O n f e r e n c e

Short paper 
The main goal of this pilot study is to explore the use of mobile text-to-speech 
synthesizers (TTS) as a pedagogical tool to improve the pronunciation teaching and 
learning of L2 French. In our investigation, we focus on the L2 acquisition of French 
liaison, a phenomenon by which a normally latent, word-final consonant is pronounced at 
the beginning of the following word if this word is vowel-initial (e.g., peti/t.a/mi => 
peti[ta]mi ‘boyfriend’). The main difficulty for learners is that the pronunciation of the 
latent consonant is not systematic. Based on the syntactic context in which it occurs, the 
production of liaison in French has traditionally been considered as compulsory, optional, 
or forbidden. Little production data are currently available to assess liaison errors of L2 
learners. Existing studies (e.g., Dejean, 1993; Mastromonaco, 1999; Thomas, 2004) 
show that L2 learners may show persistent production errors even at an advanced level. 
Furthermore, Thomas (2004) shows that a comparison of compulsory liaison with other 
pronunciations difficulties reveals that liaison accounts for nearly 20% of all the phonetic 
errors made by students. Such figures should clearly be cause for concern and remedial 
efforts on the pedagogical front. 
The literature on the pedagogical applications of TTS is scarce and, to our knowledge, 
non-existent in the context of mobile devices such as smartphones and smart media 
players (mobile TTS henceforth). However, the handful of studies available indicates that 
TTS has potential for the teaching of L2 pronunciation (Cardoso et al., 2012; Soler-
Urzúa, 2011
), particularly to raise learners’ awareness about certain language features in 
a personalized way. The quality of the synthesis has improved substantially over the 
years (Handley, 2009), and we believe that this is an appropriate time to start exploring 
this computer application, in a mobile environment, as a potential model for L2 speech. 
The main advantage of TTS is that it can be used as a mean of enhancing the L2 aural 
input both quantitatively and qualitatively. 
The use of mobile devices for language learning has sparked the interest of an increasing 
number of researchers over the last decade (e.g., Kiernan & Aizawa, 2004; Kennedy & 
Levy, 2008; Lu, 2008; Zhang, Song & Burston, 2011). Despite encouraging results, 
Kukulska-Hulme and Shield (2008) observed that Mobile-Assisted Language Learning has 
not yet been embraced on a large scale and has not yet received sufficient research 
attention toward its full potential as a pedagogic practice. Along the same lines, Joseph 
and Uther (2009) stress out that the value of using mobile devices and incorporating 
multimedia elements into language learning applications needs to be quantified with 
controlled experiments where control groups study on non-mobile platforms. The authors 
argue that these sorts of experiments should be a priority in future research. The current 
study addresses this recommendation by incorporating a control and a comparison 
(teacher-driven) group with characteristics similar to what is being proposed. To our 
knowledge, there are no studies that have investigated the use of TTS on mobile devices 
for pronunciation teaching and/or learning. 
Twenty-one adult students of French as a second language participated in this study. All 
participants were recruited from L2 French classrooms at two Anglophone universities in 
Montreal. They were either native English speakers or had native-like proficiency in the 
language. In addition, all participants had low-intermediate level in French and, 
accordingly, had not yet fully acquired the liaison. Following Cha
pelle’s (2001, 2012) 
recommendation for conducting methodologically convincing CALL research, this study 
followed a pretest/posttest/delayed posttest design within a mixed method approach to 
data collection. The participants were assigned to one of three distinct groups, each 
corresponding to an experimental group: TTS, NTTS (Non-TTS) and CTL (Control). 
During the treatment period, the participants were not informed about the nature of 
study, except that it was about “an app that could help second language 
learners to 
improve their French”. 

---

-235- 
2014 CALL Conference 
LINGUAPOLIS
www.antwerpcall.be 
The TTS Group corresponded to the group that practiced French pronunciation using 
mobile TTS on an iPod Touch, iPad or iPhone using a commercial (but free) TTS 
application. The students completed on a weekly basis, either at home or at the 
university, five 20-minute pronunciation activities consisting of noticing, listen-and-
categorize, and listen-and-repeat activities in French using the TTS software installed on 
their mobile devices. The TTS participants were asked to spend approximately one 
minute per word/phrase, depending on the level of difficulty of each target phrase, for a 
total of 20 minutes. The “Non
-
TTS Group”, on the other hand, did not have access to 
mobile TTS. However, they completed the same activities that the TTS participants did in 
individual, weekly 20-
minute sessions with a French teacher. Finally, the “Control Group” 
participated in weekly individual 20-minute meetings with the goal of practicing their 
conversation skills with a French teacher who provided no feedback. These sessions could 
be described as conversation classes, in which the participant and the teacher engaged in 
discussions of a variety of topics about school, aspirations, family, etc.
To measure the participants’ pronunciation capab
ilities in the pretest, posttest and 
delayed posttest, we employed Moodle, an interactive, multimedia tool with which the 
participants were familiar, since it is used in most courses at the university where the 
experiment took place. The task consisted of reading words and phrases aloud, which 
were recorded automatically using a Moodle plugin, Online Audio Recording, without the 
presence of the researcher or teacher. We targeted 15 occurrences of compulsory liaison 
and included a set of distractors to ensure that the participants would not become aware 
of the exact nature of the study. 
The results indicate that the group that was treated with mobile TTS outperformed both 
the Non-TTS and the Control groups in liaison production. The overall success of TTS 
group suggests that this type of learning environment is beneficial for the learning of L2 
French liaison and, we speculate, for the development of other related segmental and 
suprasegmental features.
Due to the nature and scope of the project and the hardware utilized (a mobile device), 
we encountered a number of challenges in the development of this study. Firstly, there 
was a certain lack of control over some of the learners’ time
-on-task: because students 
in the TTS group were asked to do the weekly activities using their mobile devices on 
their own (e.g., at home), we had no control over their time commitment to complete the 
assigned activities. To ensure that these participants completed the activities, they were 
asked to answer a weekly report, similar to the ones assigned to the Non-TTS group. In 
this report, the participants were asked to 1) rank how easy/hard it was to understand 
what they heard; 2) find out which words or phrases contained certain French phonetic 
features, segments or phonological phenomena (e.g., whether certain two-word 
combinations were pronounced together, whether a given word contained the rounded 
vowel /y/ as in “tu”); 3) pick the best
-sounding word, the worst-sounding word and the 
most difficult word to pronounce; etc. Secondly, because the mobile TTS app adopted 
requires constant connection to the Internet so that the target words and phrases can 
access an online database, there were reports of problems with the synthesis. In these 
cases, the participants were asked to complete the activities somewhere else, preferably 
within the university’s premises where the connection is usually more reliable. Finally, we 
acknowledge that despite the portability aspect of the hardware used in our study, its 
small size can sometimes be an obstacle, since only a small amount of information can 
be displayed at a time, and it can strain the eyes of those who use it for a long period. 
While these are centainly legitimate limitations of the type of TTS adopted, the positive 
attitudes of our participants (reported in oral interviews) suggest that they are prepared 
to tolerate them in exchange for the possibility that the technology may lead to an 
improvement in learning.

---

-236- 
2014 CALL Conference 
LINGUAPOLIS
www.antwerpcall.be