend to be ill-
formed (van Hest, 1996), possibly reflecting working memory capacity.
Although advanced learners produce fewer false starts, for that reason
their false starts may be more distracting. Comprehension is also
retarded by the occurrence of false starts in the middle of an utterance,
and near the start but introduced by a conjunction, rather than without
any marker at the beginning. In general, however, rater studies reveal
that although listeners notice false starts, they are one of the
disfluencies least likely to elicit a negative reaction (Rossiter, 2009).

1.4 Main Types of Disfluency Data

For the purposes of this survey, the data for disfluency studies is
categorised into five types: classroom interaction, corpora, elicited data,
natural language processing, and natural data outside the classroom.
The division cannot be watertight and anomalies remain, with
ambiguity, overlap, and contradiction. Some categories are confined to a
single chapter, and no category is common to all. However, even when a
data type is absent from the central section, it may feature within one of
the other sections. For example, no separate section exists for elicited
data in the chapter on silent pauses. Yet elicited data is central to the
rater studies reported within the ‘effect on listeners’ section in the
same chapter.
In fact, all six disfluencies feature in the core group of elicited rater
studies, most of which follow a similar design. Listeners rate speech
recordings of L2 learners for fluency. The recordings are subsequently
analysed for acoustic measures such as speed and pauses and the
results compared with the ratings. Inferences are made about the
features that listeners responded to in making their judgements on
fluency. These core studies are given a separate section, Effects of [the
disfluency] on listeners, in central chapters. Data from classroom
interaction is included in Chap. 2: Silent pauses for a number of
reasons. Walsh and Li (2013) illustrate the problem of assigning
between-turn pauses in interaction to one or other speaker and discuss
the implications when one of the speakers is a teacher eliciting
responses to questions and the other a student who is socially reticent
or requires time to think or both. The shorter pauses observable in
learner pair work contribute to fluency in dialogic speech (Tavakoli,
2016). Longitudinal study of learners shows a steady decline in pause
time with proficiency gains, though quantity and pause distribution
mid-clause still convey disfluency (Mora & Valls-Ferrer, 2012).
Apart from appearing in other sections such as effects on listeners
and proficiency, filled pause data in Chap. 3 is restricted to the uh and
um studies in the elicited data section. Prolongations are the only
disfluency whose study data is largely based on corpora and natural
language processing: they do not enjoy autonomous sections in any
other chapter. Although Chap. 4: Prolongations includes short sections
on classroom and elicited data, prolongations feature only in
transcripts of classroom interaction and are rarely the subject of
analysis; they also figure in elicited data collected in a classroom and
gathered into a corpus in Moniz et al. (2007). There is no classroom
data section in Chap. 3: Filled pauses or Chap. 5: Repetitions. Corpora
and natural language processing sections are included only in Chap. 4.
Natural language data collected outside the classroom is confined to
Chaps. 5, 6 and 7 on repetitions, self-corrections, and false starts.

1.5 Disfluency Types and Proficiency Level

SPs were the only fluency measure to discriminate between high- and
low- proficiency learners in Riggenbach (1991). Low-proficiency
learners have been found to produce the great majority (69%) of non-
juncture SPs, most SPs occurring in compound disfluencies, e.g. false
starts, self-corrections, and repetitions (Cenoz, 1998). Lower-
proficiency learners produce more SPs than higher proficiency leaners;
and pause duration decreases within speech units in line with higher
L2 proficiency (Riazantseva, 2001). Thus, in rater studies, absence of
pauses is associated with higher proficiency (Bosker et al., 2013), and
fluency ratings are found to fall when SP duration and frequency are
artificially increased (Bosker et al., 2014). SP length increases and the
frequency of mid-clause SPs declines with higher levels of proficiency
(Tavakoli et al., 2017). Lower-intermediate speakers are reported to
produce more SPs per minute and nearly twice the number of SPs
within self-corrections compared to advanced speakers (Williams &
Korko, 2019).
Rater studies disagree on whether FPs signal proficiency. FPs alone
were the poorest indicator of raters’ fluency judgments in Bosker et al.
(2013) and Pinget et al. (2014) but contrariwise best explained the
raters’ predictions of communicative adequacy, or proficiency, in
Révész et al. (2016). Riggenbach (1991) reports that advanced
speakers produced more lexicalised FPs such as you see than lower
fluency speakers, who overwhelmingly stuck to non-lexical fillers. Her
finding is consistent with raters’ positive response to speakers’
production of L1 lexical FPs in Préfontaine & Kormos (2016).
Based on evidence that length of pause is disregarded by raters in
fluency studies such as Bosker et al. (2014) and therefore unrelated to
proficiency level, the same premise might reasonably be applied to
prolongations.
Repetitions have been found to decline by a factor of three with
increased fluency level (Révész et al., 2016). Low-fluency speakers are
reported to make more discontinuous repetitions of the kind
sometimes found in false starts than the concatenated, i.e. immediately
adjacent, ones found outside a repair. Concatenated repetitions demand
little processing from listeners (Olynyk et al., 1987).
Self-corrections have also been found to decrease in number with
increasing proficiency (Simard et al., 2017), and are fewer than false
starts at all fluency levels (van Hest, 1996; Williams & Korko, 2019).
Self-correction may be a necessary disfluency for optimum learner
progress. The learner focus on target language structures implied by
self-corrections fosters L2 development and may be an efficient means
to acquire proficiency at the low-intermediate (CEFR B1) stage. The
majority of self-corrections in speakers judged to be low proficiency
have been reported as lexical in focus (Matea Kovač & Milatović, 2013).
In contrast, while lower-intermediate (B1) learners are the most
prolific self-correctors, post-beginner (A2) learners have been found to
produce no self-corrections and upper-intermediate (B2) level fewer
(Tavakoli et al., 2017).
False starts have been reported to show a negative correlation with
L2 proficiency and have been described as ‘productive devices’
(Verhoeven, 1989, p. 151). With greater automaticity, higher level
learners may monitor less overtly, resulting in the production of fewer
false starts. Similarly, raters have been found to associate fewer false
starts with higher proficiency learners (Révész et al., 2016), a finding
confirmed by Williams and Korko (2019). The fact that advanced
learners produce four times as many false starts as self-corrections
(van Hest, 1996) reflects the increasing accuracy of their language.
Improved automaticity at this level may also be responsible for better-
formed repairs (Kormos, 2006) with an orientation to shared discourse
(Gilabert, 2007), perhaps because advanced learners are better able to
execute timely conceptual revisions.
1.6 Disfluencies in Public Language Test
Descriptors
In their test descriptors, English language rating scales reflect socio-
cultural (dis)fluency norms associated with levels of speaker
proficiency. Each chapter maps a disfluency measure against five of the
publicly available descriptors (for Aptis, IELTS, OET, PTE Academic, and
TOEFL iBT) and the influential Common European Framework of
Reference for Languages (CEFR) (Council of Europe, 2001), to which all
five are referenced.
Pauses are cited in at least one band level descriptor for all five tests
but none except OET, which refers to ‘fillers’, specifies whether SP or FP.
Tavakoli et al. (2017) recommends that Aptis rating scales and training
materials should specify pauses as silent, filled, or mid-clause as
necessary.
Prolongations are not referred to in any of the descriptors. Like
pauses, repetitions are included in one or more descriptors from all five
scales. (Self-)corrections are referred to in the CEFR descriptors, the
IELTS and OET tests, and as the more generic ‘reformulations’ in the
Aptis test. False starts are named in the CEFR descriptors, and those for
the Aptis, PTE Academic and TOEFL iBT rating scales. Anomalously,
Aptis therefore refers to both reformulations and false starts.
The disfluencies that are mentioned most often are pauses and
repetitions. Presumably, these forms are most salient to listeners and
examiner raters.

1.7 Comment
The preceding outline has been largely descriptive, reflecting the early
state of the disfluency field. It seems that writers share a common
purpose: to better understand the causes, workings, and consequences
of disfluencies in spoken language and, in the case of L2 investigators,
the special considerations that might apply to speakers of a second
language. Yet, secondary interests, including language acquisition, task-
based performance, human-machine interfacing, test standardisation,
and listener responses, to name just a few, mean that differences in
working definitions, methodologies, and conclusions, have resulted in a
centrifugal push of knowledge concentration into a number of different
specialisms rather than a centripetal pull towards a common
understanding.
The constructs of fluency and disfluency can best be understood as
interactive, socially-produced, and jointly-constructed phenomena (cf.
Jaspers, 2016; Segalowitz, 2016) that have significance for the iterative
reproduction of social relations. In this respect, Segalowitz’s (2016)
framework offers more global insights into fluency and therefore
disfluency, and is helpful in foregrounding the crucial role of the social
context and its intimate relation with speaker motivation. The essential
components of the framework are (1) the cognitive processing systems
familiar from the upper components (conceptualiser and formulator) of
speech models like Levelt’s (1989) and relevant to psycholinguistic
investigations, e.g. in inhibitory control; (2) the lower, articulator
component of speech models that makes audible (dis)fluency features
such as speed, pausing, and repetition; and (3) the individual
motivation to produce speech of a more or less (dis)fluent nature. The
other two elements in Segalowitz’s framework represent the
communicative context in which the speaker and the other components
operate, namely the social context, and (past) perceptual and cognitive
experiences.
In adapting the framework to L2 disfluencies (Fig. 1.1), the last two
elements above are combined on the basis that the social context
creates the perceptual and cognitive experiences, whose lasting trace is
operationalised in the motivation element.
Fig. 1.1 Framework for L2 (dis)fluency and proficiency (based on Segalowitz, 2016)
There seems no advantage to adding an additional affective long-
term memory factor. The interactive communicative context explains
how and why disfluencies develop in L2 learners, and their trajectory
from bald mid-clause silent pauses to covert repairs, which mirror
developments in automaticity and consequent improvements in the
availability of working memory. The communicative context also
explains the type of disfluency the speaker learns to produce, its
position in the utterance, and the social interactions in which it is
acceptable, e.g. the present vogue for like as a filler in English. The
adapted model is sufficient to account for (dis)fluency and its gradual
orientation towards a social standard regardless of first or second
language. It contextualises disfluencies and provides an explanation for
their relationship with proficiency inasmuch as proficiency necessarily
includes (dis)fluency.
Just as Fasold and Connor-Linton (2014) note language variation in
spoken communication according to situation and speaker role and
purpose, so are different (dis)fluencies acquired according to the same
situations, roles and purposes. Thus, in relation to L1 development,

When children learn to use the socially approved variety of

spoken language in school, it is not from what their teachers
explicitly teach in class, but rather from adjusting their speech to
 match the speech of other children, in halls, on the playground,
and outside of school, and thus gain their approval.
(Fasold & Connor-Linton, 2014, p. 10).
A similar social process might apply to L2 learners and their
learning environment. For example, Hellermann’s (2009) data, all
collected in an ESL classroom but, thanks to permanently sited video
recorders and microphones, containing both task-oriented exchanges
and conversation, is a reminder that the L2 learner’s further project is
to develop language variation in the L2. At first, however, the L2
learner’s overarching concern is to join a community of practice
(Wenger, 1998), specifically a language learning classroom of English
language users. Hellermann’s focus in the data is on the participant,
Inez’s, self-initiated self-repairs (SISRs). In this context and at this stage
of her L2 language development, term 5, little difference can be seen
between Inez’s (dis)fluency in task and conversation. Rather,
Hellermann noticed an increase in the number of SISRs over Inez’s five
terms of study, implying more active monitoring of her language
production as a learner. In the first extract, Inez finds the space to make
a self-repair during a task.

Inez: teacher? uh, we have another question.

eh ea- eat is food. Drink is (.) water. (.) a:nd
have is have the (.) asp- have aspirin? I
Te: take aspirin.
Inez: take aspirin.
(Hellermann, 2009, p. 123)

And in conversation with A, a visitor to the classroom, she is equally

successful.

A: I didn’t know that.

Inez: yes Mayan people. She uh they uh speak Maya?
A: uh huh,
I: and Spanish.
(Hellermann, 2009, p. 126)

Although both self-repairs are noticeably disfluent, the first with

silent pauses, filled pauses, prolongations, and repetitions; the second
with filled pauses, Inez accomplishes her communicative objective.
Which intervention strategies speakers adopt to achieve a temporary
break in fluent speech will thus vary as to their self-ascribed identity,
the values they align with, and the interactant. In fact, the disfluencies
speakers produce in the service of repair bear only an indirect
relationship to metadiscoursal social judgements; it is rather the
speech style adopted by the speaker as a consequence of their social
judgement that shapes the speech process and defines its points of
vulnerability. That style, whether impetuous or measured, familiar or
formal, confident or hesitant—having regard to topic, social context,
size of audience, and other environmental factors—determines the
speaker’s pre-planning, their lexical range, syntactic complexity and so
on, and predicts where breakdowns and disfluencies are likely to occur.

1.8 Overview
Apart from introduction and conclusion, the inner six chapters (Chaps.
2–7) are each devoted to a recognised disfluency. Most of the chapters
are organised in a similar way, with some variation reflecting the sort of
methodology and data characteristic of work in that disfluency area.
Chapters often begin by reporting L1 groundwork in the disfluency as
first language research in these areas invariably predates L2 studies.
The next chapter (Chap. 2) defines and gives examples of silent
pauses and reports some of the influential studies of Goldman-Eisler,
who studied silent pause duration and frequency and later their
distribution in utterances. The chapter then interprets a group of L2
rating studies, common to most of the disfluency chapters, to suggest
the effect of silent pauses on listeners. A section follows that reports
studies in classroom interaction that reference silent pauses. The
chapter then revisits some of the rater studies and cites others to
examine the relation of silent pause production to L2 oral proficiency
level. The chapter ends with a section on references to silent pauses in
the descriptors of public language tests and a closing comment.
Chapter 3 on filled pauses follows a similar structure, though it
notes the prominence given by (dis)fluency researchers to filled pauses
compared to silent pauses, with many more publications on this
disfluency form. After a short introduction, a formal description of filled
pauses describes their distribution, frequency, duration and pitch. The
next section is in two parts, the first reporting studies on the effects on
listeners of the so-called uh and um forms, i.e. binary filled pause forms
claimed to have distinct functions; and the second interpreting the rater
studies introduced in the previous chapter with reference to the effects
of filled pauses. Then, three kinds of speech environment: cognitive,
interactive, and turn-taking, for FPs are considered. The section on FPs
and proficiency includes a discussion of the effects of their L1 FPs on
the FPs the same speakers produce in their L2: L2 learners gradually
adopt FPs appropriate to the target language and allow the L1 FPs to
attrite. Some difference is observed between studies that report
advanced speakers producing more FPs than lower-level learners, and
those that report the opposite. The explanation for the difference
probably lies in the pause distribution, mid- or end-juncture. The
published descriptors for the Occupational English Test (OET) are the
only ones of the English language tests to mention FPs, here in the form
of fillers.
Chapter 4 concerns prolongations, in pathological speech associated
with stuttering but in this case a form of disfluency evidenced as
unusual lengthening of phonemes and very common as a form of
hesitation to gain processing time. Precisely for this reason,
prolongations are of interest to automatic speech recognition
researchers, whose concern is to establish the maximum tolerable
length of prolongation for the listener. Prolongations have long been
recognised and reproduced in transcriptions by conversation analysts,
conventionally by a colon following the extended syllable. The
second section offers a formal description, noting the tendency of
prolongations to fall on function words such as and, the and so. The
largest section reports the effects of prolongations on listeners,
including extended forms of function words such as a and the occurring
before uh and um, the FPs discussed in the previous chapter. Apart from
studies in human-machine interaction, the majority of prolongation
data is found in corpora, e.g. the corpus in Fox Tree and Clark (1997),
which examines the incidence of thi:y as an extension of the strong form
thee (ordinarily thuh). Three more sources of data exemplified in this
section are natural language processing, interaction, elicited data.
Prolongations and proficiency level are discussed only in the context of
language test descriptors and then by inference from mention of
hesitations. The functional association of prolongations with filled
pauses is noted, along with their appearance in contrasting
environments.
Chapter 5 distinguishes immediate repeats from language repeated
after some delay and suggests that immediate or ‘concatenated’
repetition (Maclay & Osgood, 1959, p. 148) is heard as disfluent. The
second section offers a formal description of repetitions and includes
the repetition often found in repairs and a description of Plauché and
Shriberg’s (1999) taxonomy of repetitions by intonation. The following
section examines the effect of repetitions on listeners and reports
several of the rater studies with reference to findings on repetitions.
After noting L2 study outside the classroom, and in a set of empirical
studies, and the effect of task on L2 learners’ repetitions, a section on
repetition and proficiency level revisits some of the rater studies for
conclusions drawn on the association of repetitions with fluency. The
penultimate section surveys references to repetitions in the public
descriptors of English language speaking tests. The chapter ends by
noting the joint occurrence of repetitions with pauses, after which the
speaker restarts the utterance with a function word preceding the
lexical word causing trouble, and the fact that listeners often fail to
notice speakers’ repetitions of this kind, which are most likely sites of
speakers’ covert repair.
Chapter 6 concerns self-corrections, one of the two compound
reformulation disfluencies. Self-corrections are distinct from false
starts, the other kind of reformulation, for a number of reasons. In a
self-correction, the speaker attempts to repair the utterance towards a
form compliant with a social norm. For the same reason, because the
repair is usually towards a standard in pronunciation, grammar, or
lexis, the trouble source is marked as non-standard and the whole
repair heard as a disfluency. A formal description of self-corrections in
the second section refers to the seminal paper of Schegloff, Jefferson
and Sacks (1977) and their work on L1 repairs. Levelt (1983), also
working with L1 (Dutch) data, defines five categories of repair, one of
which is self-correction, and three of the others forms of false start. It is
Levelt that L2 researchers such as Kormos build on to describe features
of L2 learner self-corrections and other repairs. The next section
reports studies on the effects of self-corrections on listeners. The
following three sections discuss studies of self-corrections whose data
has been collected outside the classroom, studies of self-corrections in
classroom interaction, and the effect of task on self-corrections in
elicited data. Van Hest (1996) and Williams and Korko (2019) report
that self-corrections decline with proficiency gains; and Tavakoli et al.
(2017) note that lower-intermediate learners make abundant self-
corrections as they actively experiment with achieving target language
forms. Only two English language speaking tests, IELTS and the
Occupations English Test (OET), refer to self-correction in their public
descriptors, but the references may be to the more generic ‘repair’, i.e.
including false starts.
Chapter 7, the last chapter on disfluency types, describes false
starts, as distinct from L2 self-corrections, the other kind of repair. Of
more interest than self-corrections as a linguistic opportunity for intake
and acquisition, false starts represent the speaker’s revision of a
standard form that would not necessarily be recognised as a trouble
source by the listener. Because they represent the speaker’s choice of
linguistic revision, false starts are of great interest to language
acquisition theorists as places in the utterance where the monitoring
process is thought to involve the speaker’s selection of a repair from
one of a number of alternative revisions. This is the case because,
unlike a self-correction, no particular standard form exists for the
repair since the trouble source already adhered to target language
norms. The second section offers a formal description of both types of
repair, self-corrections and false starts, as comprising three parts: a
trouble source, editing phase, and repair (Levelt, 1983). The effects of
false starts on listeners is reported in the following section, which notes
that false starts are not simply speaker problem-solving devices. They
also benefit listeners by building ‘discourse coherence’ (Levelt, 1983, p.
42). At the same time, as ‘regressive speech markers’ (Olynyk et al.,
1987), false starts impose a processing cost on listeners, forcing them
to re-evaluate speech already heard. Compared to a middle position, or
at the beginning prefixed by ‘and’, this cost is mitigated if the false start
occurs at the start of the utterance (Fox Tree, 2010). The next section
on false start data types includes data outside the classroom, within
classroom interaction, and task effect in elicited data. A section on false
starts and speaker proficiency level notes that while lower-proficiency
learners produce greater numbers of false starts than advanced
learners, those of advanced learners tend to be more discourse-related.
Some evidence exists that false start production reflects individual
speaker style rather than proficiency level (Zuniga & Simard, 2019).
The penultimate section on false starts and public language tests notes
that only two tests: the TOEFL iBT Test and the Pearson Test of English
Academic reference false starts in their descriptors, the former within
Level A2, and the latter for the level extremes of Native-like fluency and
Disfluent.
The book will be of interest to postgraduate students of linguistics,
communication studies, media, and other social sciences, and to
specialists from other disciplines who want an overview of the range of
research in disfluency with a particular emphasis on second language
speakers and proficiency.

References
Adell, J., Bonafonte, A., & Escudero-Mancebo, D. (2010). Modelling filled pauses
prosody to synthesise disfluent speech. In Speech Prosody 2010-Fifth International
Conference.

Ahmadian, M. J., Abdolrezapour, P., & Ketabi, S. (2012). Task difficulty and self-repair
behavior in second language oral production. International Journal of Applied
Linguistics, 22(3), 310–330.

Arnold, J. E., Fagnano, M., & Tanenhaus, M. K. (2003). Disfluencies signal theee, um,
new information. Journal of Psycholinguistic Research, 32(1), 25–36.

Betz, S., Wagner, P., & Schlangen, D. (2015). Modular synthesis of disfluencies for
conversational speech systems. Studientexte zur Sprachkommunikation: Elektronische
Sprachsignalverarbeitung, 2015, 128–134.

Betz, S., Zarrieß, S., & Wagner, P. (2017). Synthesized lengthening of function words -
The fuzzy boundary between fluency and disfluency. In Proceedings of the
International Conference Fluency and Disfluency.

Bortfeld, H., Leon, S. D., Bloom, J. E., Schober, M. F., & Brennan, S. E. (2001). Disfluency
rates in spontaneous speech: Effects of age, relationship, topic, role, and gender.
Language and Speech, 44, 123–147.
Bosker, H. R., Pinget, A. F., Quené, H., Sanders, T., & De Jong, N. H. (2013). What makes
speech sound fluent? The contributions of pauses, speed and repairs. Language
Testing, 30(2), 159–175.

Bosker, H. R., Quené, H., Sanders, T., & De Jong, N. H. (2014). The perception of fluency
in native and nonnative speech. Language Learning, 64(3), 579–614.

Brennan, S. E., & Schober, M. F. (2001). How listeners compensate for disfluencies in
spontaneous speech. Journal of Memory and Language, 44(2), 274–296.

Cenoz, J. (1998). Pauses and communication strategies in second language speech.

Chiang, S. Y., & Mi, H. F. (2011). Reformulation: A verbal display of interlanguage

awareness in instructional interactions. Language Awareness, 20(2), 135–149.

Clark, H. H. (1994). Managing problems in speaking. Speech Communication, 15(3–4),

243–250.

Council of Europe. (2001). Common European framework of reference for languages:

Learning, teaching, assessment. Cambridge University Press.

Crible, L., & Pascual, E. (2020). Combinations of discourse markers with repairs and
repetitions in English, French and Spanish. Journal of Pragmatics, 156, 54–67.

Cucchiarini, C., Strik, H., & Boves, L. (2002). Quantitative assessment of second
language learners’ fluency: Comparisons between read and spontaneous speech. The
Journal of the Acoustical Society of America, 111(6), 2862–2873.

Derwing, T. M., Rossiter, M. J., Munro, M. J., & Thomson, R. I. (2004). Second language
fluency: Judgments on different tasks. Language Learning, 54(4), 655–679.

Eklund, R. (2001). Prolongations: A dark horse in the disfluency stable. In ISCA

Tutorial and Research Workshop (ITRW) on Disfluency in Spontaneous Speech.

Engelhardt, P. E., McMullon, M. E., & Corley, M. (2019). Individual differences in the
production of disfluency: A latent variable analysis of memory ability and verbal
intelligence. Quarterly Journal of Experimental Psychology, 72(5), 1084–1101.

Faerch, C., & Kasper, G. (1983). Procedural knowledge as a component of foreign

language learners’ communicative competence. In Kommunication im (Sprach-)
Unterricht. Rijksuniversiteik.

Faerch, C., & Kasper, G. (1984). Two ways of defining communication strategies.
Language Learning, 34(1), 45–63.
Fasold, R. W., & Connor-Linton, J. (Eds.). (2014). An introduction to language and
linguistics. Cambridge University Press.

Firth, A., & Wagner, J. (1997). On discourse, communication, and (some) fundamental
concepts in SLA research. The Modern Language Journal, 81(3), 285–300.

Firth, A., & Wagner, J. (2007). On discourse, communication, and (some) fundamental
concepts in SLA research. The Modern Language Journal, 91, 757–772.

Foster, P., & Skehan, P. (1996). The influence of planning and task type on second
language performance. Studies in Second Language Acquisition, 18(3), 299–323.

Fox Tree, J. E. F. (1995). The effects of false starts and repetitions on the processing of
subsequent words in spontaneous speech. Journal of Memory and Language, 34(6),
709–738.

Fox Tree, J. E. (2010). Discourse markers across speakers and settings. Language and
Linguistics Compass, 4(5), 269–281.

Fox Tree, J. E., & Clark, H. H. (1997). Pronouncing “the” as “thee” to signal problems in
speaking. Cognition, 62, 151–167.

Fraundorf, S. H., & Watson, D. G. (2011). The disfluent discourse: Effects of FPs on
recall. Journal of Memory and Language, 65(2), 161–175.

Freed, B. F. (1995). What makes us think that students who study abroad become
fluent? In B. F. Freed (Ed.), Second language acquisition in a study-abroad context (pp.
123–148). John Benjamins.

Gass, S. (2004). Conversation analysis and input-interaction. The Modern Language

Journal, 88(4), 597–602.

Gilabert, R. (2007). Effects of manipulating task complexity on self-repairs during L2

oral production. International Review of Applied Linguistics in Language Teaching,
45(3), 215–240.

Goldman-Eisler, F. (1968). Psycholinguistics: Experiments in spontaneous speech.

Goldman-Eisler, F. (1972). Pauses, clauses, sentences. Language and Speech, 15(2),

103–113.

Hellermann, J. (2009). Looking for evidence of language learning in practices for

repair: A case study of self-initiated self-repair by an adult learner of English.
Scandinavian Journal of Educational Research, 53(2), 113–132.
Hindle, D. (1983, June). Deterministic parsing of syntactic non-fluencies. In 21st
Annual Meeting of the Association for Computational Linguistics (pp. 123–128).

Jaspers, J. (2016). (Dis)fluency. Annual Review of Anthropology, 45, 147–162.

Kormos, J. (1998). A new psycholinguistic taxonomy of self-repairs in L2: A

qualitative analysis with retrospection. Even Yearbook, ELITE SEAS Working Papers in
Linguistics, 3, 43–68.

Kormos, J. (2000). The timing of self-repairs in second language speech production.

Studies in Second Language Acquisition, 22(2), 145–167.

Kormos, J. (2006). The structure of self-repairs in the speech of Hungarian learners of

English. Acta Linguistica Hungarica, 53(1), 53–76.

Lennon, P. (1990). Investigating fluency in EFL: A quantitative approach. Language

Learning, 40(3), 387–417.

Levelt, W. J. M. (1983). Monitoring and self-repair in speech. Cognition, 14, 41–103.

Levelt, W. J. M. (1989). Speaking: From intention to articulation. MIT Press.

MacGregor, L. J., Corley, M., & Donaldson, D. I. (2009). Not all disfluencies are equal:
The effects of disfluent repetitions on language comprehension. Brain and Language,
111(1), 36–45.

Maclay, H., & Osgood, C. E. (1959). Hesitation phenomena in spontaneous English

speech. Word, 15(1), 19–44.

Martin, J. G. (1970). On judging pauses in spontaneous speech. Journal of Verbal

Learning and Verbal Behavior, 9, 75–78.

Matea Kovač, M., & Milatović, B. (2013). Analysis of repair distribution, error
correction rates and repair successfulness in L2. Studia Linguistica, 67(2), 225–255.

McAllister, J., Cato-Symonds, S., & Johnson, B. (2001). Listeners’ ERP responses to
false starts and repetitions in spontaneous speech. In ISCA Tutorial and Research
Workshop (ITRW) on Disfluency in Spontaneous Speech. Edinburgh, 29–31 August.

McRobie, K. F. (1993). Perceived fluency: A study of self-correction, speech rate and

three foreign accents as components of fluency in English as a second language
(Doctoral dissertation, University of San Francisco).
Moniz, H., Mata, A. I., & Viana, M. C. (2007). On filled-pauses and prolongations in
European Portuguese. In Eighth Annual Conference of the International Speech
Communication Association.

Mora, J. C., & Valls-Ferrer, M. (2012). Oral fluency, accuracy, and complexity in formal
instruction and study abroad learning contexts. TESOL Quarterly, 46(4), 610–641.

Olynyk, M., d'Anglejan, A., & Sankoff, D. (1987). A quantitative and qualitative analysis
of speech markers in the native and second language speech of bilinguals. Applied
PsychoLinguistics, 8(2), 121–136.

Pinget, A. F., Bosker, H. R., Quené, H., & De Jong, N. H. (2014). Native speakers’
perceptions of fluency and accent in L2 speech. Language Testing, 31(3), 349–365.

Plauché, M., & Shriberg, E. (1999, August). Data-driven subclassification of disfluent

repetitions based on prosodic features. In Proceedings of International Congress of
Phonetic Sciences (Vol. 2, pp. 1513–1516).

Préfontaine, Y., & Kormos, J. (2016). A qualitative analysis of perceptions of fluency in

second language French. International Review of Applied Linguistics in Language
Teaching, 54(2), 151–169.

Préfontaine, Y., Kormos, J., & Johnson, D. E. (2016). How do utterance measures
predict raters’ perceptions of fluency in French as a second language? Language
Testing, 33(1), 53–73.

Révész, A., Ekiert, M., & Torgersen, E. N. (2016). The effects of complexity, accuracy,
and fluency on communicative adequacy in oral task performance. Applied
Linguistics, 37(6), 828–848.

Riazantseva, A. (2001). Second language proficiency and pausing a study of Russian

speakers of English. Studies in Second Language Acquisition, 23(4), 497–526.

Riggenbach, H. (1991). Toward an understanding of fluency: A microanalysis of

nonnative speaker conversations. Discourse Processes, 14(4), 423–441.

Rossiter, M. J. (2009). Perceptions of L2 fluency by native and non-native speakers of

English. Canadian Modern Language Review, 65(3), 395–412.

Schegloff, E. A. (1987). Recycled turn beginnings: A precise repair mechanism in

conversation’s turn-taking organization. Talk and Social Organisation, 1(1), 70–85.

Schegloff, E., Jefferson, G., & Sacks, H. (1977). The preference for self-correction in the
organization of repair in conversation. Language, 53, 361–382.
Another random document with
no related content on Scribd:
 CHAPTER I
INTRODUCTION

Cincinnati, the city of my birth and early memories, was, in the

‘sixties, about as begrimed and noisy and altogether unattractive as
any place well could be; yet it possessed certain attributes which
really entitled it to the proud designation of “The Queen City of the
West.”
It was prosperous; it had hardly yet been surpassed in prosperity
by Chicago; Cleveland was not even spoken of as a rival; and in many
ways it was the most important centre west of New York and east of
the Mississippi.
It owed its early development principally to its advantageous
location. It lay on the great central route from the East to the West,
which runs from Baltimore and Washington to Cumberland and over
the Alleghenies to Pittsburg, thence by the Ohio River to Cincinnati
and on west to St. Louis and south to New Orleans. It had an
important trade with New Orleans and drew commerce from a large
territory to the north. But whatever else may be said of it, its most
devoted citizen could not claim that Cincinnati was beautiful. Its
buildings were unlovely; its streets were badly paved and as badly
kept; and it lay under a pall of soft coal smoke which left its sooty
mark upon everything—inhabitants included.
Yet, ugly as it was, the city boasted an unusual society. During the
first half of the nineteenth century many young men of good stock
and great ability, drawn by the promise of rapid advancement, had
moved to Cincinnati from all parts of the East and South; New
Jersey, New England, Virginia and Kentucky contributing, perhaps,
the greatest number. There were many families of wealth and culture
which, without parade or display, maintained fine homes and
dispensed a generous hospitality. The suburbs, East Walnut Hills,
Mt. Auburn and Clifton, on the heights to the north and east, were
famous for their beautiful country places.
Then there was a large population of the best class of Germans,
many of whom were university men who left their own country after
the Revolution of 1848 and came to Cincinnati to settle. Of these,
Frederick Hassaurek, General Willich and Judge Stallo, who came to
Cincinnati when Carl Schurz went to St. Louis, are perhaps the most
prominent. The German influence upon the community was marked.
It made for a more liberal Sunday; it brought the study of German
into the public schools; and it developed a strong taste for good
music. Indeed, the musical advantages of Cincinnati in my girlhood
were better than those of any city in the United States, with the
exception of New York or Boston. Theodore Thomas was president of
the Conservatory of Music and he organised a symphony orchestra
which he continued to direct until he went to Chicago along about
1890.
Cincinnati in those days, with her educated, wealthy and public-
spirited society, was much in advance of any other city in the
Mississippi Valley in culture and refinement. There was great
interest in schools of all sorts and in every kind of intellectual
activity. Away back in 1848 the Literary Club of Cincinnati was
formed by a company of men among whom were both Mr. Taft’s
father and mine, as well as Rutherford B. Hayes, Stanley Matthews,
Manning F. Force and Mr. Spofford, later Librarian of Congress. This
club continues to be a cherished institution and in my girlhood it was
the centre of all interest in literature and intellectual pursuits.
My father, John Williamson Herron, was a graduate of Miami
University at Oxford, Ohio, and was in college with Benjamin
Harrison. He was for fifty years a trustee of that institution and was
devoted to its interests. My husband’s father, Judge Alphonso Taft,
was one of the Yale class of 1833, was for many years a member of
the Yale Corporation, and had five sons who graduated at that
university. My mother’s brother, Judge Isaac Clinton Collins, and
one of my two brothers also graduated at Yale, while my other
brother graduated at Harvard, so it will be seen that both my
husband and I grew up in the midst of strong collegiate traditions.
To write about one’s childhood is not easy. Memories by the score
come flocking up, but, dear as they are, upon examination they turn
out to be quite commonplace and hardly worth relating. My
memories are not sufficiently “early” to have any special value. The
first thing that I dimly remember is sitting on the front steps of my
home watching some sort of parade in which there were many
soldiers, but I was too young then to know that it was a peace
celebration I was witnessing at the close of the Civil War.
My father was a lawyer who came to the bar of Ohio in the ’forties.
He was United States Attorney under President Harrison, was a State
Senator, and twice declined appointments to the Bench because the
salary attached to these positions was not enough to support his
large family. I was the fourth in a family of eleven, eight girls and
three boys. One boy and two girls died before I can remember.
Our house was one of a block of grey brick houses in Pike Street, at
the east end of Cincinnati, which, at that time, was the fashionable
residence section of the city. Pike Street runs down to the river on a
rather steep incline and, as it was paved with cobblestones, my early
memories are somewhat marred by an impression of the frequent
clatter and clang of heavy wagons pulling their way up the hill from
the river landing.
While our house was not particularly distinguished, being much
like those on either side of it, across the street from us there were two
very striking and imposing residences which lent distinction to the
neighbourhood, and in which, as I grew up, were formed the
pleasantest associations of my life. The one directly opposite was a
large, square, red brick house which had an air of great dignity. It
was the home of Mr. Larz Anderson. There were ten boys in the
Anderson family and, though they were all much older than I and
most of them had gone away before I grew up, I remember that it
was a very lively household always. In my later girlhood we were
specially linked to this family by the marriage of one of the boys,
Charles, to my sister Jennie.
The house next to Mr. Anderson’s, on the north, I knew as the
Sinton home. A low, colonial structure, well set in a garden of green
lawns and finely kept shrubbery, it is still one of the most beautiful
residences in Cincinnati, and, indeed, in the whole country. Its
architecture suggests that of the White House and it was, as a matter
of fact, designed by the same architect, an Irishman named Hoban.
The Sinton house is lower than the White House, being only one
story high with a basement, but it has the same classic outlines and it
bears, moreover, the stamp of time, which gives it a character all its
own.
 It was built about 1800 by a Mr. Martin Baum, but was purchased
by the first Nicholas Longworth in the early part of the century and
was the home of the Longworth family for a generation. Long before
I can remember, it was bought by Mr. David Sinton, one of the most
successful business men in Ohio, and to me it was always the Sinton
home. When I was about twelve years old, Mr. Sinton’s daughter
Annie married my husband’s brother, Charles P. Taft, and as they
have always lived in this old house it has come to be known, since
Mr. Sinton’s death in 1901, as the Taft house. It is the only Taft house
in Cincinnati now, the house where my husband was born having
been sold after his father’s death, and it has been the scene of many
of the most important events of my life. It was there that my husband
received the announcement of his nomination for the Presidency; it
was there, in front of the house, that he made his speech of
acceptance; and it was there that Charles Taft gathered a large party
of friends on the night of November 6, 1908, to receive with us the
election returns. And it is now to this house, where my husband’s
brother Charles and his wife dispense a generous hospitality, that we
always go when we return to Cincinnati.
 MRS. TAFT’S CHILDHOOD HOME ON
PIKE STREET, IN CINCINNATI

My girlhood days were spent quite placidly in Miss Nourse’s

school, which was known in Cincinnati as “The Nursery,” and where
all the girls of the Herron family, as well as Mr. Taft’s only sister,
Fanny, received their education. Miss Nourse was a Maine woman
with a thorough New England education and with a thoroughly New
England idea of imparting it. She insisted, especially, upon languages
and literature. Much of my time, outside of that taken up in regular
school work, I devoted to the study of music, and I practised my
scales on the family piano with such persistence that I wonder the
whole neighbourhood did not rebel. Music was the absorbing interest
of my life in those days, the inspiration of all my dreams and
ambitions.
 Our house was none too large for the family, but as there was a
wide difference in our ages it happened that my oldest sister was
married while my youngest sister was still a baby in long clothes.
Then, the boys went away to college and were gone the better part of
each year, so it was not often that we were all at home together.
Nevertheless, we had our share of the happy-go-lucky and somewhat
crowded existence of a large family on a moderate income.
My mother was Harriet Collins, and when she was seventeen years
old she came with her mother to Cincinnati, from Lowville, New
York, to live with her brother, Judge Collins, who was my father’s law
partner and continued to be so for more than forty years. Her father,
Eli Collins, was a Member of Congress from the Lowville district of
New York. My mother was in many ways a remarkable, as well as a
most attractive, woman. She had an exceedingly keen wit and a mind
alert to the humour in every situation. With so many children to
nurse, to scold, to sew for and, sometimes, to cook for—in a word, to
bring up on a small income—she would seem to have had little time
for outside interests; but she was very popular in society and I
remember that in her busiest years she went out a great deal. She
had a stimulating personality and I do know that she made her
family circle a very amusing and interesting one in which to grow up.
The only incident of my girlhood which was in any way unusual
was my first visit to the White House as a guest of President and Mrs.
Hayes. Mr. and Mrs. Hayes and my father and mother had been
lifelong friends. Mr. Hayes was, at one time, a partner in my father’s
law firm. They had been closely associated for a great many years
and had a very warm regard for each other. My youngest sister was
born shortly after the election of Mr. Hayes, was named Lucy Hayes
Herron, after Mrs. Hayes, and was taken to the White House to be
christened. My mother paid several visits to the White House and
after my sister Jennie was married Mrs. Hayes invited her and Mr.
Anderson to stay a week with her and, to my intense excitement, she
added that she would like to have me accompany them. I was
seventeen years old; I had never been to Washington and to me it
was a very important event. I was not “out,” so I couldn’t spend my
time in the White House as I would have liked, in going to brilliant
parties and meeting all manner of charming people, but, fortunately
for my peace of mind, the Hayes lived very quietly, so it was not so
trying to have to devote myself to seeing the sights of the Capital like
any other tourist.
I didn’t meet my husband until I was eighteen years old. We had
been born and brought up in the same town; our fathers were warm
friends and had practised law at the same bar for more than forty
years; during that time our mothers had exchanged visits, and my
sister Maria and Fanny Taft were schoolmates and close companions
at Miss Nourse’s, but the Tafts lived at Mt. Auburn, a hill suburb of
Cincinnati, and after Will finished Woodward High School he went
for four years to Yale, so it is not at all surprising that we did not
meet.
Judge Alphonso Taft was Secretary of War, and later Attorney
General, in Grant’s Cabinet while his son Will was at college, but
before the latter graduated, the family had returned to Cincinnati, so
he came straight home and entered at once upon a law course in the
Cincinnati Law School. It was at that time, when he was still a
student and working as a law reporter on the Cincinnati
Commercial, that I met him. It was at a coasting party one winter’s
night, I remember very well, when I went with a party of young
people, including the Charles Tafts, to coast down a fine steep hill in
Mt. Auburn. Will Taft was there, and after being introduced to me he
took me down the hill on his big bobsled. After that we met very
frequently.
A small circle of us went in for amateur theatricals with much
enthusiasm and great earnestness. We launched ourselves in our
histrionic careers in “She Stoops to Conquer” which we gave at the
house of one of the company. Then came “A Scrap of Paper” in Mrs.
Charles Taft’s drawing room, in which both Will and I took part. We
had become very ambitious by this time and sent all the way to New
York for a professional stage-manager to help us with the
production. But it turned out a most nervous occasion. We were all
overtrained, I suppose. One thing after another went wrong until at
the crisis of the play, where the hero is supposed to find in the barrel
of a gun the scrap of paper upon which the whole plot hinges, the
amateur hero looked pretty foolish when he discovered there wasn’t
any gun. Another one of the company, in a fit of absentmindedness,
no doubt due to overwrought nerves, had carried it off the stage, and
just when the situation was getting tragic for the hero the culprit
came creeping back with it and carefully put it where it belonged, for
all the world as if he thought he were making himself invisible to the
audience.
But our ardour was not dampened. I remember Mr. Taft especially
in a burlesque of “The Sleeping Beauty,” which, in its legitimate
form, had been produced for charity at Pike’s Opera House. The
Unity Club, a most respectable organization of the young men of the
Unitarian Church, decided to give their version of the same story,
and it was a huge success. Mr. Taft played the title rôle and his
brother Horace, who is six feet four in his stocking feet, shared with
the Beauty the honours of the evening as a most enchanting Puck.
Then we had parties in the country, too. Many of our friends had
country places that spread along the Madison Road and the Grandin
Road on East Walnut Hills, and two of my closest friends lived out
there in a great house, looking down over the majestic but tawny
Ohio River, above the point where the sweeping curve begins that
carries it by the amphitheatre in which the business part of the city is
built. It was a long distance to East Walnut Hills and in my girlhood
we had to go the greater part of the way in a clumsy old omnibus that
clumped along over the unpaved roads at the rate of about three
miles an hour. But such little inconveniences didn’t trouble us, and
many were the vaudeville and charade parties that we had, there
being enough “talent” among us to get up an amusing performance at
a moment’s notice.
 MR. AND MRS. JOHN WILLIAMSON HERRON, MRS. TAFT’S
FATHER AND MOTHER

But in spite of all this gaiety, Mr. Taft was making very satisfactory
progress in his career. As a law reporter he showed his growing
interest in the public welfare by meeting certain elements in
Cincinnati politics with vigorous denunciation. There was a man
named Tom Campbell, a clever criminal lawyer, who had something
more than a suspicion against him of bribery and corruption of both
witnesses and juries, and he had succeeded in organising a political
machine that was running the town according to his directions.
Campbell was counsel for the defence in what was known as the
Hoffman case and was strongly suspected of tampering with the jury,
and Mr. Taft in reporting the case, took special pains to bring out all
the fine points in the lawyer’s character and methods, telling the
truth as he saw it.
This brought him into association with Mr. Miller Outcalt, the
Assistant Prosecuting Attorney, who represented the State in the
Hoffman case, and when Mr. Outcalt succeeded by election to the
position of prosecuting attorney he offered the place of assistant to
Mr. Taft, although he had been at the bar not more than seven
months. Mr. Taft served in this office for fourteen months and the
experience he had in the rough-and-ready practice in criminal trials,
in preparing cases for trial, in examining witnesses, in making
arguments to the court and in summing up to the jury, was the most
valuable experience he could possibly have in fitting him for trial
work at the bar.
But this experience was shortened by a circumstance not of his
seeking. Major Benjamin Butterworth was the Congressman from
one of the Cincinnati districts in President Arthur’s administration,
and the President being anxious to relieve the Collector of Internal
Revenue, called on Major Butterworth to suggest the name of
another man. Major Butter worth had been for a long time a warm
friend of Mr. Taft, thought he had a good family name and was too
young in politics to have many political enemies, so he suggested him
and wrote to urge him to accept the appointment which the
President immediately offered to him. He accepted the place and
held it for a year, but it proved a serious interruption in his legal
career. He resigned as soon as it was possible and began practice
with Major H. P. Lloyd who had been his father’s partner before he
went to Vienna.
Mr. Taft went abroad in the summer of 1883 to visit Judge and
Mrs. Taft in Vienna, and it was about this time, when we had all
spent several years in frivolities, that several of us became very
serious-minded and decided that we must have something by way of
occupation more satisfying than dancing and amateur theatricals. I
secured a position as school teacher and taught for two years, first at
Madame Fredin’s and then at White and Sykes, both private schools
out on Walnut Hills. Then, with two of my intimate friends, I decided
to start a “salon.” We called it a “salon” because we planned to
receive a company who were to engage in what we considered
brilliant discussion of topics intellectual and economic, and we
decided that our gathering should include only specially invited
guests. Among these were the two Taft brothers, Will and Horace,
and other men common friends of us all.
In view of the fact that two marriages resulted from this salon, Mr.
Taft has suggested ulterior motives on the part of those who got it up,
but there was no truth in the charge. We were simply bent on
“improving our minds” in the most congenial atmosphere we could
create, and if our discussions at the salon usually turned upon
subjects of immediate personal interest, to the neglect of the abstruse
topics we had selected for debate, it was because those subjects were
just then claiming the attention of the whole community.
Cincinnati, thanks to the activities of Tom Campbell and his
followers, was then in a tangle of political mismanagement of a
particularly vicious character, and our little circle developed a civic
spirit which kept us alive to local interests to the exclusion, for the
time being, of everything else. Mr. Taft was intimately connected
with the reform movement, and in all its phases, through comedy
and tragedy, disappointment and elation, we fought it out at our
salon meetings with such high feeling and enthusiasm that its history
became the history of our lives during that period.
Then came the famous Berner case. This was in 1884. Berner had
committed a deliberate murder of an unusually appalling nature and
with robbery as the motive, and there was great excitement about it.
Campbell became his counsel and, in a trial which held the attention
of the community while it lasted, he succeeded in getting the man off
for manslaughter when the unanimous opinion was that he should
have been hanged. Nobody could see how an honest jury could have
rendered any other verdict. There was intense indignation
throughout the city and a meeting was called to denounce Campbell
as an embracer of juries and a suborner of perjury.
On the evening when the meeting to denounce Campbell was
called we were having a session of the salon and our whole
discussion was of the possible developments which might grow out of
the infamous Berner trial. We were greatly excited about it. I
remember the evening distinctly because of the terrible things that
happened. We were disturbed by a great commotion in the street and
we sallied forth in a body to see what it was all about.
The mass meeting was held at Music Hall and was presided over
by Dr. Kemper, a very effective speaker. The crowd was angry and
quickly passed the condemnatory resolutions which were framed.
But with all the indignation and resentment everything might have
been carried out quite calmly had not the match been applied to the
powder. Just as the meeting was breaking up somebody shouted:
“Let’s go down to the jail and take Berner out!”
It was an appeal to the mob spirit which responds so readily in an
angry crowd; they went; and of course the worst elements
immediately came to the top. They attacked the jail, which was in the
rear of the court house, but were held back until the militia, which
had been instantly summoned, arrived. Then they went around to
the front and set fire to the court house. With the streets packed with
raging humanity it was not possible to fight the fire and the building
was completely destroyed.
The militia charged the mob and this inspired somebody with the
idea of raiding a gun store and seizing arms and ammunition with
which to make a resistance. The idea caught on and spread rapidly.
One place attacked was Powell’s gun shop near Fourth and Main. But
Powell, either forewarned or foreseeing some such development, had
quietly made preparations to meet it. He lighted up the front of the
store as brightly as he could, then, with two or three other men who
were expert shots, he put himself behind a barricade in the rear. The
mob came on and as the ringleaders broke into the shop they were
picked off by the men behind the barricade and killed in their tracks.
Four or five of them went down in a heap and the crowd behind
them, not expecting such a reception, instantly was brought to its
senses. This was in April, 1884.
Such an outbreak was a disgrace to the city of Cincinnati, but it
had the effect of bringing the Campbell controversy to a head. A bar
committee of ten men, of which both my father and Mr. Taft were
members, was formed to see what could be done to rid the
community of the evil reputation it had acquired. This committee
made a thorough investigation of Campbell’s character and record,
prepared charges against him and, with my father as chairman,
presented them, in June, 1884, to the district court of three judges,
and asked a hearing and Campbell’s disbarment if the charges were
proved.
 MEMBERS OF THE SALON. MR. TAFT IN THE CENTER, WITH
THE AUTHOR AT HIS RIGHT

Campbell had been indicted on a criminal charge of attempting to

bribe a man called on the Berner jury and the prosecutor in this case
was our intimate friend and associate, Mr. Rufus Smith, who had
been in Europe with Mr. Taft the year before. The jury hung, eight to
four, although the evidence was strong against the defendant. This
fanned the flames of popular resentment and I don’t suppose our
little salon was the only place in Cincinnati where Campbell was
carefully retried and convicted. In this criminal case Mr. Foraker,
who shortly afterward became Governor of Ohio, was counsel for
Campbell.
The disbarment hearing was set for the following November and
some six months was thus given for taking the depositions of non-
resident witnesses. Mr. Kittredge and Mr. Ramsey, leaders of the bar,
were retained as senior counsel for the committee, and Mr. Taft and
Mr. John Holmes, a warm friend of ours, were junior counsel and
were directed to prepare the evidence. In this work Mr. Taft and Mr.
Holmes went all over the country taking depositions and we kept in
constant touch with them. All the members of the committee
expected to have their reputations assailed, being perfectly certain
that Campbell would not hesitate at any measure he might be able to
take to discredit them, but they went ahead nevertheless.
 When the trial came on Mr. Ramsey, of the senior counsel,
expected to open the case, but he became quite seriously ill and was
confined to his house for days. Through his unexpected absence, the
duty of making the opening statement fell to Mr. Taft. He was taken
completely by surprise, but he rose to the opportunity, which was
certainly a splendid one for a man so young. He had then been at the
bar only four years, but having assisted throughout in the
preparation of the evidence he knew the case from beginning to end
and he made a speech which lasted four hours and a half. Mr. Taft
thinks this was an opportunity improved which had an important
influence on his career. The special part it played in his subsequent
promotion I shall speak of.
The result in the Campbell case was at first disappointing because
the Court which heard the disbarment charges found Campbell guilty
only on minor charges and, by a vote of two to one acquitted him on
those which would have required his disbarment. But the public
disapproval of the Court’s decision and the moral effect of the
proceedings drove Campbell from the city and the State and
accomplished the purpose of the bar association.
The Campbell trial was finished in December, 1884, and in
January, 1885, Mr. Rufus Smith, an old and intimate friend, entered
the office of County Solicitor and tendered to Mr. Taft the place of
Assistant County Solicitor. The advantage of this office was that it
paid $2500 a year and that, while he acted as counsel for the county,
he still was able to continue the general practice of law with his
partner, Major Lloyd.
Mr. Taft and I were engaged in May, 1885, and were married in
June of the following year.
In the summer of 1885 my mother, moved I think by some
sentimental attachment to the scenes of her childhood, decided that
she would take us all up into the Adirondacks, to a little camp near
Lowville. My two older sisters were married so there were only six of
us left in the family, but we were still something of a handful to move
in a body. However, my mother was equal to it. We packed almost a
van load of trunks and set out, and one evening we arrived, over the
worst corduroy road that was ever laid down, at a little cottage beside
a beautiful lake in a setting of pine-clad hills. The scenery indeed was
most satisfactory, but the cottage was so small that the family more
than strained its capacity. Then we took our meals at a sort of
boarding house called Fenton’s, where the only thing on the bill of
fare was fresh beef. I like what is known as “roughing it” as well as
anybody, but even the superlative appetite produced by outdoor
living demands some variety; and variety we did not get.
Mr. Taft had elected to remain in Cincinnati all summer and save
money. It was a Spartan resolution and we all applauded it, but he
probably found Lowville a long way from Pike Street; and I certainly
thought Mother was sacrificing a good deal for the sake of renewing
the memories of her youth. However, the days went on, while the
fresh beef grew less and less tempting.
I had written Mr. Taft something about the Fenton fare and he,
wanting very much to join us, but having no excuse for breaking his
admirable resolution to remain in Cincinnati, hit upon the only plan
for escaping comment on his lack of fortitude. He went down to
Peeble’s, a fancy grocer, and selecting a box as big as a Saratoga
trunk, ordered it filled with every kind of delicacy he could think of
or have pointed out to him and brought it with him to Lowville.
We went rowing on the lake about sundown the evening he
arrived, and right in the middle of a fine long stroke he suddenly
dropped his oars, reached in his pocket and drew out a letter. He
laughed a little when he handed it to me, then picking up his oars he
rowed on without a word. The letter was from his father.
Judge Taft was at this time Minister to St. Petersburg, having been
transferred from Vienna. Will had written him about his engagement
and about his plan to remain in town all summer and devote himself
strictly to business and the accumulation of funds; and this was the
answer.
There were a lot of nice complimentary things about me, with the
warmest congratulations and good wishes; then the letter closed by
saying: “I am very much pleased with your decision to remain in
Cincinnati this summer. I myself have found it not at all bad if you
take care of yourself, and there is no doubt that during the quiet
months one can make and save considerable money by staying at
home. I congratulate you on your strength of character.” We really
had a delightful summer at Fenton’s after that.
 My father had given me a very nice lot at the end of McMillan
Street on the site of an old quarry, which commanded a fine view of
the Ohio River and the surrounding country, and Mr. Taft and I
determined to build a house on it which should be ready for us when
we got back from our wedding trip. So the winter before our
marriage was filled with architects’ plans, contractors’ estimates and
all the other fascinating details of building, and we thought that we
had finally settled upon a design that met with every requirement of
good taste and modern comfort.
For our wedding trip, we went abroad, and I had my first taste of
the foreign travel of which I had always dreamed. We crossed on the
City of Chester which was the oldest, and therefore the cheapest ship
of the Inman line. We chose her for the simple reason that her rates
accorded with our means, but we found, much to our astonishment,
that we were the only people on board who had deliberately selected
her. Everybody else had been forced to take her because of some
emergency or some mishap. One man had to miss the Germanic in
order to give his dentist time to relieve a very troublesome tooth.
Another man was called to court just as he was about to board the
Britannic. Those were the proud ships of the Atlantic in those days
and it was not at all difficult to understand why anybody should
prefer them to the City of Chester, but it amused us greatly to hear
the shamefaced excuses of our fellow passengers. My husband and I
were not ashamed, nor were we so particular about our comforts that
we did not thoroughly enjoy ourselves. Besides, we had the gratifying
consciousness of the money which the low rates had left in our
pockets to be spent much more profitably abroad.
 MR. AND MRS. WILLIAM HOWARD TAFT AT THE TIME OF
THEIR MARRIAGE

The trip was full of interest to us both. We spent the greater part of
the summer in England and saw the sights of London and the
cathedral towns in great detail. Our only trip on the Continent was
through Holland to Paris. I remember that in Amsterdam I bought
some old and rather large Delft plates. They wouldn’t go into any
trunk we had, so I had them carefully packed in a wicker hamper and
this article became thereafter a part of our hand luggage, and was the
occasion for a decided disagreement between my husband and me as
to what the true object of travel was. He used to say that he “toted
that blamed thing all around Europe and after all it arrived in
Cincinnati with its contents in small pieces.” Which was true. He had
“toted” it all around Europe, but when we arrived in New York I
entrusted it to an express company with the result that when we
opened it we found its contents in such a condition that only an
accomplished porcelain mender could put a sufficient number of
pieces together to make what my husband always afterward referred
to as “the memento of our first unpleasantness.”
Our trip from Cincinnati to Cincinnati took just one hundred days
and cost us just one thousand dollars, or five dollars a day each. I
venture to say that could not be done nowadays, even by as prudent a
pair as we were.
 During a subsequent trip abroad, two years later, I was able to
indulge my desire to hear music. We went to Beyreuth, to the
Wagner festival, and heard Parsifal and The Meistersingers
gloriously rendered; after which we went to Munich and attended
operas and concerts until Mr. Taft rebelled. He said that he enjoyed a
certain amount of music just as much as anybody, but that he did
want to get something more out of European travel than a nightly
opera and a daily symphony.
So—we went to Italy and saw Rome and Florence in true
Baedecker style. When we arrived in Rome we opened our Baedecker
and read that there was almost no foundation for Rome’s awful
reputation as an unhealthy place. “Rome is a very healthy place,” said
Baedecker, “at all times of the year except the first two weeks in
August, when a visit there is attended with risk.” We had arrived for
the first two weeks in August!
When we came home from our wedding trip we found that our
house was not yet completed, so we went to stay with Judge and Mrs.
Taft for a month at the old house in Mt. Auburn. It was a nice old
place, with about three acres of ground, but the air around it was just
about as sooty as if it had been located down under the factory
chimneys. Mt. Auburn is on a sort of promontory which juts out into
the city; it is on a level with the tops of the smoke stacks and it
catches all the soot that the air can carry that far.
Judge and Mrs. Taft had come home from their European mission
in time for our wedding. Judge Taft had been ill in St. Petersburg and
had given his family a great deal of anxiety, but he was now settled
down to the business of quiet recuperation and the enjoyment of
well-earned rest.
My husband’s father was “gentle” beyond anything I ever knew. He
was a man of tremendous firmness of purpose and just as set in his
views as any one well could be, but he was one of the most lovable
men that ever lived because he had a wide tolerance and a strangely
“understanding sympathy” for everybody. He had a great many
friends, and to know him was to know why this was so.
Mr. Taft’s mother, though more formal, was also very kindly and
made my visit to her home as a bride full of pleasure. The two, the
father and mother, had created a family atmosphere in which the
children breathed in the highest ideals, and were stimulated to
sustained and strenuous intellectual and moral effort in order to
conform to the family standard. There was marked serenity in the
circle of which Judge and Mrs. Taft were the heads. They had an
abiding confidence in the future of their children which strongly
influenced the latter to justify it. They both had strong minds,
intellectual tastes, wide culture and catholic sympathies.
Not long after we arrived my husband came to me one day with an
air of great seriousness, not to say of conciliation and said:
“Nellie, Father has got himself into rather a difficulty and I hope I
can rely on you to help him out—not make it too hard for him, you
know,—make him feel as comfortable about it as you can. The truth
is he used to have a messenger at the War Department in
Washington whom he was very fond of. He was a bright man—
colored, of course—and he was very devoted to Father. Now this man
called on Father down town to-day. He’s here on a private car and
Father says he’s made a great success as a porter. Father got to
talking to him, and there were lots of things they wanted to talk
about, and besides the man said he would like very much to see
Mother,—and Father, who was just about ready to come home to
lunch said—right on the spur of the moment—you understand he
didn’t think anything about it—he said to this man, ‘Come on home
and have lunch with us.’ He’s downstairs now. Father came to me
and said he had just realised that it was something of a difficulty and
that he was sorry. He said that he could take care of Mother if I could
take care of you. So I hope you won’t mind.”
As soon as I could control my merriment caused by this halting
and very careful explanation, I went down to luncheon. I didn’t mind
and Will’s mother didn’t mind, but the expression on the face of
Jackson, the negro butler, was almost too much for my gravity. I will
say that the porter had excellent manners and the luncheon passed
off without excitement.
We made a short visit at my mother’s on Pike Street before we
moved into our new house on McMillan Street; but we began the
year of 1887 under our own roof which, though it was mortgaged,
was to us, for the time being, most satisfactory.