A few stray thoughts on the upcoming revisions to the TOEFL iBT.

1. The revisions seem to be a turning of the page on the “iBT Era” (2005 to 2025) of TOEFL.  I think that everyone reading this is familiar with the characteristics that differentiate the TOEFL iBT from the original TOEFL (and from other tests of its era). Three things come to mind.  First is the inclusion of so-called “integrated” tasks which test multiple skills at the same time.  Second is the fact that the TOEFL purports to be made up of “100% academic content.”  And third is that all test items are meant to simulate, as closely as possible, the sorts of things students do in an academic environment.  These features have been mentioned again and again in marketing material used to promote the TOEFL iBT.
2. None of the above things will be true of the revised TOEFL.  That’s not a complaint.  But it is worth noting.  Some other day we can talk about how, from a business perspective, the iBT never really worked out.  It could be argued that the switch from “classic TOEFL” to “TOEFL iBT” is what gave IELTS the opening to become the juggernaut that it is today.  But again… that’s a conversation for another day.
3. Speaking of the business and history of testing, it is worth noting that the revised TOEFL is almost identical to the TOEFL Essentials Test, which launched in 2021 as a low-cost alternative to the main TOEFL iBT. But the Essentials Test seems to have been largely rejected by both test takers and score users. That ETS has plucked this particular product from benign obscurity in the back pages of the TOEFL website to serve this new function is utterly fascinating.
4. I am very interested in cost. Everything we’ve learned so far suggests that the new TOEFL will be cheaper to develop, cheaper to deliver, and cheaper to score. Some of those savings ought to be passed on to test takers.
5. I would love to see more research about score equivalencies. The test will maintain the traditional 1-120 score scale for two years.  But this is a wholly different test. Can I be sure that 73 points on the old TOEFL is totally equivalent to 73 points on the new test?
6. Some clarification on AI vs human scoring would be welcome.
7. At this point, ETS should begin the process of winding down the test centers. The idea of hauling ass to a test center for an 85-minute test is bonkers. Eliminating test centers will be a long and tedious process, but it ought to be done.  Maintain a few test centers where necessary, but do away with the rest.  If someone really needs a test center and one isn’t available… they can take a different test.
8. Eliminating the test centers would be the first step in disentangling ETS from the NEEA. For everyone’s sake, that needs to happen eventually.
9. I’m curious what the NABP will think about this test. This could represent a golden opportunity for test makers like Michigan Language Assessment who already have a focus on health care professionals.
10. Did anyone tell IELTS about the changes when they were deciding whether to spend all that time and money on a concordance study comparing IELTS to old TOEFL?

Last month Michigan Language Assessment published a concordance report comparing MET scores to IELTS scores.  We’ve been blessed with a bunch of these in recent weeks.  They aren’t emerging out of the blue, but instead are likely linked to the ongoing REOI from the Australian Home Office.  I’m not an academic, but I imagine that it is probably a good idea to keep that in the back of your mind as you read this stuff.  There are, as always, various considerations in play.

A few things are worth noting:

1. The study collected data from 1066 test takers who took both the MET and the IELTS.  Of them, 66 were removed from the study for various reasons.
2. Based on the received scores, the authors note that:  “The tests exhibit moderate to strong correlations between the corresponding subtests, ranging from r = 0.642 for Writing to r = 0.751 for Reading. The correlation coefficient for the MET and IELTS total scores, r = 0.872, is in the “high” to “very high” range.”
3. It appears that not a single participant in the study reported an IELTS writing score of 8.5 or 9.0.  This is a recurring theme in all of the concordance studies I’ve read this week.
4. We can infer some interesting things about MET’s business model from the study.  I learned that  24% of participants were based in the Philippines, 14% in Colombia and 11% in Egypt.  I’ve already written here about the popularity of the MET among health care workers from the Philippines.  Everyone’s got a niche, right?

Here’s my score report from the Michigan English Test (MET).  I’ve also included the “digital certificate” that is provided along with the score report.

A few notes:

1. Scores usually arrive in five days.
2. I got my scores after six days, as they went into administrative review. That took about 24 hours to complete.
3. Test takers can pay extra to have a printed score report mailed to their door.
4. Michigan Assessment has a unique approach to re-scoring requests and retakes for specific skills.  When the scores are reported, the test taker has the option to accept the scores, to request a rescore, or to apply for a section retake.  If the test taker accepts the scores, they are locked in and the ability to get a rescore or apply for a retake is removed. Except in the Philippines, where you can apply to retake a section at any time.
5. Scores can be sent to an unlimited number of institutions at no extra cost.  Nice.

A few days ago, I completed the Michigan English Test (MET) from Michigan Language Assessment.  I took a few notes while everything was still fresh in my mind:

1. The pre-test process was smooth. Proctoring is handled by Prometric, and they are pretty good at this sort of thing. The process began with a check-in specialist who was visible to me. She was able to answer a few questions I had about the test.
2. All of the pre-test procedures (including a room scan) took about 35 minutes.  That’s long… but normal.  I wonder if test companies track this stat. They probably should.
3. The Prometric software includes the option to cross out answers in multiple choice questions (by right clicking on them).  I love, love, love this.  All of the other proctoring companies should steal this feature. It seems like a small thing, but it helps.
4. One potential issue is worth noting.  At the end of the speaking section I clicked a button to finish the test.  But to exit the proctoring software, I was given a warning and required to click a button that said:  “yes, exit the application and abandon the exam.”  That’s not very good phrasing.  The proctor said it was okay, but I asked him to leave a note on my account.  Update:  It was fine.  No problem.
5. I like the introductory video from Michigan Assessment at the beginning that describes how the test software works.  I also like the pleasant man who later appeared to explain the speaking section.
6. The proctor told me I could request a 10 minute break at any time, during which the test would be paused.  Weird. I didn’t take her up on the offer, as she said I’d have to do a room scan before resuming the test.
7. The test is listed at 155 minutes.  That means it is longer than the category average.  That said, I suspect most test takers won’t use all of that time.  The timers are pretty generous.
8. Note-taking on paper is not allowed, but there is a “scratch pad” built into the Prometric software.  It is kinda clunky.
9. The test includes a mix of general and academic English.  It is mostly general English.  I know there is some debate about what “academic English” even means… but I think you get the point. This isn’t the TOEFL iBT.
10. The test includes some old-school grammar questions.  You don’t see many of those on tests nowadays.
11. The listening section is likely the most challenging for test-takers.  There are short questions that really get into the nuances of word choice, and longer ones that are not as nuanced, but require the test-taker to keep a lot of balls in the air.
12. The speaking section has one “describe the picture” task and four questions that could be described as “independent speaking tasks.”
13. There are no integrated tasks in any of the sections.

That’s all for now.  I hope to take two more tests before my holiday starts in mid-January.  If anyone wants me to share a few words about their test here, remember that I’ll take any English test if it is cheap or I can get a voucher.

I’ve scheduled an appointment to take the Michigan English Test (MET) tomorrow.  I’ll take the at-home version.  I’ve wanted to take this test for ages, so I’m happy to finally find the time to do so.

The MET is the successor to the old MELAB tests, which some readers might remember. Michigan Language Assessment has been doing English language testing since the beginning of recorded human history.  Today their tests are developed in partnership with Cambridge University Press & Assessment.

Update:  you can now read my full report on the test.

A few pre-test notes are worth sharing:

1. I will take the three hour, 4-skill version of the MET.  There is a 2-skill version which can only be taken at test centers.
2. The at-home version costs between $151 and $205, depending on where it is taken from.  That’s a fairly competitive price.  In comparison, the at-home TOEFL test costs between $185 and $450.
3. At-home proctoring is by Prometric.  Test center proctoring is done by Prometric and the MET network of authorized centers.
4. The test is accepted at a decent number of schools.  The website lists 966 accepting institutions.  But you know how those things are – some schools are counted more than once.  It seems to have great coverage in Atlantic Canada.
5. Healthcare licensure seems like an area of focus for the test.  That’s wonderful.
6. Payment and registration is fairly straightforward.  That said, I’m not crazy about the two-step process wherein test takers first register and pay on the MET website, and then take a activation code over to the Prometric website to make an appointment.  I get why smaller tests prefer this approach… but it isn’t great.  Test takers really won’t know what test center/at-home slots are available until after they’ve paid. Heck, they might not even know the location of the test center.
7. I contacted MET’s customer service for help and they were really professional and pleasant to deal with.  That’s rare in this industry.
8. The cancellation/reschedule fee is $100 when done 5-29 days before the test.  Free when done 30+ days before.  Could be better.
9. There is some (but not a ton) test prep on the MET website. There is a printed prep book from 2019 which I will read if I can track down a library copy.
10. The University of King’s College started accepting MET scores in October.  I used to go there to look at the microfilms. I miss those days.

Will share a few more notes when I’m done.  Let me know if there is anything I might keep an eye out for.