SpeechRater, e-rater, AI…all these tools are fantastic for practicing for exams, but they are just tools. Nothing more!
SpeechRater, for example, listens to your speaking response, and then estimates your TOEFL score for that response, based on a number of components that it has been programmed to analyze in your speech.
➡️ Can it measure everything?
🔵 No.
➡️ Is it 100% accurate?
🔵 No, of course not. It is just a useful guide.
I’ve been working with a lot of very frustrated TOEFL students recently, who are getting high scores in SpeechRater analyses but who can’t understand why they are not getting these same scores in the TOEFL exam.
There are all sorts of reasons why this might be happening, and most of them are based on the information that SpeechRater CANNOT provide.
Here are a few:
SpeechRater doesn’t analyze topic development. It doesn’t know if you are answering the question, synthesizing and summarizing the information, and showing a logical flow of ideas. It’s not programmed to do that. That’s what the human grader(s) listen for.
SpeechRater does measure grammatical accuracy and some aspects of vocabulary use, but it’s not completely reliable. It doesn't pick up every error and it doesn't account for different grammatical styles.
I’m sure you’ve had the experience when you’ve been writing something, and autocorrect wants to change it to something else, even if your original phrase was correct. This amounts to a style choice and e-scorers are not very good at accepting different grammatical styles. Again, this is something the human grader(s) will assess.
SpeechRater does measure vocabulary components, but again they are not complete or accurate measurements.
Many students have learnt that adding ‘furthermore’, ‘moreover’, and ‘on the other hand’ are great transitions (or connecting phrases) and will add to the vocabulary diversity score. Perhaps, but they need to be used correctly.
For example, ‘on the other hand’ is used to introduce a contrasting opinion, NOT to add further information. This is an error I've seen in so many speaking responses lately.
Also these phrases are a little stilted and formal, and don't often sound natural. I never use them myself because there are hundreds of other, more suitable connecting phrases that could be used instead.
Human grader(s) are listening for normal, natural sounding speaking. If you sound like a lawyer in court using old fashioned, formal vocabulary, then that will affect your score.
If your speaking responses are full of long template phrases that don’t say anything, introductions that take up a lot of time, and unnecessary conclusions, then both your Topic Development and Language Use scores are going to be affected.
You might be speaking 160 wpm, have no pauses and clear pronunciation, but none of that matters if you’re not answering the question fully. Again, the human rater(s) assess this.
So, can you see how important the human rater(s) are in assessing your speaking?
We understand from ETS that they are responsible for 50% of your score. That’s 50% - half of your total score - that does NOT come from SpeechRater.
There is one other factor that we need to consider:
How you perform in the exam depends on how you cope with your own nerves, and the exam conditions. SpeechRater can never predict that.
This is a long post, and if you’ve made it this far, well done 😊 You’ve shown you have the motivation and persistence to be successful 😊
So, what can you do? It’s not all bad news.
You should focus on the Topic Development and Language Use parts of your response much more than focusing on delivery. Building skills in these areas will give you the foundation to create really good TOEFL speaking responses.
/version-3023/images/image-g37bb.png)