Do we say ‘t’ in witch?

I often have a discussion with people as to whether or not particular letters are pronounced in a word (e.g., the ‘t’ in witch). I have always argued that we only think we hear the ‘t’ because we know how the word is spelled.  So, for example, if I was to record you saying, “I saw a witch,” and “I didn’t know which way to go,” then isolated out the ‘witch/which’ from each sentence, you would not know whether to write ‘which’ or ‘witch’.

This view is supported in Seidenberg’s (2017, pp.26-29) book in which he makes the following points:

  • For literate individuals, our understanding of the written code influences our understanding of how speech is represented in writing. He uses the example of ‘bat’ which we think of as containing three phonemes (sounds) – /b/-/a/-/t/. Yes, when a sound wave is produced of a person saying ‘bat’ there are no clear boundaries between these three sounds and it is impossible to look at the sound wave and determine where each of these sounds begin and end. Conversely, if the sounds /b/-/a/-/t/ are spoken as individual sounds, the sounds can never be electronically melded together to sound like ‘bat’ no matter how quickly these individual sounds are combined.
  • The reason we believe we ‘hear’ three sounds in ‘bat’ is because we have learned to “treat the spoken word as if it consists of three discrete sounds because it is written with three discrete letters.” (p.27)
  • “Learning to treat spoken language as if it were composed of phonemes [sounds] is an important step in learning to read an alphabetic writing system.” (p.28)
  • Learning to read results in a feedback loop in which we start to represent speech as a series of phonemes. This in turn makes it easier to read the alphabet code, which in turn solidifies representing speech as phonemes.
  • The impact of alphabetic knowledge on spoken language is illustrated in an experiment in which participants are asked to decide if a pair of words rhyme (e.g., hit/sit) or don’t rhyme (e.g., hit/sat). There is no reading involved. The participants just listen to the words. Participants take significantly longer to decide if two words rhyme when they are spelled differently (e.g., bowl/coal) compared to when the words are spelled the same (e.g., goal/coal). In other words, the spelling of the words affects how they are ‘heard’.

So, do we actually pronounce and hear the ‘t’ in words like ‘witch’? This research would indicate that we believe we hear the ‘t’ only because our interpretation has been ‘contaminated’ by our knowledge of how the word is spelled.  This is reinforced when you look at how these types of words are spelled by children just learning to spell or by poor spellers who will most commonly leave out the ‘t’!


Seidenberg, M. (2017).  Language at the speed of sight. Basic Books: New York.

No comments yet.

Leave a Reply