ATA Certification Pass Rates 2003-2013, 2004-2014, and Statistical Trends

By Geoffrey S. Koby
Reblogged from The ATA Chronicle with permission from the author (incl. the image)

ATA CertificationThe Certification Committee is happy to report here on certification pass rates for 2003-2013 and 2004-2014. The average certification pass rates for these two sets of data have remained relatively stable, although other factors in ATA’s Certification Program have changed somewhat in the past two data sets. The four sets of 11-year data that have been published in The ATA Chronicle to date (2001-2011; 2002-2012; 2003-2013; and 2004-2014) now allow for some interesting comparisons and analyses.

To describe the results effectively and avoid distortion, the information has been divided into two groups: 1) languages with 40 or more exams in the reporting period; and 2) languages with extremely low volume (ELV), defined as language pairs with fewer than 40 exams in the reporting period. In the following, we report summary statistics for the entire set of exams for 2003-2013 and 2004-2014, broken down by these two groups.

For 2003-2013, the overall pass rate was 14.47%. A total of 6,339 candidates (previous period: 7,033) took the exam in 29 language pairs (previous period: 29), and 917 exams were rated “pass” (previous period: 1,032). Of these language pairs, 16 had 40 or more exams over this period (previous period: 18). The Polish>English and Dutch>English exams have entered ELV status due to low demand for these language pairs, while Finnish­>English is no longer represented. However, Swedish>English has started as a new language pair.

For 2004-2014, the overall pass rate was 15.45%. A total of 5,463 candidates (previous period: 6,339) took the exam in 29 language pairs (previous period: 29), and 844 examinations were rated “pass” (previous period: 1,032). Of these language pairs, 16 had 40 or more exams over this period (previous period: 16). The individual language pairs are listed in Table 1 in alphabetical order with the number of exams and the individual pass rates per language pair for both sets of Table 1 Cert

In both data sets, 13 of the 29 language pairs had fewer than 40 exams. Table 2 shows the combined results for these language pairs. The data is presented this way because these language pairs cannot be averaged reliably due to their low volume. Another reason is that exams in some languages were not offered for the entire period. The Italian>English language pair was suspended in 2007 and was only reinstated in 2015, so it will remain in the ELV category for some time. In addition, Hungarian>English, which had a low volume to begin with, has been suspended since 2008, although work is ongoing to reinstate it.

Table 2 Cert

Figures 1 and 2 show the information on the two data sets in graphical form, in a format slightly different from previous pass-rate reports. The dashed horizontal red line shows the mean pass rate. No standard deviation is provided for the pass rate percentages because the language pairs have widely divergent numbers of exams. Overall, this figure shows that the pass rates differ for each language pair.

Figure 1 shows the pass rates for 2003-2013. The pass rates for the high-volume pairs range from 8.42% for English>French to 28.42% for English>Portuguese. The ELV languages have an aggregated average pass rate of 34.15% (3.23% of all exams), which represent 13 language groups averaging two or fewer exams per year.

Figure 1

Certification Forum Revised LONG.Figure 1

Figure 2 shows the pass rates for 2004-2014. The pass rates for the high-volume pairs range from 9.00% for Arabic>English to 28.97% for English>Portuguese. The ELV languages have an aggregated average pass rate of 35.88% (3.11% of all exams), which represent 13 language groups averaging two or fewer exams per year. A slightly higher or lower number of ELV exams passing in any data set can greatly skew the individual average.

Figure 2

Certification Forum Revised LONG.Figure 2

With four data sets with which to work, it is now possible to show some trends. Figure 3 shows that the number of exams has been declining over time, from 7,585 exams in 2001-2011 to 5,463 exams in 2004-2014. This is not surprising, as the number of candidates for the exam declined in 2002 with the implementation of eligibility requirements. The number of ELV exams has remained small but relatively stable, with just under 100 exams per data set into English and just over 80% into foreign. At the same time, the number of high-volume exams has declined 28% overall, with exams into foreign languages declining 27% and exams into English declining 30%.

Figure 3

Certification Forum Revised LONG.Figure 3

Figure 4 compares pass rates over time, using four data sets (2001-2011 through 2004-2014). The overall pass rate has remained largely stable, with a high of 15.64% and a low of 14.67%. The pass rate for high-volume languages closely mirrors the overall pass rate, just slightly below it, ranging from 15.16% to 13.81%. Not surprisingly, the ELV pass rate is quite a bit higher and more variable. The shift between the low 40% range in the first two data sets and the mid-30% range in the second two sets is attributable to a couple of language pairs with moderate pass rates moving from high-volume into the ELV category, pulling the average down. This did not have a noticeable effect on the high-volume pass rate, however, which shows how small the number of ELV exams is in the overall system.

Figure 4

Certification Forum Revised LONG.Figure 4

It is now also possible to compare average pass rates over the four data sets for each language pair individually. (See Table 3 and Figure 5.) Table 3 shows the pass rates for each language pair over time, sorted by the pass rate (low to high), while Figure 5 is sorted and grouped by language for easier comparison. The standard deviation provided shows that the pass rate in each language pair has remained relatively stable over time.1 Even those language pairs with the largest fluctuations (English>Russian and English>German) have remained within a relatively narrow range over the four data sets (15.25%-21.89% and 22.30%-28.21%, respectively).

Table 3 Cert

Figure 5

Certification Forum Revised LONG.Figure 5

The stability of these pass rates indicates that, although we can calculate an overall average pass rate for each data set, the more realistic figures are the individual average pass rates over time in each language pair. This also makes sense because, although all ATA exam passages are selected, administered, and graded according to the same criteria and all ATA graders are trained in the same methodology, each language pair must be considered a separate test. This is because the populations taking the tests are composed of completely different individuals (except for a very small number of individuals who test in two languages). In addition, the language training background and linguistic-cultural contexts for candidates in each language pair vary widely. This is particularly apparent in Figure 5, where it is possible to compare pass rates where ATA offers its certification exam in both directions.

The differences in pass rates between language directions vary from a low of 0.97% for the language pairs involving Spanish to a high of 10.46% for those involving Polish. In most but not all pairings, the exam into the foreign language has a higher pass rate. Given the relatively less extensive nature and scope of foreign-language learning in the U.S., we might speculate that for many language pairs, the population taking the test into the foreign language would include large percentages of native speakers of that language, while the population taking the test into English may include both native speakers of English who learned the language and are fluent foreign speakers of English trained in other cultures. However, given the data we have, it is impossible to arrive at any conclusions as to why pass rates differ.

We hope this detailed information on pass rates is interesting and useful to our members and potential candidates for the certification exam. The Certification Committee will continue to report the figures on a regular basis.


  1. Please note that the Polish>English and Dutch>English pass rates are based on only two data sets. This is because these language pairs entered ELV status in the 2003-2013 data set due to low demand for exams in these languages.

Geoffrey S. KobyGeoffrey S. Koby is an ATA director and the immediate past chair of ATA’s Certification Committee. He is an associate professor of German/translation studies at Kent State University. Formerly the coordinator of the university’s BS in translation program and assistant to the chair, he teaches undergraduate and graduate courses in translation theory and praytice. An ATA-certified German>English and Dutch>English translator, his professional practice focuses on business, legal, and financial translation. Contact:

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s