If you want to join this month’s “Office Hours” chat just send me an email  (mgoodine@gmail.com).  It’ll happen this coming Friday. Needless to say, we’ll mostly talk about the new TOEFL this month. But there are also some other goings on in the world of testing that might be discussed, including the possible sale of the TOEFL and GRE products and the termination of the paper-delivered IELTS.

Everyone is welcome to attend the chat, as always. We generally get a mixture of teachers, test company reps, researchers… and a few students as well. I try to keep the vibe pretty casual.  Last I checked, 37 people had registered.

So, to learn how ETS converts between new and old TOEFL scores, you can consult their website.  You will see, for instance, that a score of 5.0 on the new TOEFL equals 95+ on the old TOEFL.  But that’s not exactly how the scores actually get converted on the score reports given to students. I’ve looked at dozens of TOEFL score reports since the new test launched in January.  Here’s how they convert new 1-6 scores to the old 1-120 scale:

• 6.0 = 118 of 120
• 5.5 = 110 of 120
• 5.0 = 100 of 120
• 4.5 = 90 of 120
• 4.0 = 80 of 120
• 3.5 = 65 of 120

(I haven’t seen anything lower than 3.5)

And this is it.  No other numbers are possible.  Everyone who earns 5.0 on the new TOEFL gets 100 on the old scale.  Everyone who gets 5.5 on the new TOEFL gets 110.  No one ever gets 101, 102, 103, 104, 105, etc.

And no matter how you earn a particular new scale score, it gets converted to the same old scale score.  

Just remember that due to rounding, the following is true:

• Section scores of 4.5 + 4.5 + 5.0 + 5.0 = an overall new score of 5.0 = an old score of 100
• Section scores of 4.5 + 5.0 + 5.0 + 5.0 = an overall new score of 5.0 = an old score of 100
• Section scores of 5.0 + 5.0 + 5.0 + 5.0 = an overall new score of 5.0 = an old score of 100
• Section scores of 5.5 + 5.0 + 5.0 + 5.0 = an overall new score of 5.0 = an old score of 100

In every case, the overall score of 5.0 gets converted to an old score of 110 overall, even though the English ability of the test taker is slightly different in each case.

This is only mildly interesting.  It becomes slightly more interesting when we look at how schools might use the 1-6 cut scores that have been recommended by ETS.

An institution that previously required a score of 100 points might use the aforementioned score conversion charts and set a new score requirement of 5.0.

It is likely that a greater range of students will be able to meet that new score requirement than could meet the old one. Indeed, the conversion charts suggest that anyone who previously scored between 95 and 106 points would be capable of meeting this target.

But that’s not all.  Look at the first score line I provided above (4.5, 4.5, 5.0 and 5.0). The section score conversion charts provided by ETS suggest that a student who would score only 90 points on the old test could achieve that first score line. Thus, the school which previously maintained a firm 100-point cut score is now accepting students that would have previously scored 90 points and above.

Maybe this is still just mildly interesting.

But is it meaningful? Maybe. Though a few caveats come to mind:

1. ETS hasn’t published any research in support of the score conversion tables. Accordingly, it could be wrongheaded of me to use them in such a precise way.
2. My friends at Cambridge might point out that it is a bad idea to compare scores from the new TOEFL and the old TOEFL at all since they are very different tests. For this reason, it could be totally wrong to assume that a student who scored 100 on the old test would get any particular score at all on the new test.

Anyway…

Language requirements can be tightened up by maintaining both overall and section cut scores. But for the most part, American schools don’t do this.  I googled the first ten schools that popped into my mind, and only Brigham Young University had section cut scores listed online.

In any case, given that TOEFL is still widely used, schools might be advised to carry out some longitudinal studies of students who submit TOEFL scores (and adjust reqs accordingly).

Interestingly (and not just in a mild way) this could greatly reduce the amount of re-testing done by TOEFL test takers. This is quite fun, given ongoing speculation about how financial motivations partially inspired the TOEFL revamp.

Is the new TOEFL easier than the old TOEFL?  Not really.  Some parts are easier.  Others are harder.

This article will explore both sides of the equation. First I’ll discuss a few things about the new TOEFL that could be considered easier, and then I’ll discuss some things about the new TOEFL that are actually harder.

Easy Stuff

Meeting Score Requirements is Easier

The new TOEFL has a less precise score scale than the old TOEFL.  The old TOEFL reported scores from 1-120 points.  The new TOEFL reports scores from 1 to 6 points (increasing at intervals of 0.5).  ETS has provided score conversion charts for TOEFL score users so they can adjust. Based on these charts, a school that previously required a score of 100 points might now require a score of 5.0.

Due to the lack of precision, a greater range of students will be able to meet the new score requirement.  This is because it covers a broader range of ability.  Indeed, the charts suggest that anyone who previously scored between 95 and 106 would be capable of meeting this target.

This will likely reduce the number of re-tests people do.  That’s pretty cool.

Rounding Makes Things Easier for Some Low Level Students

Rounding can now help some lower level students meet an overall score requirement.  Check this out:

• Section scores of 4.5 + 4.5 + 5.0 + 5.0 = an overall score of 5.0
• Section scores of 4.5 + 5.0 + 5.0 + 5.0 = an overall score of 5.0
• Section scores of 5.0 + 5.0 + 5.0 + 5.0 = an overall score of 5.0
• Section scores of 5.5 + 5.0 + 5.0 + 5.0 = an overall score of 5.0

So thanks to rounding, a lower level student gets a little extra credit.  Meeting the score requirement becomes a bit easier for him.  This wasn’t possible on the old test, where no rounding took place.  The last dude gets his score rounded down, but he still gets accepted.

In fact, the score conversion charts suggest that a student who scored only 90 points on the old test could achieve that first score range.  

Rounding isn’t a factor at schools which set section score requirements.  Which they really should.  I don’t want to go off on a tangent, but setting section score requirements is the only way to make these frigging tests function like they are supposed to.

There are Some Easy Questions

There are some easy questions on this test.  The “daily life” questions in the reading and listening sections are pretty straightforward.  They require less fluency and less cognitive ability than the academic reading and listening tasks that made up the entirety of the old TOEFL’s reading and listening sections.  People will take this test and then say “damn, those are some pretty easy questions.”  Indeed, they are easier than anything on the old TOEFL.

But remember that this is an adaptive test.  And test takers who end up in the “hard” module will get far fewer of these easy questions. The hard reading module doesn’t include any daily life questions, while the hard listening module contains just a handful.

Meanwhile… we don’t even know how heavily these easy questions are weighted.  They could have little impact on the overall score.

The New Speaking Section Will Be Easier for Some People

The new speaking section is a mixed bag.  All of the academic-ish content has been stripped out and replaced with two new tasks – “listen and repeat” and “take an interview.” The former task is self-explanatory.  The latter task is an interview about some everyday topic that most test takers will be familiar with.  Imagine, for example, a series of questions about your reading habits.

I’ve encountered people who speak English really well but maybe aren’t the most “cognitively impressive” folks, if you know what I mean.  Some of the kids I went to high school with (all native speakers) come to mind.

Such folks might find the old TOEFL challenging, as it required the ability to synthesize details from multiple sources along with the ability to grasp some basic academic concepts (for example, a speaking question might be about a philosophical concept like utilitarianism, or a sociological concept like “the tragedy of the commons”).  The same dudes, though, would excel at the interview task and even the listen and repeat task.

But remember that the scoring has been tightened up. The old speaking rubrics allowed for students to get a perfect score even with a highly imperfect response.  That seems to be no longer the case – the updated rubrics suggest that a perfect score now seems to require a nearly perfect response.

The New Writing Section Will Be Easier for Some People

The same is true of the writing section.  The old integrated task, which required comparing and summarizing discreet details from multiple sources has been removed.  In its place is a task which requires the test taker to write an email about an everyday concept.  For example, they might write an email to a professor asking for an extension on a project’s due date because they couldn’t get some book from the library.   The instructions are short and easy to understand.  People who speak English well but struggle with academic materials will certainly find this easier.  

But, again, the scoring rubrics have been tightened up and it looks like a perfect score requires nearly perfect grammar and language use, which was never the case before. For example, the new rubric says that a perfect response will contain “almost no lexical or grammatical errors,” while the old rubric said that a perfect response could contain “occasional language errors.”

Hard Stuff

Test Takers will Engage with Bigger Chunks of Text

The old test contained long articles, but individual reading questions were always about a single paragraph (except for the last one), and test takers were told which paragraph to check.  This means that they engaged with only about 120 words to find the answer to each question.  On the new test, they are not told which paragraph contains the answer.  That means they must engage with the whole article (about 200 words) each time.  So while the shorter articles at first glance seem to be easier, they aren’t necessarily so.

The Reading Topics are Still Hard

I know that ETS said that the test would no longer include questions about “niche topics” but I think it still does.  I took the test, and the topics I got were pretty darn niche.  Others have said the same thing.

The “Build a Sentence” Stuff is Tricky

People are finding this part of the test really hard.  I don’t know what to say about this because I thought it would be easy for a lot of people.  But it isn’t.  Like… damn.  This is actually pretty hard.  And based on what we know from the technical manual, I think it is heavily weighted.  Already, it seems from early score reports being shared online that writing is often the lowest score… mirroring, somewhat, the IELTS test.

Fin

So that’s my opinion.  Let me know what you think.  I’ll update this article as we all become more familiar with the test.

The at-home TOEFL now requires Windows 11. It can no longer be taken on computers running Windows 10. The presence of that operating system now causes the at-home TOEFL’s equipment check to fail. This change was made sometime after January 13.

In the short term, I suppose this could impact the popularity of the test, as there are still quite a few Windows 10 users. That operating system is still widely used in some key South Asia markets, where the at-home TOEFL is pretty popular. It is also widely used in South Korea, which remains a key market for the test.

The Duolingo English Test and the PTE Express test can still be taken on Windows 10. Same for the IELTS Online test.

(You can also use a Mac, of course)

A lot of people who took the TOEFL on March 1 are seeing “Due to an administrative issue, we are not able to provide a score report, please contact customer support” in their account.  I’m not sure what has gone wrong.  If I hear anything, I will post an update here.  If you get any updates, please leave a comment.

Update:  Customer support is saying that scores will be available on March 6.

Here are a few photos from outside the TOEFL test center at the Fulbright Korea building. It’s about a 500 meter walk from Gongdeok Station – near the eastern end of Gyeongui Line Forest Park. I couldn’t take any pictures inside, but you can access a cool “virtual tour” and some photos from the ETS Naver blog.

TOEFL tests are administered on three different floors. A total of 245 (!!) test takers can be accommodated at the same time. To make things easier, the seats are spread across seven separate rooms.

Despite the large size of the test center, everything went smoothly. I was happy with my keyboard, mouse and monitor. I could easily take notes, as there was sufficient elbow room between me and the test takers on either side of me. As I mentioned in an earlier post, this test center has the new noise-cancelling headphones from ETS, which do help to block noise from the rest of the room. I think they run a white-noise machine as well. That probably doesn’t make a difference, but it’s the thought that counts.

There are semi-clean bathrooms on every floor of the building and hand sanitizer just outside the testing room. There are lockers for your belongings. They can be locked using an electronic keypad, but no one bothers.

I showed up 40 minutes early and there was already quite a crowd. I was permitted to start right away (which is a standard thing in TOEFL testing).  That might be a good idea for most people, since it makes it possible to do the speaking section without a lot of background noise.

Interestingly, the administrators posted the names of everyone taking the test (and, later on, everyone taking it on Sunday). Based on that information, it looks like they administered about 250 tests over the weekend. That’s quite a lot… and it doesn’t include the two Wednesday administrations.  There are also at least 15 more test centers in the country, including a big one in Gangnam with 114 seats, and a tiny one on Jeju Island with 22 seats that hosts one administration per month. It is also worth mentioning that the at-home TOEFL is fairly popular here.

Indeed, TOEFL is a pretty big deal in Korea. It wouldn’t surprise me to learn that Fulbright hosts more than ten thousand TOEFL administrations each year. Back in 2022, Korean was the first language of 7.8% of all TOEFL test takers (the third biggest group, behind a catch-all “Indian” language category that made up 13.2% of all test takers).  Given the factors that have impacted test volumes in India since then, Korea may well be the second biggest offshore market for the TOEFL at this time (behind China). Note that Korean was the first language of only 0.8% of IELTS test takers that year.

Anyway. Below are a few of my own photos. You can see that they also run tests for Pearson, Kryterion, Inc., PSI Services LLC and Meazure Learning . EducationUSA has offices in this building as well, but the doors were locked and I couldn’t collect any pamphlets.

I took the new TOEFL yesterday in Seoul!  Below are a few notes while everything is still fresh in mind:

1. I was somewhat surprised that my fingerprints and palm prints were not taken. I guess ETS doesn’t do that, but it was done at all the PTE tests I took last year.
2. ETS has distributed custom noise-cancelling headphones to test centers around the world. They are excellent and will improve the in-person testing experience for a lot of people. 
3. I think the microphone calibration test has been adjusted, widening the acceptable range in the “audio visualizer.” That will likely reduce yelling at test centers, which is a common complaint. Regardless, I still recommend removing the visualizer entirely. I don’t think any other tests include one. 
4. I think the warning about templates given at the beginning of the test has been updated. It now says that memorized reasons and examples will receive “lower scores.” 
5. Overall, the test was broadly similar to the practice tests that ETS has published. That said, there are a few noteworthy differences. I’m not sure if I’m permitted to speak about them at length, but I suspect they will become well-known as more people take the test and more official materials are published. 
6. This is a roughly 90-minute test, including instructions. The description on the website is accurate.  I suppose most test takers will probably use about 85 minutes, as they can’t go back and use leftover time in the listening section. 
7. Speaking of timing, one insufficiently celebrated improvement is the presence of intuitive timers in the listening section.  Each listening question now has its own little timer that counts down to zero.  This compares favorably to the old TOEFL’s listening section, which had weird and unintuitive timers. 
8. The voices used throughout the test (instructions and items) were sometimes… not great. In a few cases, the AI voices used in shorter listening items had a noticeably odd cadence. On top of that, some of the conversations in that section were between speakers seemingly using microphones of wildly different quality, which was a bit jarring.  Test takers might hear a woman with a crystal clear voice (probably AI) talking to a man who sounds like he recorded his side of the conversation on his old Nokia flip phone. Thankfully, the academic lectures all sounded great.  They might have been recorded by real actors. 
9. One of the academic readings was super hard and contained a classic “stare at the screen for five minutes before picking an answer” question that the old TOEFL was known for. One of the others was about a topic I am pretty familiar with so I could breeze through it. 
10. It is worth noting how much engagement students have with the new reading passages, despite them being shorter than the ones on the old TOEFL.  In the old test, every reading question (except for one) was prefaced with something like “in paragraph 3…” and it was not necessary to go outside of that paragraph to find the answer. So while the article might have been 700 words, the test taker only engaged with ~140 words for each question (except for the last one).  That preface has been removed from the new test, so test takers engage with all ~200 words for every question. 
11. That said, with shorter articles the risk increases that pre-knowledge of the subject can be used to answer the questions without meaningful engagement with the article.  This occurred a few times on my test, but never really happened when I took the old TOEFL. Perhaps this occurs because the item writers have fewer details to work with when crafting questions.  I don’t suppose it happens frequently enough to impact score validity, but it is worth thinking about. 
12. Reduced dependence on multiple choice questions in favor of more interactive tasks that require more holistic(ish) engagement could address this. Other tests have already gone in this direction. 
13. The “build a sentence” UI is better than in the practice tests.  Clicking and dragging is much smoother.
14. The spontaneous conversation required in the “take an interview” task is pretty challenging. As expected, the questions were all about a single everyday topic… but I got one that doesn’t come up too often in my everyday life.  I actually chuckled a bit to myself when it was introduced at the beginning of the task. I fumbled around trying to quickly come up with some fake thoughts and opinions on it. This is something students will have to work on as they prepare for the test. 
15. Students should be warned about the beeps in the speaking section.  For the “listen and repeat” item they must speak after an audible beep is played.  Two minutes later they’ll start the “take an interview” item, where there is no beep.  I missed a few seconds waiting for a beep before I noticed that my timer was ticking down. 
16. I know there was much hullabaloo about the elimination of “niche topics like Greek mythology” during the TOEFL revision, but honestly it seems like the academic parts of the new test include a similar range of topics as the old one. Some were pretty darn “niche.”