0.5 The educational consultant

Rote method of learning

Obviously the Speak N Spell would teach spelling using the rote method. Our expert was emphatic that students were no longer taught using this method but used more modern techniques – learning the rules of spelling (let’s see, “i” before “e” except after “c”, etc.). After a lot of discussion back and forth on the merits of not creating such a product, that taught students how to spell improperly, Dr. Kottmeyer made an interesting comment. He suggested that there was a set of words which he called “immoral words” as they didn’t follow rules that might be applicable to such a product. He noted that these words were the ones that needed to be memorized and only rote methods would work to learn them. He also suggested that these are the words that were used on IQ tests and by knowing them a student's score would increase. This fact never showed up on the box that the Speak N Spell was sold in, nor was it used in any of the advertising literature for the product. We weren't prepared to do the required testing to prove the point. We also weren't interested in alerting the end user to the fact the product was educationally sound. We would let the parents and teachers determine the value and use that information appropriately to motivate the student to play with Speak N Spell.

Vocabulary list

In a conversation I had with Alecia Helton, she explained several of the issues that she faced in creating a proper vocabulary for the Speak N spell. In normal conversation we do not pronounce words correctly as we tend to run them together as we speak. Her example, that best shows this issue, is the way we pronounce the number "11". Here are some ways people pronounce it:

• With an initial long "e" sound: ee-leven
• With an initial short "a" sound: ah-leven
• With an initial short "e" sound: eh-leven
• With an initial short "i" sound: ih-leven

Of the four choices, the last one is the only one that is correct. To resolve the issue of correct pronunciation of words, she chose the American Heritage Dictionary of the English Language as the final authority. She than arranged with the Callier Center for communication disorders, in Dallas, Texas, for all of us who would be dealing with correct pronunciation to have our hearing tested to verify we could hear the subtle differences in speech sounds. Fortunately all of us passed. In a later chapter I'll talk about similar issues we faced on the production line and how we resolved them.

A second issue that would affect the vocabulary list was the nature of many of the words the list would include. That is many of the words would be made up of a single syllable. In many cases there wasn't enough information in the spoken word to differentiate it from similar words. Our choice of the synthesis method didn't help. For example, we chose to use an 8kHz sample rate which made the cut off frequency of the speech 4kHz. This choice, unfortunately cut off a good portion of the high frequencies needed to separate consonant sounds from each other. Some examples include (please excuse some of the words in the examples):

• Teach became Peach
• Ditch became Bitch
• Whose became Booze
• Four became Whore

There were obviously many more of which I will not include. One of Alecia's jobs was to develop the vocabulary for the Speak N Spell that did not include this issue, but still was educationally sound. I'll add a quick note that similar issues to the substitution of initial consonants as shown above also occurred with final consonants and some times with the vowel sound between the consonants.

One of the system design considerations, made to help the issue with the consonants sounds, was to not design an anti-aliasing filter on the output of the synthesizer which drove the speaker. The speaker we chose was a cheap 2 1/2 inch one. Although it had a natural roll-off at about 3,300 Hz it did allow frequencies up to about 6,000 Hz. The thought was that this would allow some of the missed higher frequencies to be heard and minimize the substitution issue. At least it was a good thought.

The final issue in created the vocabulary list was our request of having four levels of difficulty covering preschool through the third grade. It seemed that the state of the art in education was that a student learned to read in the first grade, to write in the second grade and therefore was ready to learn to spell in the third grade. The "aha" moment on the Speak N Spell was that technology could make it possible to learn to spell before learning to read or write. Fortunately, it was easy to split the vocabulary list into the four levels of difficulty we needed for the product.