The Sub-vocalization Thread

[A Leman]

I am not trying to convince anyone either way about eliminating sub vocalization since I am somewhat undecided about this for BLD myself. Sub-vocalization is the internal dialogue that is said in your mind. It’s is also a subject of debate in many different for many different applied skills such as mnemonics, speed reading, speed typing, etc.
In the Random Blindfolded Discussion thread (page 20), Chris Hardwick mentioned that it was difficult for him to listen to Hip-hop and 5BLD at the same time because it interfered with his subvocalization. He also mentioned

I know that this is a problem I have, as the memory sports group have told me flat out to stop doing this.

Also,

I really feel that vocalizing is my biggest problem with memorizing.

-Mike Hughey

This conversation(dispersed from page 20-24) and my recent post seem to be the only discussion about this on the forum. It’s seems worthy of its own thread for discussion.

Removing your dependance on subvocalization is fairly common advice from fast mnemonists and advice to stop this include humming, putting marbles in your mouth (that’s an interesting thoght!), or saying random numbers instead and again Chris’ rap practice so your mind becomes accustomed to doing the memory task without subvocalization.

There is a speed advantage to instantly recognizing an image and using any intermediate hook. From personal experience, I can recite my A images from AA-AZ in about 22 sec with subvocalization, but <12 sec without it. I can feel a speed difference for numbers, but a mind block for me comes from the fact that I am very used to a Letter Scheme on my cube and that I read the syllables while I am reading the cube.

Another possible downside is that the memo may not be as strong. NASA considers a decrease in subvocalization in reading to be related to a decrease in comprehension. On many Speed reading sites, I have seen warnings at the end that suggest to not use it for serious study like the advisory listed below:

Spoiler

Here are some times when you may want to slow your reading speed and intentionally subvocalize:
• When reading a really important document like a contract, especially if you don’t have a legal background
• When reading material that’s very challenging
• When trying to memorize something or when studying
• When reading dialogue, plays, or religious texts
When you’re in a loud distracting environment and you are having trouble concentrating on what you’re reading.

They also mention that we are slightly moving our lips (even if it is unnoticeable) while we do this. This leads me to believe that it adds some kinetic memory similar to Maskow’s hand motions. This is beside that fact that it would be very difficult to eliminate.

It should also be noted that some of our fastest BLDers (like aronpm) use subvocalization to great effect and avoid imagery as much as possible. It also seems that Nelson Dellis associates sounds to some of his cards to add feeling.

This was another interesting post/idea that I found about actually speeding up vocalization by attempting to not stress the vocal box:

Spoiler

I've always been a naturally fast reader, and when reading up on speedreading techniques, I found many things which I had done naturally already... no subvocalization being a primary one.

From what I understand, the reason that subvocalization slows stuff done is not that you're hearing the sounds, but that you're actually imaging yourself saying them, associating the words with your voice box... which makes sense giving that the word is subvocalizing and not subhearing. This means that you can read only as fast as you can speak.

anyway, I associate the words with my head, and not my voice box... I still hearthe words being said, but it's my mind that's reading them, not my mouth. Still get the benifit of hearing the sounds, but you don't have to say them out loud.

Comprehension has never been a problem.

and a suggested article by Ollie:

http://www.simplypsychology.org/working%20memory.html

Some sites:

Spoiler

http://www.speedsolving.com/forum/showthread.php?27436-Random-Blindfold-Cubing-Discussion/page20
http://mnemotechnics.org/forums/subvocalization-and-picture-training-1575.html
http://blog.mnemotechnics.org/could-a-memory-system-be-built-on-music-758.html#comment-48
http://www.stevepavlina.com/forums/personal-effectiveness/13279-eliminating-subvocalization-dangerous.html
http://www.wisegeek.com/what-is-subvocalization.htm

This is most of what I could find. What are your opinions?

 

[Noahaha]

Great post! I'm going to try removing subvocalization from different parts of my solve and look at the results. I often feel like during audio edges I am slowed down by how long it takes me to pronounce the syllables in my head, but perhaps it's necessary for audio.

 

[Ickathu]

I feel like incorporating audio into your memorization is great, but I've found that, when memorizing cards, for example, if I see the Queen of Hearts, instead of saying (mentally), "Queen of hearts - Arwen/sending/horse", then putting the the image into my route, it goes a lot faster to just see the Queen of hearts and then immediately picture Arwen/sending/a horse in my journey. However, I find it very difficult to do this, because I have to then concentrate on NOT subvocalizing, which slows down my memorization overall since I might have to review.
I think that with cubing it is even harder to stop subvocalizing, because, as you said already, we use a letter scheme on our cube. So our images for the letter pairs are usually related to those letters somehow - [for me] BE isn't sunglasses, because that wouldn't make any sense, but rather BE is a BEE, which is directly connected to the letters and the sound of the letters.

 

[Cubenovice]

I have also read some of these posts in the past and find it a very interesting matter.

But in order to have a good discussion we need a clear definition of “subvocalization”.

Wiki:
Subvocalization, or silent speech, is defined as the internal speech made when reading a word, thus allowing the reader to imagine the sound of the word as it is read.[1] This is a natural process when reading and helps to reduce cognitive load, and it helps the mind to access meanings to enable it to comprehend and remember what is read.[2] Although some people associate subvocalization with moving one's lips, the actual term refers primarily to the movement of muscles associated with speaking, not the literal moving of lips. Most subvocalization is undetectable (without the aid of machines) even by the person doing the subvocalizing.[2]

The last sentence indicates to me that we might be subvocalizing even if we think we have perfected our memo to “subhearing”.
For me personally I know that my memo (letter pairs on a journey) is more sentences / story than actually “seeing” the images.

Would love to hear from some of the very fast big / multi blindfolders how they go about this.

 

[Ickathu]

About the definition - by subvocalizing, I mean "hearing" the words in my head, either in the person's voice (if I'm familiar with how they sound, e.g., family, friends, movie characters/actors, etc), or in my own voice (for people I've never heard speak before, e.g., Cubenovice ).

[2] Although some people associate subvocalization with moving one's lips, the actual term refers primarily to the movement of muscles associated with speaking, not the literal moving of lips.

This part is interesting to me because all the time my sister actually moves her lips when she's just reading or something, but if somebody asks her what she's doing she doesn't realize that she'd been doing it.
So "normally" do the muscles move how they would normally move, just on a much smaller scale, which is why the lips and stuff don't actually change? What about when you are "hearing" someone else's voice when you read, not your own (like my example above of family, movie characters, friends, ...)? Are your muscles moving how they would move in order to imitate their voice? Because then if you were to actually try to imitate their voice it's really hard (I've tried).

 

[Schmidt]

Just a little test: See if you can read this in your own voice:

No. Try not. Do... or do not. There is no try.

 

[Ickathu]

Yes, I can. I can do it in both. I might have a hint of Yoda, but it sounds mostly like myself.

 

[Maskow]

I thought about it. I can be wrong but in this moment I think that it isn't a good idea to eliminate it. I feel that I'm faster and have better memory with VOCALIZATION.
And my PB on speaking all my words for edges is 3:19.06. 199.06/440 = 0.45s per word. Of course without list, only looking at cube. My opinion is that you simply need to train your images to be able to identify and say them faster - this kind of training helped me a lot.

Of course, eliminate sub-vocalization can be helpful, but on really really really high level.

It's only my opinion, I base it on my experiences, I can be wrong

 

[Mike Hughey]

If I recall correctly (this was a long time ago), I think when I said vocalization was a problem for me, I really meant that I would vocalize (or sub-vocalize) and not image. The danger isn't so much the sub-vocalization as it is the tendency to not actually picture the images. Sub-vocalization certainly does take time, but not very much, so I would think the memory improvement from it overwhelms the time cost of doing it in most cases.

But I think it's very important to fight to make sure you are truly imagining images (on things you're not trying to do with an audio loop) and not simply getting away with a vocalization memorization. If I had to guess, I'd guess that on a typical 5x5x5, I probably accidentally vocalize but don't properly image several locations on average per solve, which is a bad thing. That's the problem that I think I have.

I find that if I take time to psych myself up before a big solve, I can do better about having solid images on a solve (and I'm often faster). This is even better if I haven't used the room/journey recently.

 

[A Leman]

I thought about it. I can be wrong but in this moment I think that it isn't a good idea to eliminate it. I feel that I'm faster and have better memory with VOCALIZATION.
And my PB on speaking all my words for edges is 3:19.06. 199.06/440 = 0.45s per word. Of course without list, only looking at cube. My opinion is that you simply need to train your images to be able to identify and say them faster - this kind of training helped me a lot.

Of course, eliminate sub-vocalization can be helpful, but on really really really high level.

It's only my opinion, I base it on my experiences, I can be wrong

I am glad that you responded to this. I was guessing that your non-letter based system would have Forced you to not have anything to subvocalize in the first place(an unfounded random guess I had). This actually started because I was timing my list like you said except also imagining the image at the same time and I was looking for ways to become faster. I think I'll still only say the first syllable of things for 3BLD. I'll have to test for larger things.

If I recall correctly (this was a long time ago), I think when I said vocalization was a problem for me, I really meant that I would vocalize (or sub-vocalize) and not image. The danger isn't so much the sub-vocalization as it is the tendency to not actually picture the images. Sub-vocalization certainly does take time, but not very much, so I would think the memory improvement from it overwhelms the time cost of doing it in most cases.

But I think it's very important to fight to make sure you are truly imagining images (on things you're not trying to do with an audio loop) and not simply getting away with a vocalization memorization. If I had to guess, I'd guess that on a typical 5x5x5, I probably accidentally vocalize but don't properly image several locations on average per solve, which is a bad thing. That's the problem that I think I have.

I find that if I take time to psych myself up before a big solve, I can do better about having solid images on a solve (and I'm often faster). This is even better if I haven't used the room/journey recently.

That's interesting. This advice is somewhat comforting actually. I am also faster if I psych myself before a solve, but that may be because i do a run through of the future loci.

 

[bobthegiraffemonkey]

I was always a little concerned that I sub-vocalised, but if Maskow does it then it can't be that bad. I've started practicing going through my memo system and making the images/sub-vocalising with an effort to use shortened versions of the descriptions (I even made a list of the shortened words to help at first). I think it's helped me memo more securely, especially for 3BLD.

Just now, I construct 1 image/4 targets at a time, with practice to get faster I think I could construct 2 images/8 targets/1 location at a time, with help from the sub-vocalising as a sort of audio reminder of what the images should be. I'm not yet fast enough at going from targets to images to do this though. I'm not yet certain if it's a good idea and it is more difficult and potentially error prone than one image at a time, but it would potentially be faster so I'll try it when I'm capable and find out if it is good.

 

[A Leman]

I was always a little concerned that I sub-vocalised, but if Maskow does it then it can't be that bad. I've started practicing going through my memo system and making the images/sub-vocalising with an effort to use shortened versions of the descriptions (I even made a list of the shortened words to help at first). I think it's helped me memo more securely, especially for 3BLD.

Just now, I construct 1 image/4 targets at a time, with practice to get faster I think I could construct 2 images/8 targets/1 location at a time, with help from the sub-vocalising as a sort of audio reminder of what the images should be. I'm not yet fast enough at going from targets to images to do this though. I'm not yet certain if it's a good idea and it is more difficult and potentially error prone than one image at a time, but it would potentially be faster so I'll try it when I'm capable and find out if it is good.

I already do that. 1 syllable sounds of my image and then 4 targets at a time for 3BLD and 8 targets at a time for Big attempts. Since I would use 2 and 4 images per loci respectively. Noah told me that I should be going one at a time and adding to the previous image, but it feels more comfortable to do the whole loci at once because it feels like 1 piece of information instead of 2 or 4. It also has a ..- rhythm that I like. An example would be "GaNPiK"= Gandalf holding a Pikachu and getting shocked.I'll try Noah's way again, but I will only switch based upon times and accuracy.