One major topic of my research is reduced speech, which is part of how all of us communicate every day. Reduced speech refers to sounds being deleted or produced less clearly than in careful speech (e.g. voiced stops and even phonemically voiceless stops realized as approximants), and to speech with syllables or words deleted. This kind of speech is used most strongly in casual conversation, rather than when reading something out loud or talking formally. However, it can be used even in formal speech: I've heard a newscaster speaking formally on the news pronounce senator with hardly any trace of the /t/ several times in a row. In some cases words are deleted, or so many words and syllables are merged into each other that one can't tell which words are deleted and which are left. What is most interesting about this is that listeners can usually still understand the speech just fine, as long as they are native speakers of the language and they hear the reduced speech in context. In context, one doesn't even notice that there's anything unusual about the reduced part. Reduced speech has been found in every language where someone has looked for it, including English, Dutch, German, Japanese, Czech, Korean, Spanish, French, Finnish, Mandarin, Russian, Bulgarian, Italian, Greek, Polish, and Swedish. Here are some examples. The answers to what the short sound files say are at the bottom of the page, so that you can listen before you see the answer.
American English:
Dutch:
Japanese:
More English:
Answers (the portion you hear when you click on "out of context" above is bolded). Narrow transcription of bolded portion is given. Segments transcribed in parentheses have only slight acoustic or auditory evidence for their presence.
a. 'I think I can get down to a 445 this summer' (regarding someone's time in a swimming competition)
[hn̩ŋ̩]
b. 'I was (just) like, yeah, right' (some listeners hear 'just,' some do not) (continuation of utterance a)
[əʒ lʌ̰ɨ̰k]
c. 'but I was (just) like, why wouldn't you just go home?' (some listeners hear 'just')
[bɨ̰ ʌ ʒz lʌɨ̰k]
d. '...but I was like, oh, no, so he has to cut someone, 'cuz he already told Steve he was in the wedding.'
[phʌ ɨ z̥ lʌɪk]
e. 'converted into, I guess like a little computer room'
[(l)ɨʊɨ]
f. 'She wants to be a police officer, I think.'
[ʃɨ z ɪ]
g. 'And like, one of my classes is, like, a writing-intensive class, and I didn't really know that, but I need to take it to graduate.
[ʌ̃ɛ̃:ɚ]
h. '(I'm) so jealous, my Friday night was so stupid'
[fɹɛ̃:]
i. 'dacht ik van misschien dat ik hem wel daar gewoon kan kopen' "I thought that maybe I can just buy it there after all"
(d)ɪk(ə)m
j. 'a, soo iu hitotachi yoo-no supeesu-ga aru-no-ka?' あ、そういう人たち用のスペースがあるのか "Oh, is there space for people like that?"
[soiçəʃjo]
k. 'And so they were getting back to Desert right when Eric and I got there'
[gɪɾ̞ɪ̃ŋ] (flap realized as approximant)
l. 'Cuz he already told Steve he was in the wedding.'
[wɛ(ɾ̞)ɪ̃ŋ] (flap nearly deleted)
m. 'hanging out like every day' (this week so far)
[ɦẽæ̃ʊ (l)ʌɪ j̰ɛɹɨ dɛɪ] (transcription for context included)
n. 'Do you have time to talk to me for a little while?'
[tjʏ tɛ̃m] (both [t]'s slightly aspirated)
Examples for "he's" vs. "he was," "we're" vs. "we were" project
Isolation is the portion of the signal corresponding to the target "he's," "we were" etc. Limited context extends to the outer edge of the neighboring vowels. Full context provides the utterance context. Beep provides the version with full utterance context, but with the target itself replaced by a beep of standardized duration. Orthographic transcriptions of full context are provided below so that you can listen before you see the answer, with the target portion bolded.
o. Isolation Limited Full Beep
p. Isolation Limited Full Beep
q. Isolation Limited Full Beep
Answers:
o. "Er, uh, Tuesday night, uh when we were chillin' in the spa, but"
p. "When we were outside the bookstore"
q. "She's got, Chem with me, we're taking all that together"