1. SPEECH PERCEPTION DAY 18 – OCT 9, 2013 Brain & Language LING 4110-4890-5110-7960 NSCI 4110-4891-6110 Harry Howard Tulane University

2. Course organization The syllabus, these slides and my recordings are available at http://www.tulane.edu/~howard/LING4110/.http://www.tulane.edu/~howard/LING4110/ If you want to learn more about EEG and neurolinguistics, you are welcome to participate in my lab. This is also a good way to get started on an honor's thesis. The grades are posted to Blackboard. 10/07/13Brain & Language, Harry Howard, Tulane University 2

3. REVIEW 10/07/13Brain & Language, Harry Howard, Tulane University 3

4. Linguistic model, Fig. 2.1 p. 37 10/07/13Brain & Language, Harry Howard, Tulane University 4 Discourse model Syntax Sentence prosody Morphology Word prosody Segmental phonology perception Segmental phonology perception Acoustic phonetics Feature extraction Segmental phonology production Segmental phonology production Articulatory phonetics Speech motor control INPUT Sentence level Word level

5. SPEECH PERCEPTION Ingram §6 10/07/13Brain & Language, Harry Howard, Tulane University 5

6. A baby’s linguistic experience Are babies sensitive to phonological distinctions in their mothers’ speech? Yes. How can this be? They must be able to hear in the womb. In their first few months, babies prefer … their caretaker’s voice; phonological distinctions in their linguistic environment; speech rhythms in their linguistic environment. Summary: before children utter their first words, their perceptual system is being attuned to their linguistic environment. 10/07/13Brain & Language, Harry Howard, Tulane University 6

7. Head turning Baby attends to toy and ignores repetitions of a stimulus until it changes: a – a – a – i 10/07/13Brain & Language, Harry Howard, Tulane University 7

8. Perceptual magnet effects Prototype /i/ (P) and non-prototype (NP) vowels Result: reduced discrimination near prototype 10/07/13Brain & Language, Harry Howard, Tulane University 8

9. Discussion of PME Animals can’t do this. Hooray! It depends on exposure to a language – it emerges only after 6 months. 10/07/13Brain & Language, Harry Howard, Tulane University 9

10. THE SPEECH RECOGNITION LEXICON Ingram §7 10/07/13Brain & Language, Harry Howard, Tulane University 10

11. Linguistic model, Fig. 2.1 p. 37 10/07/13Brain & Language, Harry Howard, Tulane University 11 Discourse model Syntax Sentence prosody Morphology Word prosody Segmental phonology perception Segmental phonology perception Acoustic phonetics Feature extraction Segmental phonology production Segmental phonology production Articulatory phonetics Speech motor control INPUT Sentence level Word level

12. Storing on a hard disk How does a computer store files on its hard drive? By writing them in sequence or where ever there is space. 10/07/13Brain & Language, Harry Howard, Tulane University 12

13. Retrieving from a hard disk How does a computer find files on its hard drive (say, when you search for one by its name)? It searches for it in sequence or randomly. How long does it take? 10/07/13Brain & Language, Harry Howard, Tulane University 13

14. How would this work for lexical retrieval? Ingram’s example The phoneme detector department detects /k/. A comparator starts looking for all the files that begin with /k/, perhaps ordered in terms of frequency. The phoneme detector department detects /æ/. The comparator rejects the files that don’t begin with /kæ/ and starts searching the remaining files, perhaps ordered in terms of frequency. “It is an open bet whether the word cat would be retrieved before or after the detection of /t/.” (p. 143) Problems Other factors influence speed of retrieval, such as whether the target word has been seen recently. Adding such factors to a serial search model tends to make it slow down! 10/07/13Brain & Language, Harry Howard, Tulane University 14

15. An alternative: the TRACE II model 10/07/13Brain & Language, Harry Howard, Tulane University 15

16. Observations TRACE implements parallel computation, rather than serial or sequential computation. It is both bottom-up (driven by data) and top-down (driven by expectations). Bottom up The successive winnowing of a set of cohorts is modeled by decaying activation of competitors as more information is gathered. Top down Word frequency is modeled by lowering the threshold of activation of more frequent word units, so they need less activation. The phoneme restoration effect is modeled by the word units supplying the missing activation of a phoneme unit. [kæØ] can be heard as ‘cat’. The Ganong effect is modeled in the same way. [kæ ] can be heard as ‘Cass’ or ‘cash’ in the proper context. 10/07/13Brain & Language, Harry Howard, Tulane University 16

17. Simple recurrent networks Read what Ingram says to get the general idea of what it is supposed to do. 10/07/13Brain & Language, Harry Howard, Tulane University 17

18. Modeling variability We will go over it on Monday. 10/07/13Brain & Language, Harry Howard, Tulane University 18

19. NEXT TIME Finish Ingram §7. ☞ Go over questions at end of chapter. 10/07/13Brain & Language, Harry Howard, Tulane University 19