Category: Talks

Phonological learning is biased by phonetic substance

Alex Martin (CLE, University of Edinburgh)

Tuesday October 22
11:30am – 12:30pm
G32, 7 George Square

Phonological rules tend to be phonetically ‘natural’: they reflect constraints on speech production and perception. Substance-based phonological theories predict that a preference for phonetically natural rules is encoded in synchronic grammars and translates into learning biases. I will present data from two studies exploring the learning of vowel harmony compared to vowel disharmony. While these two patterns are formally similar, their typological distributions and phonetic grounding are strikingly different, with vowel disharmony being fleetingly rare and phonetically unmotivated. Learners in an artificial language learning task showed better learning of the typologically frequent, phonetically grounded pattern than of the rare, unnatural one, and readily extrapolated it to cases of multiple affixation. I will argue that this is strong evidence for phonetically-biased learning (which does not come from native language experience), which in turn may explain part of the typological asymmetry between natural and unnatural rules.

The Emergence of Sound Repetition Under Pressure for Learning

Unravelling linguistic complexity

Matt Spike (CLE, University of Edinburgh)

Tuesday, October 8,
11:30am – 12:30pm
Room G32, 7 George Square

Linguistic complexity is hard to pull apart. What are the causally important variables – utterances, grammars, individuals, or populations? How can we measure complexity – inherent factors (e.g. processing cost and ecological fitness), statistical properties (e.g. counts, frequency, and information-theoretic quantities), or system-level descriptions (e.g. MDL or generative models)? What is an appropriate model for the evolution of complexity – biologically-inspired processes (e.g. selection, mutation, drift, migration, and interactions such as mutualisms), or culture-specific ones (e.g. guided variation and biased transmission)? Finally, how can we map between the candidate variables, measures, and models? In this talk, I will focus on this mapping problem by presenting some modelling work which shows i) how seemingly minor decisions can have major implications, ii) the importance of multilevel processes, and iii) how some variables (such as population size) have effects which are robust across multiple mappings.

Acquiring fusional and agglutinating morphology can be similarly difficult

Svenja Wagner (CLE, University of Edinburgh)

Tuesday, September 24, 11:00am – 12:00pm
Room G32, 7 George Square

Agglutinating morphology is commonly predicted to be easier to learn than fusional morphology due to its compositional transparency – there is a one-to-one mapping between meaning and form in agglutinating systems, but not in fusional systems (Goldschneider & DeKeyser 2001, Brown 1973, Dressler 2003, Haspelmath & Michaelis 2017, Brighton 2002). We test this hypothesis through a series of artificial language learning experiments in which participants learn nouns expressing the grammatical features of animacy and number in either a fusional or an agglutinating way. We find that when the system is small, there is no overall difference in learnability between both types of structures. However, our results provide some evidence that learners might have an a priori assumption that morphemes are transparent. In order to be able to exploit the benefits of agglutinating structures, learners need to segment words into their individual morphemes, which could mean an additional cost. However, our experiments show that segmentation costs alone cannot explain why the agglutinating system was not learned better. It is possible that the small size of the paradigm narrows the extent of the benefit for transparency. I will conclude the talk by discussing whether the longer words in the agglutinating condition could have posed a more significant difficulty in our learning paradigm.

When do Larger Populations Enhance Cumulative Cultural Evolution?

Nicolas Fay (University of Western Australia)

Tuesday, September 17
11:00am – 12:30pm
S38, 7 George Square

The extent to which large populations enhance cumulative cultural evolution (CCE) is contentious. Due to the greater access to variation and better-adapted artefacts, larger populations have the potential to enhance CCE, but this potential is not always realised. I will discuss our recent paper showing that these potential benefits are eliminated by an associated increase in working memory (Fay et al., 2019, PNAS). I will then present a recent large-scale experiment (N=407) that tested if the ability to selectively filter variants, thereby overcoming human working memory constraints, is necessary for larger populations to enhance CCE. Participants repeatedly built virtual arrowheads over 15 trials, either individually (Individual Learning Condition) or as a member of a 3-person or 6-person population (Social Learning Conditions). There were three Social Learning Conditions: View 1-Model, View All-Models: Select Order and View-All Models: Random Order. In the Social Learning conditions participants were told the score associated with each of the arrowheads produced by the other members of their group. In the View 1-Model condition they could choose one arrowhead to learn from, i.e., they could apply a selective filter. In the View All-Model conditions participants viewed all the arrowheads produced by the other members of the their group, i.e., there was no selective filter. In the Select Order condition they could choose the order they viewed the arrowheads and in the Random Order condition they viewed the arrowheads in a prescribed random order. Performance on the arrowhead task improved over trials in all conditions (i.e., CCE was observed). Social learning outperformed Individual Learning. Finally, and as predicted, CCE was boosted by population size only in the View 1-Model condition, indicating that a selective filter is critical if larger populations are to enhance CCE.

Basic word order: Improvisation + Interaction = Conventions

Marieke Schouwstra (CLE, University of Edinburgh)

Tuesday, August 27
11:00am – 12:00pm
DSB, room 1.20

There is increasing recognition that both individual and cultural processes play a role in the evolution of language, but it is not clear how these interact. Silent gesture, an experimental paradigm in which adult hearing participants describe events using only their hands, can help us discover individual word order biases that play a role when no conventional communication system is in place. When people improvise to convey information in this way, the constituent orders they produce show variability that is dependent on the semantic properties of the meaning to be conveyed. Conventional languages, by contrast, are typically more regular.

Understanding the transition from improvised variability to conventionalised regularity is a major goal of language evolution research. I will briefly report some of the results of my British Academy postdoc fellowship, showing experimental data on the emergence of word order conventions in the lab, and naturalistic data from Nicaraguan Sign Language. Together, these two sources of evidence show that word order conventions can emerge relatively quickly in interaction, but traces of meaning-dependent variability can remain, and eventually interact with conventions of the (emerging) language.

In the last part of my talk I will give a preview of new experimental work that will be part of our ESRC grant.

The lexicon is shaped for incremental processing

Adam King, University of Arizona

Monday, June 17
11:00am – 12:30pm
DSB, room 1.17

In this talk, I will present data to show the lexicon is shaped for efficient word recognition and ask how this shaping came to pass. A cornerstone in the study of language as an efficient communication system is Zipf’s law of abbreviation: probable words are shorter, less probable words are longer. On one hand, short, probable words benefit the speaker while long, less probable words benefit the listener as listeners likely need more information from the acoustics of less probable word to accurate identify it. However, not all parts of a word contribute equal disambiguating information to word identification. Spoken word processing is incremental and competitive, meaning that sounds that distinguish a particular word from many competitors are qualitatively more informative.
From a diverse set of languages, I will show that less probable words contain qualitatively higher information sounds and that these sounds are positioned where they contribute most to word identification, i.e., early. In addition, I will present simulation data that show the lexicon can develop the patterns mentioned above from simple generation-to-generation changes based on the words themselves and not from a lexicon-wide optimization to a global maximum.

Why left/right rather than uphill/downhill? An experimental approach to the evolution of spatial referencing

Jonas Nölle (CLE, University of Edinburgh)

Tuesday, June 11
11:00am – 12:00pm
DSB, room 1.17

There is considerable variation in how languages express spatial relations between objects. Strikingly, in many globalized and WEIRD societies (hence “GEIRD”), an egocentric system is preferred to express figure-ground relations (e.g., “the ball is to the left of the car”), while many non-GEIRD societies prefer perspective-independent geocentric systems that are often directly grounded in the environment (e.g., “the ball is uphill of the car”), even for expressing relations on a smaller scale such as in tabletop configurations. These strategies are associated with different underlying conceptualizations and there has been a considerable debate about their origin. More recent fieldwork lends support to the idea that spatial language could be an example of linguistic adaptation, where linguistic features are motivated by the social or physical environment. However, there are many confounding factors that are hard to disentangle, such as topography, language contact, subsistence style etc., making it difficult to uncover straightforward causal relationships from field-work data alone. For a more mechanistic understanding of how spatial referencing strategies emerge and evolve, I propose to complement this line of research with laboratory experiments that allow isolating contributing variables. I will present two virtual reality (VR) experiments where we tested participants’ preference for egocentric/geocentric strategies in large-scale VR environments such as a mountain slope or a dense forest. Experiment 1 showed that using their native language (that has a preference for egocentric solutions such as left/right), dyads solving a spatial coordination game were more likely to produce geocentric utterances (such as uphill/downhill) in a VR environment that afforded geocentric solutions strongly, suggesting that language is potentially adaptive to such salient cues, which could give rise to geocentric systems. By contrast, one of the reasons for the relative success of egocentric systems could be their flexibility. Experiment 2 thus tested whether, when switching VR environments, dyads where more likely to abandon a geocentric strategy in favour of an egocentric strategy than vice versa. However, the results did not support this prediction. This might have been due to there being no cost for establishing a new system as participants had both strategies readily available in their native language. I will discuss the design of a third experiment (currently at the piloting stage) that tries to overcome this issue by using an Experimental Semiotics approach in which strategies have to be grounded first. We predict that introducing this cost will enable us to observe differences in the flexibility of egocentric and geocentric systems in the lab.

Multi-agent language games for language learning

Angeliki Lazaridou (Deepmind)

Tuesday, May 21
11:00am – 12:30pm
Room 1.17, DSB

Distributional models and other supervised models of language focus on the structure of language and are an excellent way to learn general statistical associations between sequences of symbols. However, they do not capture the functional aspects of communication, i.e., that humans have intentions and use words to coordinate with others and make things happen in the real world. In this talk, I will present my research program on using multi-agent language games towards achieving data-efficient (functional) natural language learning.

Bio: Angeliki Lazaridou is a senior research scientist at DeepMind. She obtained her PhD from the University of Trento under the supervision of Marco Baroni, where she worked on predictive grounded language learning. Currently, she is working on interactive methods for language learning that rely on multi-agent communication, as means of minimizing the use of supervised language data.

Categorisation: what languages – and linguists – do with it

Greville G. Corbett
with Sebastian Fedden, Mike Franjieh, Alexandra Grandison & Erich Round
(Surrey Morphology Group, University of Surrey)

Tuesday, May 14
11:00am – 12:30pm
1.17 DSB

Fascinating new systems of nominal classification keep being found, but the tools for analysis have not kept pace. We therefore propose a typology of nominal classification, encompassing gender and classifier systems of categorisation. Earlier it made sense to oppose gender and classifiers (Dixon 1982), but the opposition cannot be maintained. Miraña has characteristics of gender and of classifiers (Seifart 2005); Reid’s (1997) account of Ngan’gityemerri provides further evidence against a sharp divide, since classifiers can grammaticalize into gender, through intermediate types. Relinquishing the opposition of gender vs classifiers allows a clearer picture of the possibilities. We pull apart traditional gender characteristics, and traditional classifier characteristics, and see that these characteristics combine in many ways. This motivates a canonical perspective: we define the notion of canonical gender, and use this idealization as a baseline from which to calibrate the theoretical space of nominal classification. This allows us to situate the interesting combinations we find.

Against this typological background, we may approach the origin and nature of gender. Here the possessive classifier systems of Oceanic languages can provide a unique insight. Typically, a noun can occur with different classifiers, depending on how the possessed item is used by the possessor. But we also find, in marked contrast, languages like North Ambrym (Vanuatu), where particular nouns typically occur with a given classifier (Franjieh 2016). We argue that North Ambrym’s innovative system resembles a gender system: a noun must occur with a particular classifier regardless of contextual interactions. We seek to establish empirically whether gender systems can indeed emerge from possessive classifiers in this way. We must also uncover how and why languages would relinquish a useful, meaningful classificatory system, and adopt a rigid, apparently unmotivated gender system.

We have designed and will run seven novel experiments to compare possessive classifier systems in six Oceanic languages of Vanuatu and New Caledonia. Each of these six languages has a different inventory size of classifiers — from a simple two-way distinction to a more complex inventory of twenty-three. This combination of typology with psycholinguistics promises to shed new light on the development and functioning of systems of nominal classification. We are keen to have feedback before a round of psycholinguistic experiments in the field this summer. The Oceanic data obtained so far suggest that, in this instance, we find an interesting parallelism: diachronic change is running in the direction of canonicity.