How well do Humans Capture the Sounds of Speech in Writing? (original) (raw)
Related papers
Royal Society Open Science, 2017
In three experiments, we asked whether diverse scripts contain interpretable information about the speech sounds they represent. When presented with a pair of unfamiliar letters, adult readers correctly guess which is /i/ (the ‘ee’ sound in ‘feet’), and which is /u/ (the ‘oo’ sound in ‘shoe’) at rates higher than expected by chance, as shown in a large sample of Singaporean university students (Experiment 1) and replicated in a larger sample of international Internet users (Experiment 2). To uncover what properties of the letters contribute to different scripts' ‘guessability,’ we analysed the visual spatial frequencies in each letter (Experiment 3). We predicted that the lower spectral frequencies in the formants of the vowel /u/ would pattern with lower spatial frequencies in the corresponding letters. Instead, we found that across all spatial frequencies, the letter with more black/white cycles (i.e. more ink) was more likely to be guessed as /u/, and the larger the differenc...
2018
We demonstrate, for the first time, significant systematicity between the visual form of lettersof the Roman alphabet and their paradigmatic English pronunciation. We measure the visual distance between letters as Hausdorff distance and the phonological distance between their pronunciations as feature-edit distance. These two sets of distances are significantly positively correlated: letters that look the same tend to be pronounced the same. We discuss the implications for the teaching of the alphabetic principle in learning to read.
Linking the shapes of alphabet letters to their sounds: the case of Hebrew
Learning the sounds of letters is an important part of learning a writing system. Most previous studies of this process have examined English, focusing on variations in the phonetic iconicity of letter names as a reason why some letter sounds (such as that of b, where the sound is at the beginning of the letter's name) are easier to learn than others (such as that of w, where the sound is not in the name). The present study examined Hebrew, where variations in the phonetic iconicity of letter names are minimal. In a study of 391 Israeli children with a mean age of 5 years, 10 months, we used multilevel models to examine the factors that are associated with knowledge of letter sounds. One set of factors involved letter names: Children sometimes attributed to a letter a consonant–vowel sound consisting of the first phonemes of the letter's name. A second set of factors involved contrast: Children had difficulty when there was relatively little contrast in shape between one letter and others. Frequency was also important, encompassing both child-specific effects, such as a benefit for the first letter of a child's forename, and effects that held true across children, such as a benefit for the first letters of the alphabet. These factors reflect general properties of human learning.
Phonovisual Biases in Language: is the Lexicon Tied to the Visual World?
2021
The present paper addresses the study of crosslinguistic and cross-modal iconicity within a deep learning framework. An LSTM-based Recurrent Neural Network is trained to associate the phonetic representation of a concrete word, encoded as a sequence of feature vectors, to the visual representation of its referent, expressed as an HCNNtransformed image. The processing network is then tested, without further training, in a language that does not appear in the training set and belongs to a different language family. The performance of the model is evaluated through a comparison with a randomized baseline; we show that such an imaginative network is capable of extracting languageindependent generalizations in the mapping from linguistic sounds to visual features, providing empirical support for the hypothesis of a universal sound-symbolic substrate underlying all languages.
Nonarbitrary mappings between sound and shape (i.e., the bouba-kiki effect) have been shown across different cultures and early in development; however, the level of processing at which this effect arises remains unclear. Here we show that the mapping occurs prior to conscious awareness of the visual stimuli. Under continuous flash suppression, congruent stimuli (e.g., “kiki” inside an angular shape) broke through to conscious awareness faster than incongruent stimuli. This was true even when we trained people to pair unfamiliar letters with auditory word forms, a result showing that the effect was driven by the phonology, not the visual features, of the letters. Furthermore, visibility thresholds of the shapes decreased when they were preceded by a congruent auditory word form in a masking paradigm. Taken together, our results suggest that sound-shape mapping can occur automatically prior to conscious awareness of visual shapes, and that sensory congruence facilitates conscious awareness of a stimulus being present.