Alexander Martin

Laboratoire de linguistique formelle  |  Université de Paris

I am a postdoctoral researcher at the LLF within the Université de Paris, working with Heather Burnett. I am generally interested in understanding how constraints on language change can explain typology. I use mostly experimental techniques to test how cognitive (perceptual and grammatical) and social factors influence the shape of linguistic systems.

I was previously a postdoc with Jenny Culbertson at the Centre for Language Evolution at the University of Edinburgh. Before that I did my PhD under the supervision of Sharon Peperkamp at the Laboratoire de Sciences Cognitives et Psycholinguistique within the École Normale Supérieure.

You can download my full CV here.


Journal articles

Martin, A. & White, J. (2021). Vowel harmony and disharmony are not equivalent in learning. Linguistic Inquiry 52(1), 227–239.

General vowel harmony and disharmony rules have comparable formal complexity but differ dramatically in typological frequency and phonetic motivation. Previous studies found no difference in learning between vowel harmony and disharmony; this putative equivalence has been used to discount the view that learners are influenced by substantive learning biases. In the current study, we use a more nuanced test to show that there is a clear difference in learning between vowel harmony and disharmony: learners readily infer a vowel harmony pattern, but not a disharmony pattern. The findings suggest that vowel disharmony is in fact strongly disfavored during learning.

Martin, A., Holtz, A., Abels, K., Adger, D., & Culbertson, J. (2020). Experimental evidence for the influence of structure and meaning on linear order in the noun phrase. Glossa: a journal of general linguistics, 5(1), 97.

Recent work has used artificial language experiments to argue that hierarchical representations drive learners’ expectations about word order in complex noun phrases like these two green cars (Culbertson & Adger 2014; Martin, Ratitamkul, et al. 2019). When trained on a novel language in which individual modifiers come after the Noun, English speakers overwhelmingly assume that multiple nominal modifiers should be ordered such that Adjectives come closest to the Noun, then Numerals, then Demonstratives (i.e., N-Adj-Num-Dem or some subset thereof). This order transparently reflects a constituent structure in which Adjectives combine with Nouns to the exclusion of Numerals and Demonstratives, and Numerals combine with Noun+Adjective units to the exclusion of Demonstratives. This structure has been also claimed to derive frequency asymmetries in complex noun phrase order across languages (e.g., Cinque 2005). However, we show that features of the methodology used in these experiments potentially encourage participants to use a particular metalinguistic strategy that could yield this outcome without implicating constituency structure. Here, we use a more naturalistic artificial language learning task to investigate whether the preference for hierarchy-respecting orders is still found when participants do not use this strategy. We find that the preference still holds, and, moreover, as Culbertson & Adger (2014) speculate, that its strength reflects structural distance between modifiers. It is strongest when ordering Adjectives relative to Demonstratives, and weaker when ordering Numerals relative to Adjectives or Demonstratives relative to Numerals. Our results provide the strongest evidence yet for the psychological influence of hierarchical structure on word order preferences during learning.

Martin, A. & Culbertson, J. (2020). Revisiting the suffixing preference: Native language affixation patterns influence perception of sequences. Psychological Science 31(9), 1107–1116.

Similarities among the world’s languages may be driven by universal features of human cognition or perception. For example, many languages form complex words by adding suffixes to the ends of simpler words, but adding prefixes is much less common: why might this be? Previous research suggests this is due to a domain-general perceptual bias: sequences differing at their ends are perceived as more similar to each other than sequences differing at their beginnings. However, as is typical in psycholinguistic research, the evidence comes exclusively from one population—English speakers—who have extensive experience with suffixing. Here we provide a much stronger test of this claim, by investigating perceptual similarity judgments in speakers of Kîîtharaka, a heavily-prefixing Bantu language spoken in rural Kenya. We find that Kîîtharaka speakers (N=72) show the opposite judgments to English speakers (N=51), calling into question whether a universal bias in human perception can explain the suffixing preference in the world’s languages.

Martin, A. & Peperkamp, S. (2020). Phonetically natural rules benefit from a learning bias: a re-examination of vowel harmony and disharmony. Phonology, 37(1), 65–90.

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. Some previous work has shown evidence for such biases, but methodological concerns with those studies mean that the question warrants further investigation. We revisit this issue by focussing on the learning of palatal vowel harmony (phonetically natural) compared to disharmony (phonetically unnatural). In addition, we investigate the role of memory consolidation during sleep on rule learning. We use an artificial language learning paradigm with two test phases separated by twelve hours. We observe a robust effect of phonetic naturalness: vowel harmony is learned better than vowel disharmony. For both rules, performance remains stable after 12 hours, regardless of the presence vs absence of sleep.

Martin, A., Ratitamkul, T., Abels, K., Adger, D., & Culbertson, J. (2019). Cross-linguistic evidence for cognitive universals in the noun phrase. Linguistics Vanguard, 5(1).

Noun phrase word order varies cross-linguistically, however, two distributional asymmetries have attracted substantial attention (i.a., Greenberg 1963, Cinque 2005). First, the most common orders place adjectives closest to the noun, then numerals, then demonstratives (e.g., N-Adj-Num-Dem). Second, exceptions to this are restricted to post-nominal position (e.g., N-Dem-Num-Adj, but not Adj-Num-Dem-N). These observations have been argued to reflect constraints on cognition. Here we report two experiments, following work by Culbertson & Adger (2014), providing additional support for this claim. We taught English- and Thai-speaking participants artificial languages in which the position of modifiers relative to the noun differed from their native order (post-nominal position in English, pre-nominal in Thai). We trained participants on single-modifier phrases, and asked them to extrapolate to multiple modifier phrases. We found that both populations infer relative orders of modifiers that conform to the tendency for closest proximity of adjectives, then numerals, then demonstratives. Further, we show that Thai participants, learning pre-nominal modifiers, exhibit a stronger such preference. These results track the typology closely and are consistent with the claim that noun phrase word order reflects properties of human cognition. We discuss future research needed to rule out alternative explanations for our findings, including prior language experience.

Martin, A., & Peperkamp, S. (2017). Assessing the distinctiveness of phonological features in word recognition: prelexical and lexical influences. Journal of Phonetics, 62, 1–11.

Phonological features have been shown to differ from one another in their perceptual weight during word recognition. Here, we examine two possible sources of these asymmetries: bottom-up acoustic perception (some featural contrasts are acoustically more different than others), and top-down lexical knowledge (some contrasts are used more to distinguish words in the lexicon). We focus on French nouns, in which voicing mispronunciations are perceived as closer to canonical pronunciations than both place and manner mispronunciations, indicating that voicing is less important than place and manner for distinguishing words from one another. We find that this result can be accounted for by coalescing the two sources of bias. First, using a prelexical discrimination paradigm, we show that manner contrasts have the highest baseline perceptual salience, while there is no difference between place and voicing. Second, using a novel method to compute the functional load of phonological features, we show that the place feature is most often recruited to distinguish nouns in the French lexicon, while voicing and manner are exploited equally often.

Martin, A., & Peperkamp, S. (2015). Asymmetries in the exploitation of phonetic features for word recognition. The Journal of the Acoustical Society of America, 137(4), EL303–EL317.

French listeners’ reliance on voicing, manner, and place was tested in a mispronunciation detection task. Mispronounced words were more likely to be recognized when the mispronunciation concerned voicing rather than manner or place. This indicates that listeners rely less on the former than on the latter for the purposes of word recognition. Further, the role of visual cues to phonetic features was explored by the task being conducted in both an audio-only and an audiovisual version, but no effect of modality was found. Discussion focuses on cross-linguistic comparisons and lexical factors that might influence the weight of individual features.

Fort, M., Martin, A., & Peperkamp, S. (2015). Consonants are more important than vowels in the bouba-kiki effect. Language and Speech, 58(2), 247–266.

Adult listeners systematically associate certain speech sounds with round or spiky shapes, a sound-symbolic phenomenon known as the "bouba-kiki effect". In this study, we investigate the respective influences of consonants and vowels in this phenomenon. French participants were asked to match auditorily presented pseudowords with one of two visually presented shapes, one round and one spiky. The pseudowords were created by crossing either two consonant pairs with a wide range of vowels (Experiment 1 and 2) or two vowel pairs with a wide range of consonants (Experiment 3). Analyses showed that consonants have a greater influence than vowels in the bouba-kiki effect. This asymmetry cannot be due to an onset bias, as a strong consonantal influence is found both with CVCV (Experiment 1) and VCV (Experiment 2) stimuli. We discuss these results in terms of the differential role of consonants and vowels in speech perception.

Conference proceedings

Guevara-Rukoz, A., Martin, A., Yamauchi, Y., & Minematsu, N. (2019). Prototyping a web-based phonetic training game to improve /r/-/l/ identification by Japanese learners of English. In: Proc. SLaTE 2019: 8th ISCA Workshop on Speech and Language Technology in Education (pp. 20–24).

Even after years of study, language learners may have difficulty perceiving L2 sounds. For instance, Japanese listeners show difficulty differentiating American English /r/ and /l/. Previous research has shown that phonetic training may improve learners' perception of the contrast. While this training paradigm appears as a promising tool for language learning, its transition from the laboratory to the classroom needs to be facilitated. Not only does phonetic training require recording and/or manually editing many training exemplars, training sessions are also often long and repetitive. Given these obstacles, the long-term goal is to make phonetic training more applicable to real-life learning. In this preliminary study, we prototype a self-paced, web-based phonetic training program, featuring both identification and discrimination tasks as playable mini-games. Participants are trained using nonword minimal pairs (e.g., /lapu/-/rapu/), presented in isolation in clean speech. Their ability to identify the target phonemes is assessed before and after training, with stimuli also presented in noise and/or in sentences, to test perceptual robustness. We assess the effectiveness of the phonetic training game in its current form and discuss future improvements, notably in the context of using speech engineering to automate and augment High Variability Phonetic Training (HVPT) programs.

Martin, A., Abels, K., Adger, D., & Culbertson, J. (2019). Do learners’ word order preferences reflect hierarchical language structure?. In: A. C. Goel, C. M. Seifart, & C. Freksa (Eds.) Proceedings of the 41st Annual Meeting of the Cognitive Science Society (pp. 2303–2309). Montreal, QB: Cognitive Science Society.

Previous research has argued that learners infer word order patterns when learning a new language based on knowledge about underlying structure, rather than linear order (Culbertson & Adger, 2014). Specifically, learners prefer typologically common noun phrase word order patterns that transparently reflect how elements like nouns, adjectives, numerals, and demonstratives combine hierarchically. We test whether this result still holds after removing a potentially confounding strategy present in the original study design. We find that when learn-ers are taught a naturalistic “foreign” language, a clear preference for noun phrase word order is replicated but for a subset of modifier types originally tested. Specifically, participants preferred noun phrases with the order N-Adj-Dem (as in “mug red this”) over the order N-Dem-Adj (as in “mug this red”). However, they showed no preference between orders N-Adj-Num (as in “mugs red two”) and N-Num-Adj (as in “mugs two red”). We interpret this sensitivity as potentially reflecting an asymmetry among modifier types in the underlying hierarchical structure.

White, J., Kager., R., Linzen, T., Markopoulos, G., Martin, A., Nevins, A., Peperkamp, S., Polgárdi, K., Topintzi, N. & van de Vijver, R. (2018). Preference for locality is affected by the prefix/suffix asymmetry: Evidence from artificial language learning. In: S. Hucklebridge & M. Nelson (Eds.) Proceedings of NELS 48, Vol. 3 (pp.207–220). Amherst, MA: GLSA.

Fort, M., Weiss, A., Martin, A., & Peperkamp, S. (2013). Looking for the bouba-kiki effect in prelexical infants. In: S. Ouni, F. Berthommier & A. Jesse (Eds.) Proceedings of the 12th International Conference on Auditory-Visual Speech Processing (pp. 71–76). Lyon, France: INRIA.

Adults and toddlers systematically associate certain pseudowords, such as ‘bouba’ and ‘kiki’, with round and spiky shapes, respectively. The ontological origin of this so-called bouba-kiki effect is unknown: it could be an unlearned aspect of perception, appear with language exposure, or only emerge with the ability to produce speech sounds (i.e., babbling). We report the results of three experiments with five- and six-month-olds that found no bouba-kiki effect at all. We discuss the consequences of these findings for the emergence of cross-modal associations in infant speech perception.

You can find more information on my publications at Google scholar.