one-dimensional handwriting: inputting letters and words...
TRANSCRIPT
One-Dimensional Handwriting: Inputting Letters and Words on Smart Glasses
Chun Yu Ke Sun Mingyuan Zhong Xincheng Li Peijun Zhao Yuanchun Shi Key Laboratory of Pervasive Computing, Ministry of Education
Tsinghua National Laboratory for Information Science and Technology Department of Computer Science and Technology, Tsinghua University, Beijing, 100084, China
{chunyu, shiyc}@tsinghua.edu.cn, {k-sun14, zhongmy14, lixc12, zhaoym13}@mails.tsinghua.edu.cn
ABSTRACT We present 1D Handwriting, a unistroke gesture technique enabling text entry on a one-dimensional interface. The challenge is to map two-dimensional handwriting to a reduced one-dimensional space, while achieving a balance between memorability and performance efficiency. After an iterative design, we finally derive a set of ambiguous two-length unistroke gestures, each mapping to 1-4 letters. To input words, we design a Bayesian algorithm that takes into account the probability of gestures and the language model. To input letters, we design a pause gesture allowing users to switch into letter selection mode seamlessly. Users studies show that 1D Handwriting significantly outperforms a selection-based technique (a variation of 1Line Keyboard) for both letter input (4.67 WPM vs. 4.20 WPM) and word input (9.72 WPM vs. 8.10 WPM). With extensive training, text entry rate can reach 19.6 WPM. Users’ subjective feedback indicates 1D Handwriting is easy to learn and efficient to use. Moreover, it has several potential applications for other one-dimensional constrained interfaces.
Author Keywords Text entry; one-dimensional input; unistroke gestures.
ACM Classification Keywords H5.2. Information interfaces and presentation: User inter-faces — Input devices and strategies.
INTRODUCTION Nowadays, many new smart personal devices are emerging. Such devices typically have a constrained input interface due to the limited size of the form factor. Consequently, text entry is difficult on these devices, which prohibits their broader use. In this paper, we focus on one-dimensional text entry for devices that have only one-dimensional input
signals, or the input capacity on one dimension is much greater than that on the other. Examples include the touchable spectacle frame of a smart glass (e.g. Google Glass), the edge of a side screen of a smart phone (e.g. GALAXY Note Edge), and a smart wristband.
Researchers have proposed various techniques to enable text entry on constrained interfaces (e.g. a keypad, a joystick, a tracking ball, and air interaction etc.). These techniques include coding letters [16], unistroke gestures [2], ambiguous keyboards [18], and gesture keyboards [12, 31]. However, most of these techniques are for two-dimensional constrained interfaces or interfaces with a limited number of buttons or gestures.
In this paper, we present 1D Handwriting, which enables users to perform unistroke gestures on a one-dimensional interface to input text. We research, implement and evaluate 1D Handwriting with Google Glass, a representative one-dimensional input device. The biggest challenge encountered is that many letters look similar, if not identical, after projected into a reduced one-dimensional space. Therefore, our goal is to derive a set of one-dimensional gestures that strikes a balance between memorability, input accuracy and input speed.
After a careful and iterative design, our result is a set of ambiguous unistroke handwriting gestures based on sub-stroke of two-level lengths. To complete the input, we design a one-dimensional gesture recognition algorithm that classifies input strokes in a probabilistic way. For letter input, we design a smooth transition gesture allowing the user to first gesture a stroke (representing a group of letters) and then select the target letter (within the group) without lifting his/her finger. For word input, we use a Bayesian approach that takes both the probability of gesture and the language model into account to interpret users’ input.
To evaluate the performance of 1D Handwriting, we conduct three lab experiments. Results show that 1D Handwriting provides immediate usability: With limited training, users can achieve 4.67 WPM for letter level input (error rate = 0.78%) and 9.72 WPM for word level input (error rate = 0.25%). These speeds are comparable to those proposed for two-dimensional constrained interfaces, and significantly outperforms the one-dimensional selection-based technique (the Baseline in our experiments) by 11.4%
Permission to make digital or hard copies of all or part of this work forpersonal or classroom use is granted without fee provided that copies arenot made or distributed for profit or commercial advantage and that copiesbear this notice and the full citation on the first page. Copyrights forcomponents of this work owned by others than ACM must be honored.Abstracting with credit is permitted. To copy otherwise, or republish, topost on servers or to redistribute to lists, requires prior specific permissionand/or a fee. Request permissions from [email protected]. CHI'16, May 07-12, 2016, San Jose, CA, USA © 2016 ACM. ISBN 978-1-4503-3362-7/16/05…$15.00 DOI: http://dx.doi.org/10.1145/2858036.2858542
How Fast Can You Type on Your Phone? #chi4good, CHI 2016, San Jose, CA, USA
71
ftrlWingain
Iwpr1cf
R
LHabcMa
MoingWtoWQrs
F
UFuf[R[sWug
for letter inputraining focuseetters, 7 distin
WPM after thndicate that 1
gestures are eaapply 1D Handnterfaces.
In the remaindwork relevant tprocess for derreport three us1D Handwritinconclude our rfuture search di
RELATED WO
Letter input onH4-Writer [16]and leverages Hby considering could be achievMeanwhile, a gachieve 5.3-6.6
Multi-step interon constrained nput letters (e.
gestures (5.4 WWriter [7] uses o recursively
WPM) (FigureQWERTY keybregions. A firsspecifies the ch
Figure 1. Matrix(left and
Unistroke letteFigure 2 showunistroke vocabfast to write, a2] was desi
Researchers sh15]; but after
sessions), UnisWPM vs. 11.4used for eyes-gesturing in the
t and 20.0% foed on a singlenct gestures), the tenth input.D Handwritingasy to remembdwriting to othe
der of this papto our researchriving the 1D Hser studies thatng for inputtingresearch with irections.
RK
n constrained ] uses four buHuffman codinletter frequenc
ved after ten 35gesture-based v6 WPM using a
raction is anotuser interface.
.g. with a WiimWPM) (Figure
horizontal andreduce the lete 1, right). board on a verst swipe specifharacter within
x-based layout d middle) [11]; L
ers were first pws some samp
bularies. Unisand less prone igned to res
howed users cor extensive trstrokes could WPM) [5]. Ufree input on
e air [17].
or word input.e phrase (6 wotext entry rate. Users’ subjeg is fun to usber. In the futer one-dimensi
per, we first rh. We then introHandwriting get examine the g letters and wdiscussion of
interfaces ttons to input ng to minimizcy. A 20.4 WPM5-50 minutes pversion of H4-a touchpad and
ther means for . GesText [11]
mote) with twoe 1, left and md vertical movetter range to thSwipeboard [ry small touchfies the regionthat region.
and tri-center lLURD-Writer (
proposed for pple letters of ttrokes [8] wasto recognition
semble handwould learn Graraining (twenty
be significanUnistroke letter
mobile devic
. With extensivords, 14 distine can reach 19ective feedbacse and its strokture, we plan ional constrain
review previooduce the desigestures. We thperformance
words. We finalf limitations an
letters (base 4e key sequencM text entry ra
practice session-Writer [3] coud a Wiimote.
inputting letteallows users
o-step directionmiddle). LURDements of curshe target one [6] divides th
hscreen into ninn, and a secon
layout of GesTe(right) [7]
pen-based inputwo well-knows designed to bn errors. Graffwritten Romaaffiti in minuty fifteen-phra
ntly faster (15rs could also b
ces [21] and f
ve nct 9.6 ks ke to ed
us gn en of lly nd
4), ces ate ns. uld
ers to
nal D-sor (8 he ne nd
ext
ut. wn be fiti an. tes ase 5.8 be for
Figu
EdgeWto enabmotionThe firwhere (Figuresequendesignlearnindevice[24], an
UniGe[23], wprimitigestureEvaluaWPMvisual designi
Word iAmbigis too numbeto consto overT9) anto possof lette
Gesturusing ainputspecialhave toword. interacdifficukeyboaUsers tcontrol
One-diselectioreducereduce
ure 2. Sample g
Write [28] was ble unistroke inn stability for rst version fixusers traversed
e 3). Gesture rnce of corners ted to mimic i
ng. Edgewrite s such as joystnd spatial keys
Figure 3. Sa
est [4] definedwhere each gesive motions. es based on oation results safter a brief lsimilarity to hing the gesture
input on consguous keyboard
small to accoer of signals (estraints from brload individu
nd resolve wordsible words in ers to keys can
re keyboards [a single unistrusing both lal advantage ofo lift up the peTherefore, a g
ction (e.g. Vultult. MotionInpuard that arrangtraverse the keller.
imensional texon-based. The
e the size of thes the QWERT
gesture of Unistr(right) [5]
originally propnput while prov
individuals wxed a square hd the edges anecognition wasthat were hit. Eits handwrittencould also be
ticks and touchs [29].
mple gestures o
a gesture alpsture was desig
Writing-withon-boundary anshowed text elearning sessio
handwritten lettes of individual
strained interfads are widely uommodate all e.g. keys or geboth the deviceal keys with md ambiguity bya predefined vbe done in dif
[12, 31] enablroke gesture. anguage and f gesture keyboen or finger be
gesture keyboature [17]) wherut [19] realizeges eight ambieys to input wo
xt entry techniq1Line Keyboa
he screen occuTY keyboard i
rokes (left) and
posed by Wobbviding high ac
with motor imhole over a to
nd diagonals tos performed ba
EdgeWrite’s alpn counterpartse adapted to ohpads [26, 30]
of Edgewrite [28
habet based ogned to have ah-joystick [10nd off-boundaentry rate achon. For both tters was the gul letters.
aces used when the letters or only
estures) are avae and/or user. Tmore than one y matching keyvocabulary. Thfferent ways [1
le users to inpThe algorithmgesture infor
oard is that usetween lettersard is suitable fre inputting a des an ambiguoguous keys intords with a Le
ques already exard [13] was p
upied by the keinto one row
d Graffiti
brock et al. curacy and
mpairments. ouchscreen o input text ased on the phabet was s for quick other input , Trackball
8]
n Wiimote at most two 0] defined ary strokes. hieved 4.5 techniques, uideline for
screen size y a limited ailable due The idea is letter (e.g.
y sequence he mapping 8].
put a word m interprets rmation. A sers do not within one for mid-air delimiter is ous gesture to a circle.
eap Motion
xist; all are proposed to eyboard. It containing
How Fast Can You Type on Your Phone? #chi4good, CHI 2016, San Jose, CA, USA
72
eight keys, with each key overloaded with multiple letters. The Minuum keyboard [21], based on 1Line Keyboard, enables text entry on Google Glass. However, there is only a demo video describing it; no implementation details nor performance data are available. Circle keyboard [20] organizes letters into a circle in alphabetical order. Users rotate a Wiimote to select target letters. The text entry rate was 10.2 WPM. Walmsley et al. [22] applied the source-channel approach to Circle keyboard to allow fast and imprecise input. Experiment results showed expert users could reach 21WPM after 6.7 hours practicing.
In sum, an input interface can be constraint in terms of the number of input signals (keys and usable gestures) or the number of dimensions of the input space. In this work, we target the constraint interface where touch can be performed only in one-dimensional space. Besides, we focus on handwriting approaches rather than selection-based ones for text entry.
STUDY 1: DESIGN OF 1D HANDWRITING GESTURES The goal was to gain an understanding about user acceptance of 1D Handwriting gestures as well as the input capacity of Google Glass, the device we used to experiment our idea in this research. To achieve this, we first derived a set of 1D Handwriting stroke gestures, and then asked individual participant to assess their intuitiveness by performing them on Google Glass. Based on the subjective and objective results, we agreed on the design guidelines for the final handwriting gestures, and revised our design.
Phase 1 —Initial design of handwriting gestures In our first trial, we adopted a user-participatory approach [25]. We recruited eight participants and asked them to design a set of one-dimensional gestures for Google Glass. The process lasted for one hour for each participant to design gestures for all twenty-six letters of the alphabet. In the end, four participants proposed a code-based design, while the other four proposed a handwriting design. Unfortunately, none of the obtained gesture sets was satisfactory, because they were not memorable. This result revealed the difficulty of designing appropriate 1D Handwriting stroke gestures, especially within a very limited time. However, according to the post-experiment interviews, participants (including those who proposed the coding-based design) consistently agreed that the handwriting design was much more recognizable and memorable than the code-based design.
We then decided to design a set of handwriting stroke gestures by ourselves. The guideline was to design a 1D Handwriting stroke gesture for each letter to best mimic their two-dimensional counterparts. Figure 4 summarizes the handwriting gestures. For some letters, the design was easy and intuitive. For example, the handwriting stroke gesture of “w” should be a four-stroke gesture “back—forward—back—forward”. However, for others it was not an easy task since several letters were similar, if not identical, when projected into a reduced one-dimensional
space. In such cases, we designed strokes with more details (e.g. adding a short stroke at the end to distinguish “q” from “y”), or use a dot to distinguish them, such as “o” and “v”, “b” and “k”, “i” and “s”, and “j” and “l”. The design of handwriting gestures employed strokes of three levels of lengths (i.e. short, medium and long) and a dot.
Figure 4. Handwriting gestures used in Study 1. Dots (“·”) were included in this design. However, these were
consequently dropped in favor of faster input speed. Arrows represent the directions for each sub-stroke in a gesture. They
are drawn continuously in a gesture.
Phase 2 — User assessment and data collection
Apparatus We used Google Glass for this experiment. As shown in Figure 5, there is a touchpad mounted on the right spectacle frame. The touchpad (7.0 centimeters long and 0.8 centimeters wide) supports up to three touch points with an input resolution of 1366×160. Although Google Glass is not a strict 1D input device, the vertical space is very limited for input. In addition, there is a virtual screen (resolution: 640×360) and a built-in speaker on the device. The Google Glass system was XE 22 running on Android 4.4.4. We developed a software that recorded all touch events, including time and coordinate information.
Figure 5. The touchpad configuration of Google Glass
Design and procedure Twelve participants (six males and six females; Age: 18-25) were recruited from Tsinghua University campus. None had previously used a Google Glass. The experiment took about one hour to complete. At first, each participant was asked to memorize the handwriting gesture set (Figure 4). Then, they were instructed to perform these gestures (five times per letter) on Google Glass at their comfortable speed. For reference purposes, a sheet containing every letter-gesture pair was placed beside participants in case they forgot a gesture during the experiment. At the end of the
How Fast Can You Type on Your Phone? #chi4good, CHI 2016, San Jose, CA, USA
73
eow
RMraucaoUlcdmsinin
Idsssbpbmino
Fm
PBrwsetowrwggd
experiment, paof each individwe collected 15
Results Memorability. resembled twadjustments (i.eusers to rememcomments, the an arbitrary waout. ParticipanUnfortunately, earning the s
constantly referdifficulty. Mormemorability wstroke gesturesndividual partnterpreting the
Input accuracdistinguish betwshows the diststrokes after besub-stroke witbetween the thparticipants occbeginning or emore difficult tntend to drop s
of a two-level d
Figure 6. Beta medium and sho
Phase 3 — FinBased on the requirement ofwhich posed astroke gestureefficiency. Insto represent mo
was inspired bresolving the lwith a small nugathered duringgestures, we fodesign.
articipants rateual gesture on
560 handwritin
Although thewo-dimensionae. short strokes
mber actual strosmall adjustm
ay in that no cnts had to rem
this was not estroke gesturerred to the geseover, participwas diverse, es, (e.g. “q”, “ticipants shoule handwriting s
cy. It was dween three levtribution of seing normalizethin a gestureree-level lengtcasionally perfend of a strokto recognize byshort strokes frdesign of strok
distribution foort strokes.
nal design of tabove results
f one-to-one ma major challenes with gooead, we alloweore than one ley an ambiguolimitation of inumber of buttog the initial dermed the follow
ed the perceivea 5-point Like
ng gestures for
e overall handl handwritins and dots) mad
okes. Accordingments seemed tconsistent rule member the steasy given the s. The fact tsture sheet alsoants’ subjectivven for the sa
“h”, “g” etc.). d have discrep
stroke gestures
difficult for vels of stroke short, mediumed to the lengte. There wasths of sub-stroformed unintenke, making shoy the algorithmrom our designe length (medi
or the relative
he handwritins, we decided mapping of genge for designd memorabiled a handwritinetter if necessarus keyboard thnput resourcesons). Based onesign of the hanwing guideline
ed memorabiliert scale. In tota26 letters.
dwriting strokng, the smade it difficult fg to participanto be definedcould be draw
trokes by healimited time f
that participano confirmed th
ve assessment oame handwritin
This suggestpant models f.
participantslength. Figure
m and long suth of the longes much overlkes. In additio
nded hooks at thort strokes evm. Therefore, wn and be in favium and long).
lengths of lon
ng gestures to remove th
stures to letterning handwritinity and usagng stroke gestury. This decisiohat was also fs (e.g. a keyp
n the informationdwriting strokes for the gestu
ity al,
kes all for ts’ in
wn art. for nts his on ng ed for
to e 6 ub-est ap
on, he en we vor
ng,
he rs, ng ge
ure on for ad on ke
ure
(1) Mgetraea
(2) Mstr
(3) Mpe
(4) Sineff
The debetweeprovidpromotwe firsletter performin eacselectiostroke dimens
Figureof Goo
stahan
counte
Figure The demost owe des26 lettlettersdealt din a olettersdegreedimensthus as“n” to
Mimic traditionaestures shouldaditional handwasy to be remem
Minimize levelsroke gestures a
Minimize numberform stroke gngle stroke inp
fficiently.
esign goal waen speed, accue immediate te simplicity rst designed a “b
without concmed a few adjch group to ron within a ggestures to m
sional handwri
e 7. Gesture desogle Glass is plaarting with a cirndwriting. Letteerclockwise. Bla
for on
7 illustrates tesign was baseof these are pesigned 13 strokters of the alpassociated wit
differently withone-dimensiona“e”, “z”, and
es counterclocsional space. Lssigned the ge it. We also i
al handwritingd be easy towriting, the obmbered. s of stroke lengaccurately and er of sub-strok
gestures efficieput: In order to
as to obtain a uracy, learnabil
usage. One rather than distbest” stroke gecerning abouustments to lim
reduce ambigugroup. We de
mimic the sub-iting and to be
sign for individuaced to the left frcle are the firsters “e”, “s”, “z”ack arrows reprne-dimensional
the final stroked on glyphs oerformed with ke gestures, wh
phabet. There th each handwh letters that lal space if m“s”, their glyp
ckwise and thLetter “x” has esture of anothintroduce shor
g: 1D Handwrio learn. By btained strokes
gths: In order efficiently.
kes per letter: ntly. o perform strok
comprehensivlity and memobasic principtinction. To acesture for each
ut ambiguity.mit the numbeuity and facilesigned 1D H-stroke sequenas succinct as
ual characters. for reference. Rt strokes in trad” are rotated 90resent the propol gestures.
ke design for ef lower-case lea single strokhich were mapwere no more
writing stroke glacked optimal
mapped directlyphs were first hen mapped a two-stroke her two sub-strthand in the
iting stroke mimicking
s should be
to perform
In order to
ke gestures
ve trade-off orability to le was to
chieve this, h individual
We then er of letters litate letter
Handwriting nce in two-
possible.
The image Red arrows ditional degrees osed design
each letter. etters since
ke. In total, pped to the e than four gesture. We l mappings y. For the rotated 90 to a one-glyph. We
troke letter design. In
How Fast Can You Type on Your Phone? #chi4good, CHI 2016, San Jose, CA, USA
74
trawo
TIthapltaWin
GThrNreb
T
SWoppdspa
SIarp
SWdln
SWspws
wg
raditional handa backward (towith a backwaour design for s
THE TECHNIQIn this section,he technique u
algorithm to recprobability of etters. We thenarget words b
We also descndividual char
Gesture recogThe purpose ihandwriting strrather than to dNote that accorepresent moreestimated resulbased word-lev
The recognition
Step #1: FilterinWe first identioccur when strpoints of a stroproceed to elidistance is lessegmentation ispoints. The deras … .
Step #2. RemoIn order to deaat the beginniremove the firspixels (7.5 mm
Step #3. NormaWe normalize tdividing the lenength of sub
normalized in a
Step #4. CompWe only compsub-strokes is eprobability of we assume thestrokes to be in
where the probgiven according
dwriting, letteroward the bodyard stroke. Thesimplicity and
QUE , we describe
used in 1D Hancognize gesturan observed
n present our y taking advaribe the interacters and wor
gnition s to estimate roke input ( )discretely classording to our de than one lt to produce le
vel input.
n algorithm has
ng and segmenify the “turninroke direction oke are two deiminate consess than 20 pixs performed brived sequence
val of unintentil with unintentng and end ost and last sub) in length.
alization the length of angth of the lonb-strokes in advance.
utation of a pospute the templaequal to the deother templatee estimation o
ndependent from|bability densityg to the result o
rs “n”, “m”, any) stroke, whileese strokes weusability.
the implemenndwriting. We fres, which estim
gesture givenBayesian algo
antage of the lraction designrds respectively
a posteriori p) given a stroksify it into a spdesign, a strokletter. We canetter-level inpu
s four steps:
ntation. ng points” in tchanges. The efault turning pecutive turningxels (1mm) asased on the ree of sub-stroke
ional strokes tional hooks thof a handwriti-strokes that a
an individual sungest sub-strok
the gesture
steriori probabate gestures werived gesture. es is set to zerof individual nm each other. W|
y function (PDof Study 1.
nd “p” start wie letter “d” en
ere eliminated
ntation details first describe thmates a posterin by individuorithm to predilanguage moden, for inputtiny.
probability of ke template (pecific templatke template man then use tht and dictionar
the stroke, thestarting and enpoints. Then wg points whos noise. Finallemaining turnines is represent
hat usually occing gesture, w
are less than 15
ub-stroke ( ) bke. Note that th
templates a
ility whose number
The a posterioro. Furthermornormalized suWe have
DF) of |
ith nds
in
of he ior ual ict el. ng
f a ),
te. ay he ry-
ese nd we ose ly, ng ed
cur we 50
by he
are
of ori re,
ub-
(1)
is
PredicA userone wletters,numbeinput aapproacan be
Becaus
We theother. W
That is
where
If no mauto-cothat hawordspresent
InteracFigure on Goois dividrow) ddisplaybottom
The hatranspaText reof the time ldimensdimensthe scr
cting target wor inputs a sequ
word ( ) from, that is, er of likely woand language
ach [9], the poscomputed as f|
se we have no ,en assume eachWe have
,s,
, is the stro
matches are foompletions byave more letters
are sorted,ted to the user
ction design 8 shows the in
ogle Glass thaded into three
displays the inpys the recogn
m row) displays
Figure 8. Inte
andwriting stroarent overlay oegion is directl
touchpad. Thline. Therefosional stroke gsional visual freen, providing
ords uence of strok
m a vocabulary… . Thords based on t
model. Similsterior probabilfollows: , ∗knowledge of ∗h letter input i
∗∗
oke template o
found, the algoy computing ths than the inpuaccording to in descending
nterface of theat users can seee regions. The put text. The Lnized letters. s the five most
erface design of
oke is renderedon top of Text ly mapped to thhe horizontal
ore, although gesture, users feedback of thg guidance duri
kes ( …y ( ). The w
he algorithmthe probabilitylar to Goodmlity of I given
, we can com/
is independent
|
|
of letter . orithm attempthe probability
ut sequence. Altheir probab
order [27].
e 1D Handwrite on the virtuaText region (
etter region (thThe Word relikely words.
f 1D Handwriti
d in real-time region. The hehe entire touchdirection reprthe input i
can still obsehe handwritinging input.
) to input word has n
predicts a y of gesture
man et al.’s a word (w)
(2)
mpute
(3)
from each
(4)
(5)
ts to make y of words ll predicted bility, and
ting system al screen. It the middle
he top row) egion (the
ing
on a semi-eight of the hable range resents the is a one-rve a two-
g stroke on
How Fast Can You Type on Your Phone? #chi4good, CHI 2016, San Jose, CA, USA
75
Irthfstopwwto
Imfasrththg
Femsof
DboMpto
AFbasstosdsthawtass
TGcrthfs
Input a wordrespective strokhat most likel
five) are showselect the mosouchpad with
performing a tword, followedword is selecteo the end.
Input a letter. mode. In orderfinger on the toafter performinswitch to letterregion highlighhe touchpad tohe selection.
gesture for inpu
For some letterend near one omovement in thsystem arrangopposite direcfrequently used
Deletion of a leby clicking observation, mMost likely duperpendicular to be performed
A VARIATION For the sake obased keyboardan ambiguous shown in Figusimilar to 1D ouch-selection
switch betweendown gesture switched). To ihe ambiguous
and then tap twword, users swap one finger t
select a word,selection.
To improve tarGlass, we impcontrol/display respectively. What users coul
fast stroke or strokes.
d. Users inpuke gestures of ly correspond
wn in the Worst likely word
two fingers, two-finger mod by a two-fined, the system
The system isr to input a leouchpad for 30ng a stroke. Ar input mode hted) allowing o select letters.
This design utting letters.
rs, the handwriof the edges ofhat direction des letters in
ction of the d letter selected
etter is performthe camera
most users prefue to the facto the main inpd on the touchp
OF THE 1LINEof comparison,d, referring to
keyboard basure 9, The inte
Handwriting n, rather than n letter mode a
(in our expnput a letter, ukey, swipe tw
wo fingers to cowipe one fingeto confirm the and then tap
rget selection lemented cursratios for slo
We empirically d traverse the
perform acc
ut words by each letter in sto the current
rd region. Usd out of them
or select the ovement to senger tap to coautomatically
s set by defauetter, users hav00ms (empiricAt that point, (with the framusers to moveLifting the finoffers a seam
iting stroke gef the touchpaddifficult. To de
frequency orlast stroke,
d at the endpoin
med with a swibutton. Acc
fer to use the ct that a swipput direction, thpad.
E KEYBOARD we implemen1Line Keybo
sed on QWERerface and theexcept that a 1D gesturing: and word moderiment, the
users swipe onewo fingers to onfirm the seler to select an selection, swip
p two fingers
with the touchsor accelerationow and fast fin
determined thentire keyboa
curate selectio
performing thsequence. Wort input (at moers can directby tapping thother words b
elect the desironfirm. When appends a spa
lt to word inpve to place theally determinethe system w
me of the Lette their finger o
nger will confirmless one-fing
esture is likely , making furth
eal with this, thrder toward thwith the mo
nt of the stroke
ipe-up gesture ording to ocamera butto
pe-up gestureherefore not ea
D nted a selectioard design [13RTY layout. Ae interaction a
user inputs bUsers manual
de with a swipmode was n
e finger to seleselect the lette
ection. To inputambiguous ke
pe two fingers to confirm th
hpad on Googn. We used twnger movemenhe parameters ard with a singns with slow
he rds ost tly he by ed
a ace
put eir ed)
will ter on rm ger
to her he he ost e.
or our on.
is asy
on-3], As are by lly pe-not ect er, t a ey, to he
gle wo nts so
gle wer
Figselecte
t
Becaus“ambigsimulaWe usthe soillustraKeybovocabuexperimdiction
Fi
USER In ordeinputtinof 1Lin
ApparaSixteenfrom campubefore.
F
gure 9. A variatied key (“edc”) ithe interface de
se both 1Lineguous” for ation to quantifed a dictionar
ource code ofates the resultoard exhibited ularies containments, we unary.
igure 10. Accur
STUDY 2: INPer to evaluate tng letters, we ne Keyboard.
atus and partin participants 18-30) were
us. None of the.
Figure 11. The
ion of 1Line Keis highlighted. Fesign is similar
Keyboard aninputting wofy their abilityry (with frequf the Androidts. Both 1D high Top-1 a
ning as many aused the top
racy for vocabul
PUT LETTERSthe performanccompared its
icipants (eight males recruited fro
em had previou
experiment sett
eyboard. The cuFor consistencyto 1D Handwri
nd 1D Handwrords, we pery to specify tarency data) derd Keyboard. Handwriting accuracy (ove
as 178,000 wor10,000 word
laries of varyin
S ce of 1D Handperformance a
and eight femom Tsinghua usly used a Go
ting of User Stu
urrently y purposes, iting.
riting were rformed a rget words. rived from Figure 10 and 1Line r 95% for rds). In the ds as the
ng sizes
dwriting for against that
males; aged University
oogle Glass
udy 2
How Fast Can You Type on Your Phone? #chi4good, CHI 2016, San Jose, CA, USA
76
Fsearrsth
DWo(fMsthinbrigt
BptfGBwDpoetm
TthcpppwlKGmst
RTapwbop
Figure 11 illusetting. The syenter on a laptaway from therested their elrequired inputssynchronized ahe laptop PC th
Design and PrWe used a witonly one indep(1D Handwritinfor each parMacKenzie ansame phrases fhe order of phrn one session.
block containirequired to breagnored for pechnique was c
Before the expparticipant. Weechniques wa
familiarize theGlass, which Before each tewarm-up sessiDuring the exphrases as fast of the experievaluation forechniques. Th
minutes.
The warm-up shat of 1Line
containing eveparticipants. Npractice the performing eacwere asked to etters appeared
Keyboard, partGlass until thmovement coselecting/deletiechniques laste
Results The main depeand uncorrecteprovide more dwe further divbetween the coof the next performing the
ustrates the eystem displayetop PC. Partici
laptop PC. Thlbow onto thes. The softwareall touch eventhrough Wi-Fi.
rocedure thin-subject dependent factor ng and 1Line Krticipant was d Soukoreff's for the two terases randomiz Sessions cons
ing two phrasak after finishinpractice. The counterbalance
periment, we e did not informs ours. Moreomselves with took approxi
echnique was ion with the xperiment, sub
and accurate aiment, particirm and provhe total experim
session of 1D Keyboard. F
ry letter-gestuext, we asked strokes on
ch gesture fivcomplete a shd twice in a raticipants practhey became ontrol and ng a letter). ed for about th
endent measureed errors), inpdetailed inform
vided its input ompletion of th
stroke), gestue handwriting
experiment ened phrases foripants sat apprhey wore a Ge desk while e on Google Gts during the e
esign for this ewas the Tech
Keyboard). Thrandomly g
phrase set [14chniques per pzed. Each technsisted of six bses for input.ng a block. Thpresentation
ed.
introduced itsm participants wover, we allowthe user intermately five tpresented, patested text en
bjects were reas possible. Fiipants filled
vided feedbackment lasted fo
Handwriting For 1D Handwure pair was fi
participants toGoogle Glasse times. Fina
hort practice taandom order. Aiced input phrfamiliarized wother input
Warm-up sehirty minutes.
es were input eput speed and mation about 1
time into elahe previous strouring time (stroke gesture
nvironment anr participants roximately 0.5oogle Glass anperforming th
Glass logged anexperiment wi
experiment. Thhnique employhe tested phrasgenerated fro4]. We used thparticipant, winique was testlocks, with ea Subjects we
he first block worder of ea
s goals to eacwhich of the twwed subjects rface of Googto ten minute
articipants had ntry algorithmequired to inpinally, at the en
a NASA-TLk on the twr roughly nine
was longer thwriting, a sheirst presented o memorize ans by correctally, participanask where all 2As for the 1Linrases on Googwith the tou
gestures (e.essions of bo
errors (correctinput time. T
1D Handwritinapsed time (timoke and the sta(time spent oe) and selectio
nd to
5m nd he nd ith
he ed
ses om he ith ed ch
ere was
ch
ch wo to
gle es.
a ms. put nd
LX wo ety
an eet to nd tly nts 26 ne
gle ch .g.
oth
ed To ng, me art on on
time (importnecessagesturelocatiohandw
CorrecFigure error ra
There werror rlow unHandwThe reif they
ANOV(F1,15=6correctTechniParticipthan w(SD=3performp<.005effect p=.29)
Figur
Input SFigure over thincludeaverag1D HaKeyboinput a4.20 WHandwaveragblock significtechniqKeyboeffect.(F4,60=
(time spent otant to note ary pause bees (to completon) and the t
writing stroke ge
cted and uncorr12 illustrates
ates migrated w
was no significratio (F1,15=1.7ncorrected erro
writing and 0.4eason was that
noticed it on t
VA showed 68.92, p<.000ted error rique Block wapants had more
with 1Line Key.69%). Howevm better (i.e. h5) with 1D Hafor 1Line Key.
re 12. The error
Speed 13 shows the
he five blocked the time u
ge, participantsandwriting an
oard. If we conat 4.67 WPM
WPM (SD=0.6writing was 8ge (F1,15=5.31,
(F1,15=8.70, pcant main effeques (1D Han
oard: F4,60=15.3There was no 1.37, p=.26) on
on performingthat the elap
etween succeste a stroke gestime participaestures.
rected errors s how the cowith blocks for
cant effect of T76, p=.20). Boor ratios: 0.78
48% (SD=0.65participants te
the virtual scre
significant 01) and Blockratio. The as also significe corrected err
yboard, 17.67%ver, through prhaving less corrandwriting. In yboard was no
rs for both techntesting result
e text entry raks. The calculusers spent ons input at 4.04
nd 3.71 WPMsidered only th(SD=0.61) w
61) with 1Lin8.9% faster th
p<.05), and p<.01). Mean
ect of block onndwriting: F4,60
39, p<.0001), interaction eff
n text entry rat
g the selectipsed time incsive handwritsture and resetants spent to
orrected and ur letter-level in
Technique on uoth techniques8% (SD=1.12%5%) for 1Line ended to correcen.
effects of k (F4,60=3.80,
interaction cant (F4,60=4.22rors with 1D H% (SD=6.10%)ractice, participrected errors) (contrast, such
ot significant (
niques with thets.
ates for both lation of text n correcting e4 WPM (SD=
M (SD=0.46) whe last block, p
with 1D Handwe Keyboard. H
han 1Line Ke11.2% faster
nwhile, we on text entry rat0=14.50, p<.00suggesting th
ffect of Techniqtes.
on). It is cludes the ting stroke t the finger
recall the
uncorrected nput.
uncorrected s exhibited %) for 1D Keyboard.
ct mistakes
Technique p<.01) on effect of 2, p<.005).
Handwriting vs. 5.51%
pants could (F4,60=4.68,
a learning (F4,60=1.39,
e statistical
techniques entry rate
errors. On =0.58) with with 1Line participants writing and Hence, 1D eyboard in in the last
observed a es for both 001; 1Line he learning que Block
How Fast Can You Type on Your Phone? #chi4good, CHI 2016, San Jose, CA, USA
77
i
InIti(Efinscsptile
Fslinthmdgwd“pgtir
Figure 13. Textand 1Line Keyin the 4th block
resulted fro
nput times of 1In average, elaime was 504m
(SD=151) inclElapsed time, gfor 45%, 16% as, the majority
next letter, recselecting lettercould be impsignificant effep<.0001), gestuime (F4,60= 11.earning curve
elapsed time an
Figure 14. Elaacross five bloc
Furthermore, aselection time fetter. From Fncreased as thehe structure o
medium lengthdependent on gestures as welwas no obvioudetected that fo“z”) elapsed participants spegesture. Interesime was short,
rather large.
t entry rates acryboard. There w
for 1D Handwrom uncontrolled
1D Handwritingapsed time wasms (SD=96), anluding 300ms gesturing timeand 39% of th
y of input timecalling the hans. Fortunately,
proved (Figureects of Block uring time (F4,6
04, p<.0001). Aof gesturing ti
nd selection tim
apsed time, gestcks for letter-lev
analysis of elapfor individual igure 15, we e number of suf sub-strokes i
h sub-strokes). the number oll as the order ous pattern foror some letterstime was relent more time rstingly, for the , both gesturing
ross blocks for was a sudden periting. We interd factors in the
g s 1397ms (SD=nd selection tim
spent on swe and selectionhe input time ree was spent onndwriting stro with practicee 14): RM-A
on elapsed t60= 5.41, p<.00A noteworthy ime was not asme.
turing time andvel 1D Handwr
psed time, gesletters was concan see that,
ub-strokes or thincreased (withSelection tim
f letters repreof letters insidr elapsed times that were selatively high, recalling its haletter “j”, althog time and sele
1D Handwritinerformance droprpret it as a nois
experiment.
=354), gesturinme was 1206m
witching moden time accountespectively. Thn considering thoke gesture, an, all input tim
ANOVA showtime (F4,60=8.701) and selectioaspect is that ths steep as that
d selection time riting in Study 2
sturing time annducted for ea, gesturing timhe complexity h both long an
me was primariesented by inpe a group. Thee. However, wldom input (e.indicating th
andwriting strokough the elapsection time we
ng p se
ng ms es. ed
hat he nd
mes ed
73, on he of
2.
nd ch
me of nd ily put ere we .g. hat ke ed
ere
Figurediffere
SubjecNonpacompaHandwp<.000much other m
Althouto permanagpracticintervigestureIn addifatigueon the 1D Hascreen participThese1D Han
USER The goHandwsixteenrecruiteseven appara
The dethat paletter-lrequire
Prior tohandwand to that pa
e 15. Elapsed timent letters. The c
same handwr
ctive feedbacksarametric Wilcared to 1Linwriting to be 01). On the othbetter evaluat
measures were
ugh at first it srform the hanged to quickly ce rounds for ew, every parties to be intuiti
dition, most pare with 1Line Ke keyboard to andwriting req
while performpants were mfindings sugge
andwriting.
STUDY 3: INPoal of this stu
writing and 1Lin participants ed again for tto fourteen da
atus as in Study
esign and procarticipants perflevel input. Ined to enter 18 p
o the experimewriting stroke ge
write them dowarticipants cou
me, selection timcolors separate
riting stroke for
s coxon signed-rne Keyboard,
more mentalher hand, theytion (z=-42.0, not statisticall
seemed difficulndwriting strok
master the ter each letter. icipant reporteve to recognizrticipants repo
Keyboard since control cursor
quired a lesserming stroke g
more familiar wested eyes-free
PUT WORDS udy was to evine Keyboard
who participthis study. Thays after Study 2.
cedure were siformed word-ln this experimphrases (divide
ent, participantestures for all 2wn on a piece
uld correctly r
me and gesturine letters with resr easy comparis
rank tests sugg, participantsly demanding
y gave 1D Hanp<.01). Diffe
y significant.
lt to rememberke gestures, pechnique in les
In the post-eed 1D Handwrize and easy to orted higher lev
more focus war movement. Ir degree of focestures, espec
with the strokee input was po
valuate and cofor inputting w
pated in Stude experiment
dy 2. We used
imilar to Studylevel input as oment, participed into nine blo
ts were asked to26 letters of thof paper. Resu
recall 89% of
ng time for spect to the on.
gested that felt 1D
g (z=-52.5, ndwriting a erences on
r and learn participants ss than ten experiment iting stroke remember. vels of eye as required In contrast, cus on the ially when e gestures. ssible with
ompare 1D words. The dy 2 were took place
d the same
y 2, except opposed to pants were ocks).
o recall the he alphabet, ults showed
the stroke
How Fast Can You Type on Your Phone? #chi4good, CHI 2016, San Jose, CA, USA
78
gwtredo
PatorT
RTas
CFe
TeloH
Wceas(s(
F
InFo
I1K1la9
gestures. Four well were “x”raining each p
each letter in adays, this resulof 1D Handwri
Participants fama warm-up sesso fifteen minu
required to inpThe total exper
Results The main depeand uncorrectedstandard deviat
Corrected and uFigure 16 illuerror rates migr
There was no serror ratio (F1,
ow uncorrecteHandwriting an
We observed acorrected errorerror ratio wasand 11.87% (SDsignificant eff(F7,105=1.15, p=significant int(F7,105=0.46, p=
Figure 16. The
nput Speed Figure 17 showover the eight b
In average, par1D HandwritinKeyboard. Tha1Line Keyboarast two blocks
9.72 WPM (SD
letters that pa, “d”, “j” andparticipant hadverage) and thlt positively inditing stroke ges
miliarized themsion before theutes for each
put phrases as riment lasted fo
endent measured errors) and intions. Therefor
uncorrected erstrates how thrated with bloc
ignificant effec,15=1.21, p=.29ed error ratiosnd 0.07% (SD=
a marginal signr ratio (F1,15=s 17.38% (SDD=5.81%) for fect of Block=.034), suggeteraction effe
=.86) was obser
corrected and utechniques a
ws the text enblocks.
rticipants inputng and 7.14 Wat is, 1D Handwrd in average (Fs were consideD=2.67) for 1D
rticipants could “t”. Consided (no more thhe interval of sdicated the goostures.
mselves with the task, which l
technique. Pfast and accur
or about eighty
es were input enput speed. Ale, no outliers w
rrors he corrected acks for word-le
ct of Techniqu9). Both techns: 0.25% (SD==1.42%) for 1-L
nificant effect o4.51, p=.051)
D=8.67%) for 1Line Keyboa
k on correctsting the learect of Techrved.
uncorrected erracross blocks
ntry rates for
t at 8.84 WPMWPM (SD=1.writing was 23F1,15= 6.92, p<ered, text entryD Handwriting
ld not remembering the limitehan 20 times f
even to fourteod memorabili
he techniques asted around tarticipants werate as possibl
y minutes.
errors (correctll data are withwere removed.
and uncorrectevel input.
e on uncorrectniques exhibit=0.07%) for 1Line Keyboard
of Technique o. The correct1D Handwritinard. There wasted error rat
rning effect. Nhnique Blo
ror rates for bot
both techniqu
M (SD=2.79) wi.21) with 1Lin3.8% faster tha
<.05). If only thy rates raised
g and 8.10 WP
ber ed for en ity
in en
ere le.
ed h 2
ed
ed ed
1D d.
on ed ng s a tio No ck
th
ues
ith ne an he to M
(SD=1was 20blocks
ANOVtext enlearninof Tech
F
Input tiOn ave(SD=6be intewhen c1Line 1D HaStudy techniq
Figua
Auto-coAuto-c1D Hadifferenp<.000input vphrase5.17 WThe recompleHandwstroke the lap
.34) for 1Line0.0% faster th (F1,15=5.86, p<
VA also showedntry rates (F7
ng effect. Therhnique Block
Figure 17. The
imes erage, the elaps
66) longer thanerpreted as thecompared to tKeyboard, the
andwriting was2, we observe
ques (p<.05), a
ure 18. Slide timacross eight blo
ompletion completion accandwriting an
ence was st01). In other wvia auto-comps at 8.30 WPM
WPM (SD= 0.8eason why partetion for 1Lin
writing, as pargestures, they
ptop PC screen
e Keyboard. Than 1Line Ke<.05).
d a significant7,105=11.43, p<re was no sign
k on text entry r
learning curve
se time of 1D n that of 1Linee recall time fthe slide time e gesturing tims much smallered significant as shown in Fig
me, gesturing timocks for both tec
counted for 6.nd 38.10% fortatistically sigwords, if we
pletion, particiM (SD=2.70) f86) for 1-Line kticipants took
ne Keyboard warticipants perfy tended to loon instead of th
That is, 1D Heyboard in th
t main effect o<.0001), suggnificant interacrates (F7,105=1.2
of both techniq
Handwriting we Keyboard, wfor gestures. In
(1,593 ms, Sme (427 ms, Sr. Similar to thelearning effectgure 18.
me and elapsed chniques in Stu
51% of inputr 1Line Keybgnificant (F1
did not consiipants would hfor 1D Handwkeyboard over more advantag
was as followsformed the hok at the test phe Google Gla
Handwriting e last two
of Block on gesting the ction effect 25, p=.28).
ques.
was 289 ms which could
n addition, D=132) of
SD=38) for e results of ts for both
time and udy 3.
letters for board. The ,15=191.49, ider letters have input
writing, and all blocks.
ge of auto-s: With 1D handwriting phrases on
ass display.
How Fast Can You Type on Your Phone? #chi4good, CHI 2016, San Jose, CA, USA
79
Tnpcpthwpath
UTbfinatwn[wsao
RWa1lrla
AorposMthsru
LTp
They only turneeded to selecparticipants cocontrol cursor mparticipants to he post-experim
with participanparticipants maauto-completioheir eyes reste
USER STUDY The goal of thibe following exfive participantnput a single
and Soukoreff’wo trials were
number of word6]. However, i
was based on aserved as a rouauto-completioof a word with
Figure 19. The
Results showedWPM (SD=3.5achieved 24 W16 WPM. As earning effect
rate. In averagast block than
According to pobservations, arates was the liperforming fasovershot the toslow down finMoreover, we he design of
speed and expreasonably reflused 1D Handw
LIMITATIONS AThis research point to the dire
rned to the vct words. In constantly lookmovement. Thnotice the au
ment interviewnts. We think ay be more liken function witd on the virtua
4: EXTENSIVEs study was to xtensive trainits who participphrase (random’s phrase set [1e ignored as prds, the trainingit is important
a single phrase,ugh estimation.n so that partichandwriting st
learning curve
d the mean tex50) for the last
WPM, while theFigure 19 shof Block (F9,6
ge, participantsin the first one
participants’ suba major factor imited input cast gestures, thouchpad. In sunger movemenconfirmed wiexperiment: P
perience obtainlect their profiwriting for a lon
AND FUTUREhas the follo
ection of future
virtual screen ontrast, with 1
ked at the viherefore, it wasuto-completion w, we confirme
in actual reaely to take full th 1D Handwr
al screen.
E TRAINING estimate how
ing. To achievpated the Studymly selected f14]) for twelveactice. By focu
g time could beto note that gi
, the obtained s In this study, cipants had to troke gestures.
e of 1D Handwr
xt entry rate cophrase. The fae slowest part
hows, there w3=6.16, p<.000
s entered 33.95e.
bjective feedbafor prohibiting
apacity of Goohe finger ofteuch cases, parnts to produceth participants
Participants thned in the ex
ficient input spnger period.
E WORK owing limitatioe works.
whenever th1Line Keyboarirtual screens more likely f
suggestions. ed this differenal-life scenarioadvantage of thriting by havin
fast a user coue this, we asky 3 to repeatedfrom MacKenze times. The firusing on a smae greatly reduciven the practispeed could onwe switched oinput each lett
riting in Study 4
ould reach 19.6astest participaicipant achiev
was a significa01) on text ent5% faster in th
ack and our owg high text entgle Glass. When went off rticipants had e correct inpus with regardhought the inpxperiment coupeed if they h
ons, which al
ey rd, to
for In
nce os, he ng
uld ed
dly zie rst all ed ice nly off ter
4.
61 ant ed
ant try he
wn try en or to ts. to
put uld ad
so
First, ofor lowmore cnon-Labetter band 1Dto remeoverloaaddrescharactthe “fllower c
Secondparticipand rethere iemploy[1]) to to testtraining1D Hlearnin
Third,handwbe notadvantas the sensorshandwsmartpedge),
Finallyinputtin1D hakeyboa
CONCWe exform interfachandwanalyzeHandwimplemGlass. suggesrememfeasibi
ACKNO
This wof ChiTsinghand No
our current 1Dwer-case Englcharacter setsatin alphabets.be an easy-to-r
D strokes; otheember. In addiading a single s this with ter sets. For elip up” gesturcase, or even b
d, regarding lpants could lea
ecall 89% of tis still space foy a well-desigfacilitate gestu
t the performg. Now, in allandwriting di
ng time.
in this researcwriting. Howev
t optimal for tage of the vert
three-point tos) for input.
writing techniqphones with a
smart wristban
y, in current resng letters. It isandwriting forard. This might
LUSION xplored the po
of constraince. The basic
writing strokesed the usabili
writing gestumented and ev
In user studiested 1D Hand
mber and efficiility of 1D Han
OWLEDGMEN
work is supportina under Grahua Universityo. 2015130077
D handwritinglish letters. It , such as num One requiremrecognize maperwise, the 1Dition, supportin stroke with mmode selectio
example, on Gre to switch bbetween numbe
learnability, ouarn the 1D strothe strokes aft
for improvemengned learning iure learning. M
mance after lol three experimid not conver
ch, we confirmver, the current
Google Glasstical input and ouch and even
It is also que on other
curved-edgednds, or even 2D
search, we focus interesting to r word level t further impro
ssibility of texned interface:
idea was to into a one-ity issues of
ures. Subsequvaluated 1D Hs, both objecti
dwriting strokeient to use. O
ndwriting.
NTS
ted by the Natant No. 61303y Research Fun71.
g gestures wereis important
mbers, punctument is that thpping between D strokes will bng more characmore characteron to specify
Google Glass, wbetween upperers and punctua
ur research shokes with a fewfter one week.nt. For exampinterface (e.g. Moreover, it is ong-term and
ments, the inpurge due to th
med the feasibt interaction ds: We did notother input senn the embeddinteresting to1D interface
d screen (e.g. D touchscreens
us on 1D handexplore the poinput, i.e. 1
ove the text ent
xt entry on an: the one-dproject two-d
-dimensional sdesigning uniuently, we
Handwriting wiive and subjecte gestures we
Our research v
tural Science F076 and No. nding No. 20
e designed to support
uations and here should 2D strokes be difficult cters means s. One can y different we can use r case and ations
howed that w practices, However,
ple, we can OctoPocus interesting extensive
ut speed of he limited
ility of 1D design may t take full nsors (such
ded motion o test 1D es such as Galaxy S6 s.
dwriting for ossibility of D gesture
try speed.
n emerging imensional imensional space. We istroke 1D
designed, ith Google tive results re easy to
verified the
Foundation 61272230, 151080408
How Fast Can You Type on Your Phone? #chi4good, CHI 2016, San Jose, CA, USA
80
REFERENCES 1. Olivier Bau and Wendy E. Mackay. 2008. OctoPocus: a
dynamic guide for learning gesture-based command sets. In Proceedings of the 21st annual ACM symposium on User interface software and technology (UIST '08), 37-46. http://dx.doi.org/10.1145/1449715.1449724
2. Conrad H. Blickenstorfer. 1995. Graffiti: Wow! Pen Computing Magazine, pp. 30-31.
3. Steven J. Castellucci and I. Scott MacKenzie. 2013. Gestural text entry using Huffman codes. In Proceedings of the International Conference on Multimedia and Human-Computer Interaction (MHCI '13), 119.1-119.8. http://www.yorku.ca/mack/mhci2013e.html
4. Steven J. Castellucci and I. Scott MacKenzie. 2008. Unigest: text entry using three degrees of motion. In CHI '08 Extended Abstracts on Human Factors in Computing Systems (CHI EA '08), 3549-3554. http://dx.doi.org/10.1145/1358628.1358889
5. Steven J. Castellucci and I. Scott MacKenzie. 2008. Graffiti vs. unistrokes: an empirical comparison. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (CHI '08), 305-308. http://dx.doi.org/10.1145/1357054.1357106
6. Xiang 'Anthony' Chen, Tovi Grossman, and George Fitzmaurice. 2014. Swipeboard: a text entry technique for ultra-small interfaces that supports novice to expert transitions. In Proceedings of the 27th annual ACM symposium on User interface software and technology (UIST '14), 615-620. http://dx.doi.org/10.1145/2642918.2647354
7. Torsten Felzer and Rainer Nordmann. 2006. Alternative text entry using different input methods. In Proceedings of the 8th international ACM SIGACCESS conference on Computers and accessibility (Assets '06), 10-17. http://dx.doi.org/10.1145/1168987.1168991
8. David Goldberg and Cate Richardson. 1993. Touch-typing with a stylus. In Proceedings of the INTERACT '93 and CHI '93 Conference on Human Factors in Computing Systems (CHI '93), 80-87. http://dx.doi.org/10.1145/169059.169093
9. Joshua Goodman, Gina Venolia, Keith Steury, and Chauncey Parker. 2002. Language modeling for soft keyboards. In Proceedings of the 7th international conference on Intelligent user interfaces (IUI '02), 194-195. http://dx.doi.org/10.1145/502716.502753
10. Zhenyu Gu , Xinya Xu, Chen Chu, Yuchen Zhang. 2015. To Write not Select, a New Text Entry Method Using Joystick. Human-Computer Interaction: Interaction Technologies, 35-43.
11. Eleanor Jones, Jason Alexander, Andreas Andreou, Pourang Irani, and Sriram Subramanian. 2010. GesText: accelerometer-based gestural text-entry systems. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (CHI '10), 2173-2182. http://dx.doi.org/10.1145/1753326.1753655
12. Per-Ola Kristensson and Shumin Zhai. 2004. SHARK2: a large vocabulary shorthand writing system for pen-based computers. In Proceedings of the 17th annual ACM symposium on User interface software and technology (UIST '04), 43-52. http://dx.doi.org/10.1145/1029632.1029640
13. Frank Chun Yat Li, Richard T. Guy, Koji Yatani, and Khai N. Truong. 2011. The 1line keyboard: a QWERTY layout in a single line. In Proceedings of the 24th annual ACM symposium on User interface software and technology (UIST '11), 461-470. http://dx.doi.org/10.1145/2047196.2047257
14. I. Scott MacKenzie and R. William Soukoreff. 2003. Phrase sets for evaluating text entry techniques. In CHI '03 Extended Abstracts on Human Factors in Computing Systems (CHI EA '03), 754-755. http://dx.doi.org/10.1145/765891.765971
15. I. Scott MacKenzie and Shawn X. Zhang. 1997. The immediate usability of Graffiti. Graphics Interface 97: 129-137.
16. I. Scott MacKenzie, R. William Soukoreff, and Joanna Helga. 2011. 1 thumb, 4 buttons, 20 words per minute: design and evaluation of H4-writer. In Proceedings of the 24th annual ACM symposium on User interface software and technology (UIST '11), 471-480. http://dx.doi.org/10.1145/2047196.2047258
17. Anders Markussen, Mikkel Rønne Jakobsen, and Kasper Hornbæk. 2014. Vulture: a mid-air word-gesture keyboard. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (CHI '14), 1073-1082. http://dx.doi.org/10.1145/2556288.2556964
18. Franck Poirier and Mohammed Belatar. 2007. UniGlyph: only one keystroke per character on a 4-button minimal keypad for key-based text entry. HCI International 2007, 479-483.
19. Shuo Qiu, Kyle Rego, Lei Zhang, Feifei Zhong, and Michael Zhong. MotionInput Gestural Text Entry in the Air. http://www.columbia.edu/~sq2144/MotionInput.pdf
How Fast Can You Type on Your Phone? #chi4good, CHI 2016, San Jose, CA, USA
81
20. Garth Shoemaker, Leah Findlater, Jessica Q. Dawson, and Kellogg S. Booth. 2009. Mid-air text input techniques for very large wall displays. In Proceedings of Graphics Interface 2009 (GI '09), 231-238.
21. The Minuum Keyboard. http://minuum.com/
22. William S. Walmsley, W. Xavier Snelgrove, and Khai N. Truong. 2014. Disambiguation of imprecise input with one-dimensional rotational text entry. ACM Trans. Comput.-Hum. Interact. 21, 1:4 http://dx.doi.org/10.1145/2542544
23. Wiimote, https://en.wikipedia.org/wiki/Wii_Remote
24. Jacob Wobbrock and Brad Myers. 2006. Trackball text entry for people with motor impairments. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (CHI '06), 479-488. http://dx.doi.org/10.1145/1124772.1124845
25. Jacob O. Wobbrock, Htet Htet Aung, Brandon Rothrock, and Brad A. Myers. 2005. Maximizing the guessability of symbolic input. In CHI '05 Extended Abstracts on Human Factors in Computing Systems (CHI EA '05), 1869-1872. http://dx.doi.org/10.1145/1056808.1057043
26. Jacob O. Wobbrock, Duen Horng Chau, and Brad A. Myers. 2007. An alternative to push, press, and tap-tap-tap: gesturing on an isometric joystick for mobile phone text entry. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (CHI '07), 667-676. http://dx.doi.org/10.1145/1240624.1240728
27. Jacob O. Wobbrock, Brad A. Myers, and Duen Horng Chau. 2006. In-stroke word completion. In Proceedings of the 19th annual ACM symposium on User interface software and technology (UIST '06), 333-336. http://dx.doi.org/10.1145/1166253.1166305
28. Jacob O. Wobbrock, Brad A. Myers, and John A. Kembel. 2003. EdgeWrite: a stylus-based text entry method designed for high accuracy and stability of motion. In Proceedings of the 16th annual ACM symposium on User interface software and technology (UIST '03), 61-70. http://dx.doi.org/10.1145/964696.964703
29. Jacob Wobbrock, Brad Myers, and Brandon Rothrock. 2006. Few-key text entry revisited: mnemonic gestures on four keys. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (CHI '06), 489-492. http://dx.doi.org/10.1145/1124772.1124846
30. Jacob O. Wobbrock, Brad A. Myers, Htet Htet Aung, and Edmund F. LoPresti. 2003. Text entry from power wheelchairs: edgewrite for joysticks and touchpads. In Proceedings of the 6th international ACM SIGACCESS conference on Computers and accessibility (Assets '04), 110-117. http://dx.doi.org/10.1145/1028630.1028650
31. Shumin Zhai and Per-Ola Kristensson. 2003. Shorthand writing on stylus keyboard. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (CHI '03), 97-104. http://dx.doi.org/10.1145/642611.642630
How Fast Can You Type on Your Phone? #chi4good, CHI 2016, San Jose, CA, USA
82