Regarding: “…speech categories are probably inborn, their boundaries can be modified or even lost as a result of learning, and weaker secondary boundaries can be generated by learning alone…”
This brings to mind an example of “phonologization” discussed in one of my previous linguistics courses regarding the historical loss of the phonemic distinction between [ba] and [pa] in many Southeast Asian languages. As discussed by Harnad (2003), [pa] morphs into [ba] when the VOT is adjusted. Stop consonant voicing is an extrinsic feature that speakers are able to control. Contrarily, the pitch height of the following vowel, -a, is an intrinsic feature that varies depends on the voicing of the preceding consonant. Over time, /p/ and /b/ merged into the single phoneme /p/ and the intrinsic phonetic feature of pitch developed into an extrinsic phonemic one, which speakers have control over.
To tie this back to the above quote on categorical perception, tone contrasts replaced the perceptual distinction between voicing contrasts, illustrating the modification of phonemic boundaries and the incorporation of an additional feature boundary.
Kreidler, Charles W., ed. Phonology: Critical concepts in linguistics. Vol. 1. Taylor & Francis, 2001.
RE: Evolved CP Categories are not simple. Some categories are learned and some are innate? Vowels, colors, and speech sounds are said to be innate and biased by evolution. I wonder what makes certain categories innate where others are learned. Is it due to an aspect of survival where the categories that help us survive are passed on through evolution and therefore innate? Or is there another explanation?
I think this has to do with the idea that predictable thing can be offloaded to the environment, so genes don't like to code for more than they have to.
“These language-induced CP-effects remain to be directly demonstrated in human subjects; so far only learned and innate sensorimotor CP have been demonstrated (Pevtzow & Harnad 1997; Livingston et al. 1998). The latter shows the Whorfian power of naming and categorization, in warping our perception of the world. That is enough to rehabilitate the Whorf Hypothesis from its apparent failure on color terms (and perhaps also from its apparent failure on eskimo snow terms, Pullum 1989), but to show that it is a full-blown language effect, and not merely a vocabulary effect, it will have to be shown that our perception of the world can also be warped, not just by how things are named but by what we are told about them.”
If there are language-induced CP effects that are Whorfian in nature, would we be happy to show this just at the level of speech-sounds (which have more obvious physical/acoustic features)? At the vocabulary or vocab-Boolean-combined level?
The last line of the text above, I’m not sure that “what we are told about [the things that are named” is so different from the Boolean combinations of different vocabulary words. As we make our categories more specific – or if we rely on different feature weights to define the categories – we really are just compounding more and different vocabulary.
I think in order to really get at the way someone may see the world, we need to also ask about “how” they are told about things, not just “what” information they receive. For example, many languages use different verb-argument relations in their syntax; how the agent or doer of a particular verb interacts with an object can be very different cross- and intra-linguistically. Think of the passive example in English: “I threw the ball” vs. “The ball was thrown by me.” Also, other languages, which are unlike English, do not have these active/passive voicing alternations at all (ergative-absolutive languages), and the way the thrower may interact with the ball in those types of sentences may look, superficially anyway, very different, maybe schematically like: ball—THING throw—VERB I—AGENT/ACTOR (ignoring differences in word order; these relationships could still be understood if the word order was rearranged).
The roles of the arguments in these sentences, based on presentation, could potentially influence how important a listener thinks the ball is, or the throwing is, or the thrower. I think in order to investigate possible Whorfian-ness in language-induced CPs, we would not only have to consider “what” names are used, “what” other words we use to talk about those names, but also “how” (syntactically, semantically) they are presented to the listener.
Given my studies in linguistics and phonetics in particular, I was very interested in the section about the motor theory of speech perception, and specifically on the distinction made between ba and pa.
This led me to thinking about something I learned when taking acoustic phonetics--the McGurk effect. Generally, when speaking to someone, it is likely that you won’t confuse whether the speech sound they produced was ba, da, or ga, because those sounds are all distinct due to how they are formed by the mouth (b is labial, d is alveolar, and g is velar). But the McGurk effect demonstrates that if you hear someone saying ‘ba’ while their lips and mouth are moving as if to pronounce ‘ga', there is a chance that your brain will combine this mismatched auditory and visual information into thinking they said ‘da’. Thus, I don’t think our ability to distinguish between speech sounds is completely innate--it is also based on learned visual perceptions, such as knowing how one’s mouth should be moving when they are producing a certain speech sound.
In his paper Harnad describes categorical perception, which should be distinguished from perceptions that exist along a continuum (e.g. size), as being both innate or learned. He also discusses how even categorical perceptions (CP) which appear to us as "all or nothing" often exist along a continuum at a microscopic level. For example, colour and phonetic sounds such as ba/pa exist along a continuum and yet our perception of them is not gradated.
I think the example given about the "ba/pa" CP is an interesting one especially considering more recent research in infants which demonstrated that the point along the continuum in which your CP switches varies according to the language the infant is exposed to. There is evidence showing that linguistic CPs are language dependent, which suggests that they are not entirely innately determined. I think intercultural studies provide good models to assess the extent to which a behaviour or cognitive capacity is innate or influenced by learned experience. It also suggests the possibility that we may be predisposed to some CPs due to innate factors but that these CPs can still be shaped by learned experience. It would be interesting to see if there are similar cultural differences in colour perception or if for some reason colour CPs are less maleable due to experience.
The nets learning to ignore other variation reminds me of the schema theory of categorization. What if you have something very different from the normal schema? For example, if you encounter some abstract design for a chair? Would you add that to the schema and modify it or would that stand our as a learned and isolated incident?
Berlin and Kay (1969) challenged the Whorf hypothesis, which said that CP for colours is learned through the way language and culture denotes the colours. According to Harnad’s reading, Berlin and Kay showed that CP is largely innate because different cultures classified colours the same way. If CP is largely innate, could this explain why apes cannot learn English as fast as a toddler? Learning English would require CP to be able to identify the continuum of sounds producible by a mouth as discrete sections. Perhaps the reason why apes cannot learn English is simply, by extension of the innate quality of CP in humans, that apes are not evolved to categorically perceive the same way as humans. Of course, pinpointing what exactly allows us to CP and what causes apes to CP differently boils down to the hard problem.
The Whorf Hypothesis seems very far-fetched, and has no proof. It cannot have any effect on how we see things, but might have an effect on our attention, and be influenced by expertise. Having so many different words for snow won’t lead us to see the differences, but seeing the differences as a result of expertise will lead us to create new names for it. How could we have named colors if we weren’t seeing them categorically in the first place and have the categorization process so similar with different languages? I believe that naming always comes after, as we must feel the need to categorize as a result of seeing something.
RE: There are even recent demonstrations that although the primary color and speech categories are probably inborn, their boundaries can be modified or even lost as a result of learning, and weaker secondary boundaries can be generated by learning alone. The article defines an “accordion effect” in categorical perception as having “within-category compression and/or between-category separation”. In other words, the differences in same-category items are less salient than the differences in across-category items. The article also makes a distinction between innate categories and learned ones and makes the point that both types of categories share sensorimotor origins, whether referring to sensorimotor experience during the life of the organism or at some point in history of the species. Phonemes are distinct units of sound in a specific language; the accordion effect implies differences between phonemes are highlighted in a specific language. Allophones are non-significant variants of a phoneme in a specific language, i.e. the sounds are phonetically similar but when found within the same environment (inside the same word) they do not change the meaning of the word. The accordion effect predicts that differences in allophones are less salient to the speakers of that language. Day-old babies have the ability to discriminate between all language sounds, even if they are allophones of the same phoneme in what will be their native tongue. This suggests they have innate categories of language sounds. As the babies are exposed to more language, this sensitivity for different allophones decreases to the point where they are no longer able to discriminate between them, i.e. the sounds undergo a within-category compression. On the other hand, their sensitivity for phonemes increases, resulting in a between-category separation. I find this fascinating because it shows us how arbitrary categories are, and how easily they are molded by our environment.
Regarding: “…speech categories are probably inborn, their boundaries can be modified or even lost as a result of learning, and weaker secondary boundaries can be generated by learning alone…”
ReplyDeleteThis brings to mind an example of “phonologization” discussed in one of my previous linguistics courses regarding the historical loss of the phonemic distinction between [ba] and [pa] in many Southeast Asian languages. As discussed by Harnad (2003), [pa] morphs into [ba] when the VOT is adjusted. Stop consonant voicing is an extrinsic feature that speakers are able to control. Contrarily, the pitch height of the following vowel, -a, is an intrinsic feature that varies depends on the voicing of the preceding consonant. Over time, /p/ and /b/ merged into the single phoneme /p/ and the intrinsic phonetic feature of pitch developed into an extrinsic phonemic one, which speakers have control over.
To tie this back to the above quote on categorical perception, tone contrasts replaced the perceptual distinction between voicing contrasts, illustrating the modification of phonemic boundaries and the incorporation of an additional feature boundary.
Kreidler, Charles W., ed. Phonology: Critical concepts in linguistics. Vol. 1. Taylor & Francis, 2001.
RE: Evolved CP
ReplyDeleteCategories are not simple. Some categories are learned and some are innate? Vowels, colors, and speech sounds are said to be innate and biased by evolution. I wonder what makes certain categories innate where others are learned. Is it due to an aspect of survival where the categories that help us survive are passed on through evolution and therefore innate? Or is there another explanation?
Hi Jenna,
DeleteI think this has to do with the idea that predictable thing can be offloaded to the environment, so genes don't like to code for more than they have to.
“These language-induced CP-effects remain to be directly demonstrated in human subjects; so far only learned and innate sensorimotor CP have been demonstrated (Pevtzow & Harnad 1997; Livingston et al. 1998). The latter shows the Whorfian power of naming and categorization, in warping our perception of the world. That is enough to rehabilitate the Whorf Hypothesis from its apparent failure on color terms (and perhaps also from its apparent failure on eskimo snow terms, Pullum 1989), but to show that it is a full-blown language effect, and not merely a vocabulary effect, it will have to be shown that our perception of the world can also be warped, not just by how things are named but by what we are told about them.”
ReplyDeleteIf there are language-induced CP effects that are Whorfian in nature, would we be happy to show this just at the level of speech-sounds (which have more obvious physical/acoustic features)? At the vocabulary or vocab-Boolean-combined level?
The last line of the text above, I’m not sure that “what we are told about [the things that are named” is so different from the Boolean combinations of different vocabulary words. As we make our categories more specific – or if we rely on different feature weights to define the categories – we really are just compounding more and different vocabulary.
I think in order to really get at the way someone may see the world, we need to also ask about “how” they are told about things, not just “what” information they receive. For example, many languages use different verb-argument relations in their syntax; how the agent or doer of a particular verb interacts with an object can be very different cross- and intra-linguistically. Think of the passive example in English: “I threw the ball” vs. “The ball was thrown by me.” Also, other languages, which are unlike English, do not have these active/passive voicing alternations at all (ergative-absolutive languages), and the way the thrower may interact with the ball in those types of sentences may look, superficially anyway, very different, maybe schematically like: ball—THING throw—VERB I—AGENT/ACTOR (ignoring differences in word order; these relationships could still be understood if the word order was rearranged).
The roles of the arguments in these sentences, based on presentation, could potentially influence how important a listener thinks the ball is, or the throwing is, or the thrower. I think in order to investigate possible Whorfian-ness in language-induced CPs, we would not only have to consider “what” names are used, “what” other words we use to talk about those names, but also “how” (syntactically, semantically) they are presented to the listener.
Given my studies in linguistics and phonetics in particular, I was very interested in the section about the motor theory of speech perception, and specifically on the distinction made between ba and pa.
ReplyDeleteThis led me to thinking about something I learned when taking acoustic phonetics--the McGurk effect. Generally, when speaking to someone, it is likely that you won’t confuse whether the speech sound they produced was ba, da, or ga, because those sounds are all distinct due to how they are formed by the mouth (b is labial, d is alveolar, and g is velar). But the McGurk effect demonstrates that if you hear someone saying ‘ba’ while their lips and mouth are moving as if to pronounce ‘ga', there is a chance that your brain will combine this mismatched auditory and visual information into thinking they said ‘da’. Thus, I don’t think our ability to distinguish between speech sounds is completely innate--it is also based on learned visual perceptions, such as knowing how one’s mouth should be moving when they are producing a certain speech sound.
In his paper Harnad describes categorical perception, which should be distinguished from perceptions that exist along a continuum (e.g. size), as being both innate or learned. He also discusses how even categorical perceptions (CP) which appear to us as "all or nothing" often exist along a continuum at a microscopic level. For example, colour and phonetic sounds such as ba/pa exist along a continuum and yet our perception of them is not gradated.
ReplyDeleteI think the example given about the "ba/pa" CP is an interesting one especially considering more recent research in infants which demonstrated that the point along the continuum in which your CP switches varies according to the language the infant is exposed to. There is evidence showing that linguistic CPs are language dependent, which suggests that they are not entirely innately determined. I think intercultural studies provide good models to assess the extent to which a behaviour or cognitive capacity is innate or influenced by learned experience. It also suggests the possibility that we may be predisposed to some CPs due to innate factors but that these CPs can still be shaped by learned experience. It would be interesting to see if there are similar cultural differences in colour perception or if for some reason colour CPs are less maleable due to experience.
The nets learning to ignore other variation reminds me of the schema theory of categorization. What if you have something very different from the normal schema? For example, if you encounter some abstract design for a chair? Would you add that to the schema and modify it or would that stand our as a learned and isolated incident?
ReplyDeleteBerlin and Kay (1969) challenged the Whorf hypothesis, which said that CP for colours is learned through the way language and culture denotes the colours. According to Harnad’s reading, Berlin and Kay showed that CP is largely innate because different cultures classified colours the same way. If CP is largely innate, could this explain why apes cannot learn English as fast as a toddler? Learning English would require CP to be able to identify the continuum of sounds producible by a mouth as discrete sections. Perhaps the reason why apes cannot learn English is simply, by extension of the innate quality of CP in humans, that apes are not evolved to categorically perceive the same way as humans. Of course, pinpointing what exactly allows us to CP and what causes apes to CP differently boils down to the hard problem.
ReplyDeleteThe Whorf Hypothesis seems very far-fetched, and has no proof. It cannot have any effect on how we see things, but might have an effect on our attention, and be influenced by expertise. Having so many different words for snow won’t lead us to see the differences, but seeing the differences as a result of expertise will lead us to create new names for it. How could we have named colors if we weren’t seeing them categorically in the first place and have the categorization process so similar with different languages? I believe that naming always comes after, as we must feel the need to categorize as a result of seeing something.
ReplyDeleteRE: There are even recent demonstrations that although the primary color and speech categories are probably inborn, their boundaries can be modified or even lost as a result of learning, and weaker secondary boundaries can be generated by learning alone.
ReplyDeleteThe article defines an “accordion effect” in categorical perception as having “within-category compression and/or between-category separation”. In other words, the differences in same-category items are less salient than the differences in across-category items. The article also makes a distinction between innate categories and learned ones and makes the point that both types of categories share sensorimotor origins, whether referring to sensorimotor experience during the life of the organism or at some point in history of the species.
Phonemes are distinct units of sound in a specific language; the accordion effect implies differences between phonemes are highlighted in a specific language. Allophones are non-significant variants of a phoneme in a specific language, i.e. the sounds are phonetically similar but when found within the same environment (inside the same word) they do not change the meaning of the word. The accordion effect predicts that differences in allophones are less salient to the speakers of that language. Day-old babies have the ability to discriminate between all language sounds, even if they are allophones of the same phoneme in what will be their native tongue. This suggests they have innate categories of language sounds. As the babies are exposed to more language, this sensitivity for different allophones decreases to the point where they are no longer able to discriminate between them, i.e. the sounds undergo a within-category compression. On the other hand, their sensitivity for phonemes increases, resulting in a between-category separation. I find this fascinating because it shows us how arbitrary categories are, and how easily they are molded by our environment.