Phonetics, Phonology, and Morphology

Sande and Oakley publish in Language and Speech

November 30, 2022

Congrats to Hannah Sande and Madeleine Oakley, whose article on the perception and phonological patterning of implosives has just been published in Language and Speech! Click here to read it.

Beguš speaks at UCL

November 7, 2022

Gašper Beguš gave a talk at the Speech Science Forum at University College London. More info about the talk is available here.

Congrats to Dr. Eric Wilbanks!

November 1, 2022

Congratulations to Eric Wilbanks, whose doctoral dissertation, "The Integration of Social and Acoustic Cues During Speech Perception," was signed, sealed, and delivered last week!

Melguy and Johnson article in JASA

October 11, 2022

Congratulations to Yevgeniy Melguy and Keith Johnson, whose article "Perceptual adaptation to novel accent: Phonetic category expansion or category shift?" was published online this week in The Journal of the Acoustical Society of America (Vol. 152, Issue 4). It may be accessed via this link.

Berkeley linguists @ AMP 2022

October 3, 2022

The Annual Meeting on Phonology (AMP 2022), taking place at UCLA from October 21 to 23, will feature presentations by the following Berkeley linguists:

Gašper Beguš: "Deep Phonology: How to model phonology with deep learning" [abstract] Gašper Beguš and Peter Jurgec (Toronto): "Interactions of tone, stress, quantity, and vowel quality in Žiri Slovenian" [abstract] Katherine Russell: "Paraguayan Guaraní reduplication: A novel prosodic analysis" [abstract] Maksymilian Dąbkowski: "A'ingae reduplication is phonologically optimizing" [abstract] Hannah Sande: "Discontinuous harmony is movement after local phonology" (invited plenary)

Congrats, all!

Beguš speaks at Yale

September 27, 2022

On October 3, Gašper Beguš will be giving a colloquium talk at the Yale University Department of Linguistics titled "Deep Phonology: Modeling language from raw acoustic data in a fully unsupervised manner." More information is available here.

Pfiffner colloquium

September 26, 2022

The 2022-2023 colloquium series continues on Monday, October 3, with a talk by our very own Alexandra Pfiffner, taking place in Dwinelle 370 and synchronously via Zoom (passcode: lxcolloq) from 3:10-5pm. Her talk is entitled "Features, cues, and phonological contrast: A look at plosive voicing in Afrikaans," and the abstract is as follows:

Phonological voicing in obstruents is signaled by numerous acoustic cues, both spectral and temporal. Voicing contrasts have been featurally described as [±voice], [±spread glottis], fortis versus lenis, or a combination of features such as [±spread] and [±slack] vocal folds, depending on the cues utilized in a particular language. The problem that arises is that describing obstruent voicing contrasts with only cues or features, to the exclusion of the other, misses larger cross-linguistic patterns.

In this talk, I examine plosive voicing contrasts and positional neutralization in Afrikaans. Using data from perception and production experiments with native speakers, I show that acoustic cues (that are not necessarily linked to the definition of a distinctive feature) are integral to the realization of phonological contrast. To account for this data and unite the two views on describing voicing contrasts, I propose a new framework of cue-based features.

Modeling speech recognition and synthesis simultaneously: Encoding and decoding lexical and sublexical semantic information into speech with no direct access to speech data

Gašper Beguš
Alan Zhou

Human speakers encode information into raw speech which is then decoded by the listeners. This complex relationship between encoding (production) and decoding (perception) is often modeled separately. Here, we test how encoding and decoding of lexical semantic information can emerge automatically from raw speech in unsupervised generative deep convolutional networks that combine the production and perception principles of speech. We introduce, to our knowledge, the most challenging objective in unsupervised lexical learning: a network that must learn unique representations for...