Text Entry Does Not Imply “English Text Entry”

Abstract

The title above is an actual sentence borrowed from MacKenzie and Soukoreff's seminal article on Text Entry for Mobile Computing (2002), which I found to be enlightening many years ago, and has inspired my latest work on text entry.

-

Permission to make digital or hard copies of part or all of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for third-party components of this work must be honored. For all other uses, contact the owner/author(s).

Copyright held by the owner/author(s).

MobileHCI, 2018 Barcelona, Spain.

My latest work on text entry has been directed at improving the stimulus phrases when conducting evaluations with non-native English speakers. Together with my co-authors, I devised a couple of automated phrase sampling methods that produce phrases that are memorable (easy to remember), representative of the language or task, and easy to assimilate. This paper provides a brief overview of such methods as well as other contributions I have made to the text entry community.

Author Keywords

Text Entry; Phrases; Sampling; Multilingualism; Memorability; Representativeness; Semantics.

ACM Classification Keywords

H.5.m [Information interfaces and presentation]: Miscellaneous.

Biography

Luis Leiva is working as researcher and Chief Technological Officer at Sciling, a Machine Learning company based in Spain. He is a former research staff member of the PRHLT research center at the Technical University of Valencia (UPV), where he worked as Principal Investigator. He got his PhD degree (with honors, international mention, extraordinary doctorate award) in Computer Science in 2012. He also has two undergraduate degrees—Industrial Design and Industrial Engineering—and an M.Sc. in Communications and Mobile Services Development (2006) from the UPV. His research interests include Human-Computer Interaction, Machine Learning, and the intersection between the two. His recent work on text entry includes Back-of-Device interaction [ 1, 3, 4, 5 ], new miniaturized QWERTY keyboard layouts [ 6, 7 ], and multilingual phrase sampling methods [ 2, 8, 9 ].

Overview of Past and Current Work

While this position paper will focus on phrase sampling, I believe it is worth mentioning my previous work on text entry as well.

Back-of-Device Interaction

I have investigated two contexts of use for Back-of-Device (BoD) interaction. The first one is BoD Taps [ 1, 5 ], an unlock technique that was found to be very usable and theoretically more secure than their peers. The second one is βTap [ 3, 4 ], a tap-based BoD input technique that uses a low-cost yet highly discriminative set of features from commodity sensors, without the need to instrument the device.

Tiny QWERTY Keyboards

With the ongoing breakthrough of wearables, such as smartwatches or digital jewelry, text entry on devices with very small screens (1” wide or less) becomes increasingly relevant and a challenging issue, simply because space is at a premium. I contributed in this regard with ZShift [ 7 ], a callout-based text entry technique for diminutive QWERTY keyboards. I also investigated cost-efficient error auto-correction mechanisms [ 6 ], that can be easily implemented in current wearables. Barcelona,

Multilingual Phrase Sampling

This is the current line of research I am mostly involved in, motivated by the fact that text entr y is fundamentally [S] mWIKI:EN:Inuteraction_l(albtum)ilingual. Many text entr y researchers are conducting user studies in many languages different from English, and the tasks given to par ticipants are often language-sensitive. This suggests that, when conducting text entr y experiments with non-native English speakers or in ver y specialized domains (e.g. medical devices, where technic a[S]WIKlI:EN:Hvumano_(Goldcfrap W_asIoKnI:gE)Nb:Wil _GuregorylarWyIKI:EN:Eilecstronic_music commonplace), we either use a standWaIKI:ENr:Golddfrap phrase set and accept that there will be differences in performance across studies, or we have to develop language- or d o[S]WIKI:mEN:Human_(Tahe_Kil ersi_sonng)-specific phrase sets. To solve this problem, automated phrase sampling methods like MemRep and Kaps become necessar y.

On the one hand, MemRep [ 8, 9 ] was the first phrase sampling method that optimized both for memorability and representativeness (dual-objective function). Our experi mWIKI:ENe:Human_(nStargate_tUnivaerse) [S] l results showed that it performed [S] WIKI:EsN:Compiuterg_(magazinne)ificantly better than other sampling procedures. the other hand, Kaps [ 2 ] sought a balance between memorability, representativeness, and complexity (grammar and semantics subtleties). We found th introducing this third variable, the method not only better p[S]journral#n#o5per ties than MemRep but also that the resulting phrases exhibited better performance. at, by had

On

Proposed Scenario

I believe my [S] rWIKIe:EN:Humacn_(Branedy_Norwo nd_albumt) rese[S] aWIKI:ENr:Humcan_(banhd) on text entr y is of special relevance to text entr y researchers interested in conducting experiments tailored to the linguistic capabilities of their par ticipants. Therefore, I would tcoonccornettreiblyu,t eI wtoo uthlde lwikoer ktos[S] WIhKId:EN:Huomain_(vBabpylone_5)winit hthtehiismtpoopri cta. nMceortehat like [S] WIKI:EN:Lights_(El ie_Goulding_album) [S] WIKI:EN:Projected [S] daybo k#n#1 datum#n#1 [S] read#v#11 computveirr_tugaral_prheciaoclmsit#ypn#ur#ente1s#ro1_lunteiotwn#onrk##7nte#l1ephroandeio#t#nen#l#2e1cinofomrmmatucionnoi_cmteaecthplieonuonclot#tgednryoar##_nt1n#as1i#ccnsime##panr1tnohpgce#rroaem1gma#rmatiwmncreersrm#ia#t#idnen1ni#ngn#1__g1rl#aennc[S#og]1ruudan#gdnee#r#s1nta#n1d#v#3 engi njoeuerrinnga#l#n#nr2e#a2d#v#1 expert#n#1 lIcsFeaotvicnigmamrguelmuuumrausiecaalaol_tiunsetrtung3mseindto_d:eigcitalpn_isnTosterfahscenho#no#.1rceanfeMaKslpyreoa,thsrprwieneda[suSst]mrehiaolpWelv_cetpIrnthKigcyiiarvsnIeeil:ec_aEcaeerl_niNnpgegchi:#neChnyens#eoo1bmrhimrnegnep#o,nnue#h#nrnt1ge#en1nreet_ecqloireeuaeaeschalme_qnt[ntnlrSriecgopt]ml_ueaamlibnevemtniarnvaauecei_grtgakynmcregn#ry#tyrndhee_erov#piut#i#cee1nisianes4hrmdcngmcttaoi_hcinprp#lia#aesirenlcd#git1ins#nseari##u#2ntupyei1onnnh#n#n#h#ei#4svnnm1Wynese###osscr11t1IsiaiiorKaheectnree#t#nIsncpeaic:fnanoeleti#Ec[rh#sc#Ss#yc1eino1N]ansn##itls#atneu_:1plsnmtN_di1henrce##arotananemu1oc#otlaptrtuet1yoiuikgpoatmreemennarr#cd#a#nwnslnraent#nc_no#on1b#letei1eadnr3#a#te#cu#sinsnnnnyrrne#b#v#o##m#n#io11in2#nnaar2r#1gctinao#3#hiimn3nwnmlgh#ecsma1#etsopalnmi#nmro#elntoppm1ei#cue#nfnltrtre1i#ieisntnecgt4roep_#sendItuaayv#e#snlo_en#nnpnbdgrl##uo#id#nigontss1ii1#t1#ooax1mnlnv1_1u#dli#argnusn#rs#i##i1tnsn2rtnea#ash#boftcn1eeoauitp#r1cummrtnaoeel##iacrgog1iaann-nlh#####unaangm3##n#En#11ae11nertststf#oueanrc#re1.#pnmhcp#yrcaur2enma#tanouo#nrhe3h_#bmno#aod2plrebys##jnennsc##ttt2#a1adennita#iiaiar1sylso#qt#nina#c#2s1s#nunn#1aeo[S]sjoufrnal#n#4 the current living languages bstuudy#n#t2 cer tainly hotg#nh#3blo ed#n#1re are may more languages wor th of evaluation. I look for[S] wread#va#5rd to discussing this with the workshop attendees .intestine#n#1 regression_analysis#n#1

personal#a#1 diarilsitte#rna#te1#n#1 reardeeard#enr##3n#1statistic#n#1 number_cruncher#n#1 proportionality#n#1subatrdaceter#r#nn##11

statistician#n#2 record#n#5 give#v#18

character#n#4

Pro least_squares#n#1 WIKWI:EIKNI::EBNia:sC_oonf_fidaenn_cees_timintaetrovral [S] WIKI:EN:Interaction_( WIKI:EN:Multivariate_normal_distribution

WIKI:EN:Mean_squared_er or covariance#na#n1alysis_of_variance#n#1 WIKI:EN:Gaus –Markov_theorem

variance#n#3 normal_distribution#n#1

References

1. Catalá , A. , and Leiva , L. A. Back-of-device authentication with bod taps and bod shapes . In ACM Conf . on Human-computer interaction with mobile devices and ser vices (MobileHCI) ( 2014 ).

2. Franco-Salvador , M. , and Leiva , L. A. Multilingual phrase sampling for text entr y evaluations . International Journal of Human-Computer Stu 113 , 1 ( 2018 ).

3. Granell , E. , and Leiva , L. A. Less is more: Efficie back-of-device tap input detection using built-in smar tphone sensors . In Proceedings of the ACM Conf. on Interactive Surfaces and Spaces (ISS) ( 2016 ).

4. Granell , E. , and Leiva , L. A. βTap: Back-of-device tap input with built-in sensors . In Proc. ACM Conf . on Human-computer interaction with mobile devices and services (MobileHCI) ( 2017 ).

5. Leiva , L. A. , and Catalá , A. Bod taps: An improved back-of-device authentication technique on smartphones . In Proc. ACM Conf . on Human-computer interaction with mobile devices and services (MobileHCI) ( 2014 ).

6. Leiva , L. A. , Sahami , A. , Catalá , A. , Henze , N. , and Schmidt , A. Error auto-correction mechanisms on tiny qwerty soft keyboards . In CHI Workshop on Text Entry on the Edge ( 2015 ).

7. Leiva , L. A. , Sahami , A. , Catalá , A. , Henze , N. , and Schmidt , A. Text entry on tiny qwerty soft keyboards . In Proc. SIGCHI Conf. on Human Factors in Computing Systems (CHI) ( 2015 ).

8. Leiva , L. A. , and Sanchis-Trilles , G. Representatively memorable: sampling the right phrase set to get the text entry experiment right . In Proc. SIGCHI Conf. on Human Factors in Computing Systems (CHI) ( 2014 ).

9. Sanchis-Trilles , G. , and Leiva , L. A. A systematic comparison of 3 phrase sampling methods for text entry experiments in 10 languages . In Proc. ACM Conf . on Human-computer interaction with mobile devices and services (MobileHCI) ( 2014 ).