Our next meeting will be on November 20, 2019 at 4PM, featuring a talk by Timo Roettger, a postdoctoral fellow in the Linguistics department. Abstract & title below. As usual, a happy hour will follow at Stacked & Folded Evaston (824 Noyes).
Preregistration – What is it? Why should we do it? And what’s in it for us?
The current publication system incentivizes neither publishing null results nor direct replication attempts. This state of affairs biases the scientific record toward novel findings that appear to support presented hypotheses (referred to as “publication bias”). Moreover, flexibility in data collection, measurement, and analysis (referred to as “researcher degrees of freedom”) can lead to overconfident beliefs in the robustness of a statistical relationship. This flexibility is particularly pronounced in speech sciences, potentially increasing the rate of false discoveries in our own publication record.
One strategy to systematically decrease publication bias and the harmful impact of researcher degrees of freedom is preregistration. A preregistration is a time-stamped document that specifies how data is to be collected, measured, and analyzed prior to data collection. Preregistration is a powerful tool to reduce bias and to facilitate transparency in decision making. This talk introduces the concept of preregistration and discusses its benefits and potential disadvantages for both our scientific field and individual researchers.
November 20, 2019, 4PM to 5PM
Cresap 101, Cresap Laboratory 2029 Sheridan Rd
Our next speaker will be by Kasia Hitczenko, a postdoctoral fellow in the department of Linguistics at Northwestern.
How context can help in learning sounds from naturalistic speech
Infants learn the sound categories of their language and adults successfully process the sounds they hear, even though sound categories often overlap in their acoustics. Most researchers agree that listeners use context (e.g. who the speaker was, what the neighboring sounds were, etc.) to help disambiguate overlapping categories, and have put forth a number of theories about how contextual information could be used. However, for the most part these theories have been developed by studying simplified speech (synthetic or well-enunciated, controlled lab speech), so it is unclear to what extent these ideas extend to naturalistic speech. Here, I ask how contextual information could be helpful for processing and learning from naturalistic speech of the type that listeners actually hear. I implement two main ways of using context and test their efficacy in separating overlapping categories on naturalistic speech, focusing on the test case of Japanese vowel length. Our results show that well-established results from lab speech do not necessarily generalize to naturalistic speech, and lead to a new proposal for how infants could learn the sounds of their language. Overall, our results reveal the importance of studying infants’ naturalistic input and highlight the value of tools that allow us to do so.
October 30th, 2019, 4PM to 5PM
Cresap 101, Cresap Laboratory 2029 Sheridan Rd
Next meeting (5/1), Oriana Kilbourn-Ceron (LING) will be talking about “Phonological variability at word boundaries: the effect of speech production planning”.
“Connected speech processes have played a major role in shaping theories about phonological organization, and how phonology interacts with other components of the grammar. Presenting evidence from English /t/-realizations and French liaison, we argue that the effect of lexical frequency on variability can be understood as a consequence of the narrow window of phonological encoding during speech production planning. By connecting the study of phonological alternations with the study of factors influencing speech production planning, we can derive novel predictions about patterns of variability in external sandhi, and better understand the data that drive the development of phonological theories.
Hope the talk at Rutgers went well!”
Our meeting will take place at the regular time and place on Wednesday 05/01 from 4-5pm in Cresap 101. Afterwards, we will have our happy hour at the World of Beer in Evanston.
Next meeting (4/17), Olga Dmitrieva (Purdue University) will be talking about “Phonetic first language attrition – where next?”. Our meeting will take place at the regular time and place on Wednesday from 4-5pm in Cresap 101.
Abstract: “In this talk I will review some of my recent work investigating the effects of second language (L2) learning on the production and perception of first language (L1) sounds. One project addressed the effects of learning French or Russian as a second language on the production of word-initial and word-final voicing distinctions in English-speaking learners. Another study investigated how the perception of word-final voicing in Russian was affected by second language experience in English.
I will focus in particular on the insights these findings may offer about the circumstances under which L2 effects on L1 are likely to occur and factors which may be at play in intensifying or mitigating such effects. I would like to finish by highlighting the questions that still remain to be addressed and initiating a discussion about the productive directions in which research in this area should develop in the near future.”
Next meeting (3/13), Jacob Phillips (University of Chicago) will be talking about “Asymmetrical sound change actuation and propagation: Production and perception of /s/-retraction in American English”. Our meeting will take place at the regular time and place on Wednesday from 4-5pm in Cresap 101.
Models of sound change actuation propose that change emerges from both speech production and perception, with potential sources including the persistence and accumulation of short term shifts and/or the failure to compensate for extreme coarticulation. In this talk, I question how context-dependent variation can approach the threshold of a sound change in one phonological environment, but not other similar environments. I examine the production and perception of /s/-retraction in American English, a sound change in progress in which /s/ approaches /sh/ in /str/ clusters (such that street may sound like shtreet), but rarely /spr/ or /skr/ clusters (scream rarely sounds like shcream). I ask whether evidence from imitation, phoneme categorization and eye tracking tasks can shed light on this asymmetrical distribution.
Next meeting (3/6), Yossi Keshet will be talking about “Hide and Speak: Deep Neural Networks for Speech Steganography”. Our meeting will take place at the regular time and place on Wednesday from 4-5pm in Cresap 101.
Abstract: “Steganography (“steganos” – concealed or covered plus “graphein” – writing) is the science of concealing messages inside other messages. It is generally used to convey concealed “secret” messages to recipients who are aware of its presence while keeping even their existence hidden from other unaware parties who only see the “public” or “carrier” message.
In the talk I will propose the use of deep neural networks as learnable steganographic functions, that learn to exploit redundancies in audio data to conceal messages optimally. I will demonstrate quantitatively that the proposed method can both effectively hide secret messages into a carrier and recover them from the carrier. Qualitative experiments suggest that modifications to the carrier are unnoticeable by human listeners and that the decoded messages are highly intelligible as well.”
Next meeting (2/20), Elizabeth Pillion will be talking about “Clicks in American English: Understudied Aspects of Sound Systems”. Our meeting will take place at the regular time and place on Wednesday from 4-5pm in Cresap 101.
Clicks in English are known to occur appear in a variety of contexts, such as transitional periods of talk, during word searches and as indicators of speaker affect. Less well established are the phonetic properties of these clicks, phonetic variation with respect to the click’s discourse role, and phonetic variation between speakers. This study contributes to understandings of click acoustics by examining clicks used within the Buckeye Corpus (Pitt et al. 2007) of American English.
Percussive clicks varied significantly in intensity from those with a discourse role such as conveying affect or turn-management. Speakers also varied in the extent to which clicks of all discourse types were employed: male speakers clicked percussively at a significantly higher rate than female speakers, whereas female speakers were more likely to use turn-management and affect-conveying clicks. Manual auditory categorizations of articulatory properties are not strongly supported by acoustic evidence. This research helps illuminate an understudied aspect of sound systems, and gives insight into the extent of intra- and inter-speaker variation in paraphonemic sounds.
Our next speakers will be Annette D’Onofrio and Jaime Benheim (LING), presenting on: “Contextualizing Reversal: Sociohistorical Dynamics and the Northern Cities Shift in a Chicago Neighborhood”. Our meeting will take place at the regular time and place on Wednesday 01/23 from 4-5pm in Cresap 101.
Abstract: Despite Chicago’s status as the largest urban center in the Inland North, recent dynamics of the region’s Northern Cities Vowel Shift have remained relatively understudied in this city. This study examines the vowel systems of 40 speakers from one Chicago neighborhood area. Results reveal a reversal of the NCS in apparent time, paralleling findings in other NCS locales. However, while results indicate a robust community-wide trend, not all speakers engage with the shift in the same way. Through a qualitative examination of individual speakers’ vowel spaces, we find that shifting demographics and ideological concerns across the neighborhood’s history help explain which community members are likely to use NCS-shifted vowels at different points in apparent time. More broadly, we suggest that reversals of local sound changes are not always indicators of increased supralocal orientation or contact, but instead can be driven by shifts in what it means to index local identity.
Our next speaker will be Ryan Cotterell (University of Cambridge), presenting on: “Probabilistic Typology: Deep Generative Models of Vowel Inventories”. Our meeting will take place at the regular time and place on Wednesday 01/09 from 4-5pm in Cresap 101.
Linguistic typology studies the range of structures present in human language. The main goal of the field is to discover which sets of possible phenomena are universal, and which are merely frequent. For example, all languages have vowels, while most—but not all—languages have an [u] sound. In this paper we present the first probabilistic treatment of a basic question in phonological typology: What makes a natural vowel inventory? We introduce a series of deep stochastic point processes, and contrast them with previous computational, simulation-based approaches. We provide a comprehensive suite of experiments on over 200 distinct languages.
Our next speaker will be Jennifer Cole (LING), presenting on: “Conventionalization in the prosodic encoding of information structure: An information-theoretic approach”.
Early accounts of phrasal prominence [Bolinger 1972; Chafe 1974; Chomsky & Halle 1968; Gussenhoven 1983; Ladd 1980] point to informational criteria (related to focus and discourse-givenness) and structural criteria (related to position in the prosodic phrase) as determining which word(s) within a prosodic phrase are assigned phrasal prominence. Yet empirical evidence from recent studies calls for reconsideration of analyses that directly and deterministically link information structure (IS) meaning with phrasal prominence and/or pitch accent. The argument against traditional accounts of the prosodic encoding of IS meaning rests on findings from experiments on English, French, German, Hindi, Italian, Russian, Spanish and Berber that collectively show: (1) the relationship between IS categories and phonological pitch accents is not deterministic—e.g., in English and German, accent may occur on given as well as new words, and many accent types (e.g, L*, H*, L+H*) are attested for each information structure condition. (2) There is systematic acoustic prosodic enhancement associated with gradations in informativity across IS categories—e.g., focused and/or new words are acoustically enhanced relative to given words—in languages with typologically diverse prosodic systems. (3) The gradient effect of informativity on acoustic prosodic enhancement can be greater for phrase-final (nuclear) prominence compared to non-final prominences, an asymmetry that is reflected in an overall structural bias that privileges the phrase-final position in perceptual processing of phrasal prominence.
These findings are hard to reconcile in traditional accounts that directly and deterministically link prosodic phonological structures with IS meaning. I argue for an alternative approach, adopting an information-theoretic framework where predictability and conventionalization interact to shape systematic variation in prosodic expression and its association with IS meaning. The predictability of a word and its referent influences linguistic expression at the lexical, syntactic, phonological and phonetic levels [e.g., Aylett & Turk 2004; Levy & Jaeger 2007], to varying degrees across languages. Traditional IS distinctions, re-cast in in terms of predictability, influence a speaker’s choice of prosodic expression at the phonological and phonetic levels, and variation in the prosodic expression of predictability is potentially offset by expression via lexical or syntactic choices. The conventionalized pairing of acoustic prosodic marking and IS meaning varies across IS categories and across languages. For instance, in English, a phonological pitch accent may come to function as a pragmatic morpheme due to the highly conventionalized pairing of acoustic prosodic enhancement (e.g., a sharply rising pitch excursion) with a salient information status distinction (e.g., corrective focus), while less conventionalized associations result in probabilistic and phonetically gradient patterns in speech production. In other words, the prosodic encoding of information structure is phonological only in the most conventionalized cases. This information-theoretic account offers insight into observed structural and pragmatic biases in the perceptual processing of prosody, and reconciles the apparent conflict between recent experimental findings in diverse languages and traditional views of the prosody-IS relationship.