You are on page 1of 8

Perception of Vowels and Suprasegmental Features in Cross-language

Research: Literate Review


Literature Review

Among the literature on the topic of L2 phonology most of the research has been done

in the area of production of L2 sounds. In recent decades the research on the perception has

become widely spread and with this the connection between the perception and production is

being investigated more closely. None of the researchers is completely separating these two

phenomena, but there are disagreements to which extent the perception is influencing the

production. Since the topic of this literature review is limited to the problem of perception I

will not further discuss the literature on the production or the connection between production

and perception. As an introduction to the problems, theories and findings of L2 phonology

there are a few collections of these issues summarized. One of the most important ones is the

collection made by W. Strange in 1995 (Strange, 1995). Some of the most prominent theories

are presented in this collection and it is mostly based on perception problems. Among the recent

publications, the collection made by M. L. Zampini and J. G. H. Edwards is a good summary

of the achievements in the phonology of L2 in general (Zampini and Edwards 2008). For the

summary of research in the field of production a literature review made by T. Piske, I. R. A.

MacKay and J. E. Flege in 2001 can be a good starting point.

When it comes to the perception, the most common approaches are research on the

infants’ perception or cross-language research. In the first type, linguists are mainly interested

in the development of the perceptual space and its reorganization, and some of the most

important questions is at what age is the perceptual space finally organized. Research has

shown that there is high probability that the perceptual reorganization of perceptual space

according to the native language happens during the first year of life (Best, McRoberts, Sithole,

1988; Werker and Tees, 1984). After the first year of life the ability to perceive new sounds is

becoming smaller due to the lack of attention given to the categories or features that were not

acquired during the first year.

1
The cross-language research, which is the topic of this literature review, is interested in

the possibility of changing the perceptual space that was acquired in the childhood: the

questions that are of the highest interest include the possibility of expanding the space with

new categories and the factors that would influence changes in this space. In the next part I will

present some of the most important characteristics and findings of recent cross language

research.

1.1. Segmental and suprasegmental level in cross-language research

Until now, segmental level was investigated in more details than suprasegmental level

even though it has been shown that the instruction of prosody is more helpful for the

comprehensibility of speech in both production and perception than the instruction in the

production and production of phonemes of L2 (Lee, B., Guion, S. G, Harada, T., 2006). For

example, Flege, Munro and Fox (1994) have examined if the distance in the vowel space of

first (F1) and second formant (F2) of foreign categories influences the ability to make

distinctions among the vowels. Their finding that it is easier to perceive difference among

vowels that are further in the space of F1 and F2 then difference among closer vowels is taken

as a fact now days. Another research on the segmental level was conducted by Barrios, Jiang

and Idsardi (2016). They were investigating if there is difference in Spanish speakers’

perception of pairs of English vowels i-ɪ and a-æ. Following the theory that availability of

certain feature in native language (front/back vowel contrast in this case) would allow that

feature to be used for easier acquisition of foreign categories that use that feature for contrast

(a-æ contrast in this case) compared to the foreign categories that are not using that feature (i-

ɪ categories in this case), they made a research, which, in the end, gave results that actually

show no difference in perception of these two contrasts.

In regards to research of suprasegmental level, I have payed attention to the research

concerning tone, duration and stress placement because those are the features used in Serbian

2
prosody for lexical accent (Peco, 1988). Perception of tone was examined in languages that use

tome to signal lexical stress like Mandarin (Hallé, Chang and Best, 2004; Hao, 2012),

Cantonese (Francis, Ciocca, Ma and Fenn, 2008) or Thai (Kaan, Wayland, Bao and Barkley

2007) while duration was investigated in languages that use duration as a distinctive feature

like Japanese and Arabic (Tsukada, 2012). Since the topic of my research proposal is Serbian

language, research on the duration of Serbian is also important (Krebs-Lazendic and Best 2013).

There is not much research done on the stress placement, but one of the papers investigated

perception of the stress placement among speakers of languages with predictable stress

(Peperkamp, Vendelin and Dupoux 2010). The authors predicted that speakers of Hungarian,

Polish, French and Finnish would have difficulties in perception of stress because in their native

languages stress is always found in the same position in word and the participants were

confirmed to show deafness to stress.

1.2. Theories

Most of the researchers conducting experiments about perception of foreign sounds

base their research hypothesis on prominent theories. The most commonly used theories are

Flege’s Speech Learning Model and Best’s Perceptual Assimilation Model. Here we present

two more theories that were introduced for theoretic base of the research in this literature

review. Most of the papers presented all of these theories as a base for the research and used

more than one to interpret results. The reason for this can be found in the fact that all of the

theories are based on the effect of L1 on the perception of phonology of L2. All of them agree

that effect of the native language is the most important factor and they are predicting the extent

and direction of this influence, while the researchers using them are trying to show the where

are the limits of this influence. I will explain the theories based on the theoretic part of the

research of Barrios, Jiang and Idsardi (2016).

3
J. E. Flege’s Speech Learning Model (SLM) is based on the proximity of L2 sounds to

L1 phonetic categories. The focus of this theory is on the learning: L2 learners’ success can be

predicted by whether or not a target language phone is ‘identical’, ‘new’, or ‘similar’ to the

native categories. The most difficult would be similar categories because it is hard to

distinguish them from native categories and make new distinction in our perceptual space. The

first part the results of Krebs-Lazendic and Best (2013) is discussed through the perspective of

SLM.

In the C. Best’s Perceptual Assimilation Model (PAM) it is considered that nonnative

sounds are mapped onto a listener’s native categories on the basis of articulatory similarities.

It predicts that the more similar the sounds are to the native ones the bigger is the probability

of their assimilation to native categories. Results from Hallé, Chang and Best (2004) were

interpreted in the framework of this theory.

P. K. Kuhl’s Native Language Magnet Model (NLM) is mostly concerned with the

development of the perceptual space with age. It is claiming that infants' ability to discriminate

speech sounds becomes increasingly specific to their native language as they age. One research

that is based on this model is Pajak and Levy (2014).

Brown’s Feature-based Model (FBM) is based on the belief that learner’s native

grammar constrains influence which nonnative contrasts he or she will be able to accurately

perceive and, therefore, limits which nonnative contrasts the learner will successfully acquire.

This approach was used as a theoretic base in Barrios, Jiang and Idsardi (2016).

1.3. Methods

1.3.1. Participants

In the experiments of the cross-language perception research one group of the

participants usually consists of native speakers of the target language and they are considered

a control group. The control group is compared to the other groups in the research. The profile

4
of the participants involved in other group(s) depends on the topic of the research. The

experiments are usually examining how are the categories of nonnative language perceived by

speakers of the language that doesn’t have the investigated categories. That’s why the groups

of nonnative speakers in these experiments are usually native speakers of languages that fit this

description. For example, Hallé, Chang and Best (2004) investigated only perception of

Mandarin tones among two groups, one native Mandarin group and one French group because

French language doesn’t use tone for lexical stress. The common case is also using three groups:

one groups of native speakers of the target language, one group of speakers that have the

examined categories in their native language and one group of speakers that don’t have

examined categories in their language. In this case, two latter groups are compared to each

other and they are compared to the native control group. An example can be found in Tsukada

(2012). Sometimes the interest of the research is only the difference between the nonnative

speakers, so they don’t use the control group (Hao, 2012; Pajak and Levy 2014). From the

perspective of the second language research, I believe that control group of native speakers is

necessary in order to show if there are difference from native speakers, what are those

differences and in which direction the instruction of the L2 perception and pronunciation

should go for the learners to be able to acquire native perception and pronunciation.

1.3.2. Stimuli

In regards to their production, there are two different kinds of stimuli. First ones are

stimuli pronounced by native speakers (Krebs-Lazendic and Best 2013; Hao, 2012) and the

others are synthetized stimuli (Hallé, Chang and Best, 2004; Escudero, Benders and Lipski

2009; Peng, Zheng, Gong, Yang, Kong and Wang, 2010). Both are used often, but researchers

prefer using synthetized stimuli in the research in which they are looking for the boundaries

between categories. In this case they would synthetize a certain amount of sounds that are in

between two prototypical categories and they would examine where is the boundary between

5
two categories in perception, as Hallé, Chang and Best (2004) did with Mandarin tones for

example. There is also difference if the stimuli used in the research are real words of the target

language or non-words were use as in Krebs-Lazendic and Best (2013) or Peperkamp, Vendelin

and Dupoux (2010).

1.3.3. Tasks

There are two types of tasks that are most commonly used in the perception experiments:

discrimination task and identification task. In the discrimination tasks it is only investigated if

participants can distinguish between the two stimuli, while in the identification task it is

questioned if subjects can assign stimuli to a certain category – if they can recognize the

category of the sound they hear.

Most common discrimination task formats are:

1. AX task – in this task there are two stimuli presented to the participants and

their assignment is to decide whether the two sounds are the same or different

(Pajak and Levy, 2014; Barrios, Jiang and Idsardi, 2016)

2. AXB (also XAB or ABX) task – after the participants hear the three stimuli,

this task requires from participants to decide if the target stimuli X is the same

as A or it is the same as B (Hallé, Chang and Best, 2004, Escudero, Benders

and Lipski 2009)

3. Oddity task – participants are given three stimuli and their task is to decide

which of the presented stimuli is the odd one among the three (Krebs-Lazendic

and Best, 2013; Flege, Munro and Fox, 1994)

For the identification task there are two forms that are used (Zampini and Edwards,

2008):

6
1. Open-set task – in this task participants are expected to write the category of

the sound they hear by themselves, there are no options from which they would

choose.

2. Closed-set task – in this task participants are offered options and after hearing

the stimuli they have to choose among those options which is the sound they

heared (Hallé, Chang and Best, 2004).

1.4. Factors influencing L2 perception

Among the examined factors that have an effect on the perception of foreign sound the

most important is age of acquisition. It is commonly agreed that the earlier the age of

acquisition is the better results in perception and production are expected, even though it is not

a rule and the other factors are being investigated. Some of the other factors that are considered

important or that are being investigated are length or residence (Flege, 1988; Flege, Munro and

Fox, 1994; Cebrian, 2006), formal instruction (Kaan, Wayland, Bao and Barkley 2007,

Wayland and Li, 2008), learner engagement and self-regulation (Moyer, 2014) and context

(Kingston, Kawahara, Chambless, Mash and Brenner-Alsop, 2009; Lee, Tao and Bond 2009).

For the length of residence none of the mentioned research is giving strong evidence for the

positive effect. Formal instruction and engagement and self-regulation have been proved to

help in the L2 perception, while for context there are different results and it should be further

investigated for any definite conclusions.

You might also like