Thursday, January 12, 2012

Revision Week: Day 4

So, today is big brother's birthday! We're getting so old. Haha!
Happy Birthday Big Brother GREG!!!!

Over the course of the day I have knocked my bowl of cereal onto my carpeted floor (which led to the discovery that wet Wheetabix is really hard to get off of the floor), fell off of the stair leading up to my flat, ran into a doorway, and fell asleep at the computer while studying. What has all of this led me to do? Take a long break.

I think I've discovered something. There is such a thing as over studying and not giving yourself enough down time. Now, I've been giving myself down time after studying in the afternoon. But I've been up at 7am every morning since Monday and have been studying for 6 hours a day. I guess I never fully understood "over studying". Did you know that you can do it before the night right before the exam?! But today, I just studying for 4 hours and I'm going to let myself sleep for as long as I need tonight. *smile*

Right well...
6 days until the English exam
7 days until the Psychology exam

Let's go to spoken speech recognition. "There are a few models which try to explain how humans can process and understand speech. One of these theories is called the Cohort Model. A "cohort" is a group of possible words a word can be before you figure out what the word is. Let me try to explain that a little better. Within the Cohort Model model there is something called the "uniqueness point." The uniqueness point is the point where that word can only be that word. Some words have early uniqueness points and others have late uniqueness points. Words with early uniqueness point are easier to identify if speech gets distorted in any way. This uniqueness point is what the Cohort Model is based around. Basically, if you hear the sound /c/ your mind thinks of a million words that start with the sound /c/ (like cat, cabbage, carrot, crate, crab, etc.) Then you hear the /ca/ sound, so your mind eliminates the words that don't fit with that sound and you're left with those words that do. Then you finally hear /cat/ and you're left with only one word that can be. The word "cat" has a late uniqueness point. There are early and late models. The main difference between these two models is that early versions believe that we use top-down processing (we use the context to help in word recognition) and later versions believe we use bottom-up processing (it's all data driven)."

Odd question to ask...does that make any sense? I'm trying to figure out how I would explain the Cohort Model if it came up in a Short Answer Question in the exam.

No comments:

Post a Comment