skip to main content
US FlagAn official website of the United States government
dot gov icon
Official websites use .gov
A .gov website belongs to an official government organization in the United States.
https lock icon
Secure .gov websites use HTTPS
A lock ( lock ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites.


Title: Typological gaps in iambic nonfinality correlate with learning difficulty
This paper discusses gaps in stress typology that are unexpected from the perspective of a foot-based theory and shows that the patterns pose difficulties for a computationally implemented learning algorithm. The unattested patterns result from combining theoretical elements whose effects are generally well-attested, including iambic footing, nonfinality, word edge alignment and a foot binarity requirement. The patterns can be found amongst the 124 target stress systems constructed by Tesar and Smolensky (2000) as a test of their approach to hidden structure learning. A learner with a Maximum Entropy grammar that uses a form of Expectation Maximization to deal with hidden structure was found to often fail on these unattested languages.  more » « less
Award ID(s):
1650957
PAR ID:
10424605
Author(s) / Creator(s):
;
Date Published:
Journal Name:
Proceedings of the Annual Meetings on Phonology
Volume:
9
ISSN:
2377-3324
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. Ettinger, Allyson; Pavlich, Ellie; Prickett, Brandon (Ed.)
    Morphological patterns can involve simple concatenation of fixed strings (e.g., unkind, kindness) or ‘nonconcatenative’ processes such as infixation (e.g., Chamorro l-um-iʔeʔ ‘saw (actor-focus)’, Topping, 1973) and reduplication (e.g., Amele ba-bagawen ‘as he came out’, Roberts, 1987), among many others (e.g., Anderson, 1992; Inkelas, 2014). Recent work has established that deep neural networks are capable of inducing both concatenative and nonconatenative patterns (e.g., Kannand Schütze, 2017; Nelson et al., 2020). In this paper, we verify that encoder-decoder networks can learn and generalize attested types of infixation and reduplication from modest training sets. We show further that the same networks readily learn many infixation and reduplication patterns that are unattested in natural languages, raising questions about their relationship to linguistic theory and viability as models of human learning. 
    more » « less
  2. Guha, Ishani; Kidwai, Sana; Schwarz, Martha (Ed.)
    I argue that so-called ‘mora stress’ in Kohistani Shina (spoken in Northern Pakistan, realized as falling vs. rising accent) is best analyzed as a difference in the alignment of a LH* pitch accent to two types of feet (moraic vs. syllabic trochees). This paper offers a formalization of the mapping of (intonational) tones to foot structure and argues that independent evidence for a foot-based approach comes from a process of stress advancement, where stem stress predictably shifts to a following suffix when the final mora of the stem is accented; yet it remains on the stem when the stem accent is on a non-final mora. I end by briefly discussing typological and theoretical implications of our analysis, also drawing on a comparison with a similar advancement pattern in Lithuanian (known as de Saussure’s Law). 
    more » « less
  3. This paper examines the ternary vowel length contrast in East Frisian Low German, with a focus on the distinction between long and overlong vowels. While traditional accounts have analyzed overlong vowels as trimoraic or as carrying a tonal distinction, we argue that their behavior is better understood in terms of metrical structure rather than purely segmental or moraic properties. Drawing on phonetic evidence and prosodic patterns, we show that overlong vowels interact with stress assignment and foot structure, suggesting that their distinctive status arises from higher-level prosodic organization. Unlike long vowels, which occur predictably in stressed syllables of disyllabic feet, overlong vowels appear in monosyllabic feet, affecting the rhythm and duration of the prosodic unit. This metrical perspective provides a more unified explanation for the distribution and realization of overlong vowels and aligns with broader cross-linguistic patterns in vowel length contrasts. By reconsidering overlength as a prosodic rather than a strictly segmental phenomenon, this study offers new insights into the structural underpinnings of vowel length distinctions. 
    more » « less
  4. null (Ed.)
    The emergence of bipedalism had profound effects on human evolutionary history, but the evolution of locomotor patterns within the hominin clade remains poorly understood. Fossil tracks record in vivo behaviours of extinct hominins, and they offer great potential to reveal locomotor patterns at various times and places across the human fossil record. However, there is no consensus on how to interpret anatomical or biomechanical patterns from tracks due to limited knowledge of the complex foot–substrate interactions through which they are produced. Here, we implement engineering-based methods to understand human track formation with the ultimate goal of unlocking invaluable information on hominin locomotion from fossil tracks. We first developed biplanar X-ray and three-dimensional animation techniques that permit visualization of subsurface foot motion as tracks are produced, and that allow for direct comparisons of foot kinematics to final track morphology. We then applied the discrete element method to accurately simulate the process of human track formation, allowing for direct study of human track ontogeny. This window lets us observe how specific anatomical and/or kinematic variables shape human track morphology, and it offers a new avenue for robust hypothesis testing in order to infer patterns of foot anatomy and motion from fossil hominin tracks. 
    more » « less
  5. Abstract An automatic complex topology lightweight structure generation method (ACTLSGM) is presented to automatically generate 3D models of lightweight truss structures with a boundary surface of any shape. The core idea of the ACTLSGM is to use the PIMesh, a mesh generation algorithm developed by the authors, to generate node distributions inside the object representing the boundary surface of the target complex topology structures; raw lightweight truss structures are then generated based on the node distributions; the resulting lightweight truss structure is then created by adjusting the radius of the raw truss structures using an optimization algorithm based on finite element truss analysis. The finite element analysis-based optimization algorithm can ensure that the resulting structures satisfy the design requirements on stress distributions or stiffness. Three demos, including a lightweight structure for a cantilever beam, a femur bone scaffold, and a 3D shoe sole model with adaptive stiffness, can be used to adjust foot pressure distributions for patients with diabetic foot problems and are generated to demonstrate the performance of the ACTLSGM. The ACTLSGM is not limited to generating 3D models of medical devices, but can be applied in many other fields, including 3D printing infills and other fields where customized lightweight structures are required. 
    more » « less