Research Limitations

3.6.1 Limited Sample Size

Sample size is a key factor in reducing variability and increasing reliability of

findings. Generally, larger sample sizes are more reliable and produce lower

variability and error. However, limited population size has always been a grounded

limitation of research in conference interpreting. Apart from this present study, only

one other study has attempted to target the population size of conference interpreters

in Taiwan (Government Information Bureau, 2004). By way of snowball sampling, a

target population of no more than 53 conference interpreters was generated. Similar to

other observational or experimental studies in conference interpreting, the small

sample size in this research is largely due to the relatively small interpreter population

to begin with, which is a limitation constitutional to researches of the kind.

3.6.2 Limited Research Scope

The target population of the research is limited to conference interpreters who

simultaneously translated for a conference during the data collection period of the

research. First, the samples do not include court, medical, escort, or other types of

interpreters. This is largely due to the fact that court, medical, and escort interpreters

in Taiwan come with a wide spectrum of background in terms of their education,

training, interpreting techniques, and the actual difficulty of interpretation they are

required to perform on an ad hoc basis. Including them in the study would be to

investigate a population of which the size and members are yet to be defined.

Second, to investigate interpreters’ anxiety at work, the researcher has also

limited the research scope to a particular interpreting mode: simultaneous conference

interpreting for the convenience of data collecting. This is mainly because, first, over

90% of the interpreters approached by the researcher indicated that their next

interpreting assignment would be simultaneous rather than consecutive. Second, since

a team of two interpreters are required to perform simultaneous interpretation,

studying simultaneous conference interpretation greatly increases the researcher’s

chance to collect two research samples, instead of one in the case of consecutive

interpreting. Third, one sampling error often resulted from snowball sampling is that

certain groups out of the social or economic network of the initial contacts may be

accidentally left out. However, the fact that conference interpreters work in a team of

two helps minimize such sampling error. As a result, the personality profile and

anxiety level of court, medical, escort interpreters and consecutive conference

interpreters remain to be investigated.

3.6.3 Non-random Sampling Techniques

The representativeness of the 30 volunteer interpreters in this research remains

to be explored. Due to the absence of any interpretation certification exam and

conference interpreter’s labor unions or organizations of the kind, definition of

conference interpreters in Taiwan vary from one to another, and the actual population

of Taiwan-based conference interpreters have never been delimited. Faced with a

hard-to-reach population that is yet to be defined, the researcher’s only resort was to

employ non-random sampling techniques such as snowball sampling.

Snowball sampling has proved to be useful in identifying participants for the

research since it is a highly affordable and simple way for the researcher to build up

samples for data collection. To minimize sampling errors due to the non-probability

sampling method as mentioned in section 3.3.2, the researcher first strived to generate

as many contacts as possible in order to reduce the chance that the interviewees

belong to a special or minority group within the target population. Second, to make

sure that the initial interviewees are truly members of the target population defined by

this study, the researcher asked every interviewee to identify whether s/he is an active

conference interpreter. Third, to make sure that certain interpreters in different market

segments are not left out of the survey, the researcher requested interviewees to

nominate active interpreters outside their direct social and economic network, such as

asking interviewees to nominate those who did not graduate from the same

interpreters’ school as them.

Moreover, the research sampling is a direct result of the subjects’ willingness

and availability to participate. The researcher is only able to investigate the mental

quality of those who volunteered instead of every member known to the researcher.

The personality profile and anxiety level of those who declined the survey remain to

be examined.

As previously noted, the sampling technique is a non-probability method, which

therefore reduced external validity and the ability to generalize to the larger

population of conference interpreters in Taiwan. However, tentative broader

conclusions pertaining to conference interpreters’ personality and anxiety level may

be drawn which then can be tested in future research.

3.6.4 Lack of Long Term Observation

The research is a cross sectional empirical study that involved the measurement

of the subjects’ mental status within a narrow time span so that the measurements may

be collected and viewed as contemporaneous. In this research, data regarding

conference interpreters’ mental status were collected at only one point in time.

However, since one’s Trait anxiety and personality are “the relatively enduring

pattern of recurrent interpersonal situations which characterize a human life,” (Harry

Stack Sullivan, 1953) the cross-sectional can be considered acceptable under such

circumstances. The State Anxiety Test was administered one time before a volunteer

interpreter enters an interpreting situation. Limitations inevitably arise due to such

cross sectional research design: First, the test score is only representative of the

subject’s anxiety level prior to that particular conference; hence, the subject’s average

anxiety level in all conference interpreting situations remains unknown. Second, the

State Anxiety Test result is representative of the subject’s anxiety level no earlier than

one hour before the conference started. The test result can not be generalized to

represent the subject’s anxiety level out of that designated time frame. Nevertheless,

limiting the completion of the State Anxiety Test to one hour prior to a conference, as

previously noted, minimizes the potential influence of the interpreter’s

non-occupation-related experiences on his or her anxiety level as the interpreter fills

out the anxiety questionnaire. Third, a conference interpreter’s anxiety level at work

fluctuates as s/he gains experience and competency. Change in interpreters’ anxiety

level at work can only be examined with long term observation instead of a one-time

test.

It’s noteworthy that cross sectional design in this research is appealing for

reasons of economy of time and cost. The researcher was not face with the difficulty

and cost of maintaining contact with the subjects. For the participants, there was only

one period for data collection. It is also worth noting that the conference interpreters’

willingness to complete the questionnaire on a busy working day greatly increased

when they learned that there was going to be only one period for data collection.