Source features
These are related to the source (vocal fold vibration that sustains voiced sounds or a turbulent airstream that sustains many voiceless sounds).
The feature [voice] is self-explanatory (with or without vocal fold vibration). The feature [spread glottis] is used to distinguish aspirated from unaspirated stops (aspirated stops are initially produced with the vocal folds drawn apart). We can therefore make the following distinctions:
voiced
spread glottis
p
-
-
b
+
-
pʰ
-
+
The [strident] feature is used by Halle and Clements for those fricatives produced with high-intensity fricative noise: supposedly labiodentals, alveolars, palato-alveolars, and uvulars are [+strident]. There seems to be little acoustic phonetic basis to the claim that labiodentals and alveolars pattern acoustically (as opposed to dentals). In this course, we will use Ladefoged's feature [sibilant] which is defined by Ladefoged (1971) in acoustic terms as including those fricatives with 'large amounts of acoustic energy at high frequencies' i.e. [s ʃ z ʒ]. The English affricates would therefore also be [+sibilant]:
cont
sibilant
oral stops
-
-
affricates
-
+
sibilant fricatives
+
+
non-sibilant fricatives
+
-
This may be an oversimplification, however, as the alveolar oral stops might also be described as sibilant, so sibilant isn't sufficient to separate oral stops and affricates (and we still need spread glottis).
Monday, April 16, 2012
Chomsky and Halle (1968-1983)
An overview of commonly used distinctive features
The features described in Halle & Clements (1983) have been commonly used in the phonology literature in their analyses of the sound patterns of various languages. They incorporate many insights of the original features devised by Jakobson (1941) but are mostly based on those of the Sound Pattern of English, taking into account some modifications suggested by Halle & Stevens (1971). Most of these are also discussed below.
i. Major class features
Four features [syll], [cons], [son], [cont] (syllabic, consonantal, sonorant, continuant) are used to divide up speech sounds into major classes, as follows. Note that [syll] means "syllabic" (syllable nucleus), [cons] means "consonantal", [son] means "sonorant" (periodic low frequency energy), [cont] means "continuant" (continuous airflow through oral cavity), and [delrel] means delayed release (release is not "delayed", but there is a longer aspiration phase than oral stops - nb. voice onset is what's actually delayed).
syll
cons
son
cont
delrel
vowels
+
-
+
+
0
oral stops
-
+
-
-
-
affricates
-
+
-
-
+
nasal stops
-
+
+
-
0
fricatives
-
+
-
+
0
liquids
-
+
+
+
0
semi-vowels
-
-
+
+
0
Note that the approximants have been divided into liquids (eg. in English /r, l/) and semi-vowels (eg. in English /w, j/). In this, and most other distinctive feature sets derived from Chomsky and Halle. Semi-vowels (being [-syll, -cons]) form a class of sounds intermediate between vowels ([+syll]) and consonants ([+cons]). The approximants can be defined as a class by the features [-syll, +son, +cont] and can be further sub-divided into liquids and semi-vowels using the [cons] feature. Note that "0" means irrelevant feature for these classes of sounds (there's nothing to release).
We also have a feature [nasal] which, as its name suggests, separates nasal from oral sounds. In the above table, [nasal] would have been redundant as the nasal stops are already defined uniquely as [-syll, +cons, +son, -cont] (ie. as sonorant stops). However. the feature [nasal] is required to define nasal stops, nasalised vowels and nasalised approximants as a single natural class.
The features described in Halle & Clements (1983) have been commonly used in the phonology literature in their analyses of the sound patterns of various languages. They incorporate many insights of the original features devised by Jakobson (1941) but are mostly based on those of the Sound Pattern of English, taking into account some modifications suggested by Halle & Stevens (1971). Most of these are also discussed below.
i. Major class features
Four features [syll], [cons], [son], [cont] (syllabic, consonantal, sonorant, continuant) are used to divide up speech sounds into major classes, as follows. Note that [syll] means "syllabic" (syllable nucleus), [cons] means "consonantal", [son] means "sonorant" (periodic low frequency energy), [cont] means "continuant" (continuous airflow through oral cavity), and [delrel] means delayed release (release is not "delayed", but there is a longer aspiration phase than oral stops - nb. voice onset is what's actually delayed).
syll
cons
son
cont
delrel
vowels
+
-
+
+
0
oral stops
-
+
-
-
-
affricates
-
+
-
-
+
nasal stops
-
+
+
-
0
fricatives
-
+
-
+
0
liquids
-
+
+
+
0
semi-vowels
-
-
+
+
0
Note that the approximants have been divided into liquids (eg. in English /r, l/) and semi-vowels (eg. in English /w, j/). In this, and most other distinctive feature sets derived from Chomsky and Halle. Semi-vowels (being [-syll, -cons]) form a class of sounds intermediate between vowels ([+syll]) and consonants ([+cons]). The approximants can be defined as a class by the features [-syll, +son, +cont] and can be further sub-divided into liquids and semi-vowels using the [cons] feature. Note that "0" means irrelevant feature for these classes of sounds (there's nothing to release).
We also have a feature [nasal] which, as its name suggests, separates nasal from oral sounds. In the above table, [nasal] would have been redundant as the nasal stops are already defined uniquely as [-syll, +cons, +son, -cont] (ie. as sonorant stops). However. the feature [nasal] is required to define nasal stops, nasalised vowels and nasalised approximants as a single natural class.
Chomsky and Halle (1968-1983)
Phonetic interpretation
According to Jakobson (1941), the distinctive features should have definable articulatory and acoustic correlates. For example, [+nasal] implies a lowering of the soft-palate and also an increase in the ratio of energy in the low to the high part of the spectrum. Chomsky & Halle (1968) abandoned the acoustic definitions of phonological features (inappropriately, as Ladefoged, 1971 and many others have argued: for example [f] and [x] are related acoustically but not articulatorily and they participated in the sound change by which the pronunciation of 'gh' spellings in English changed from a velar to a labiodental fricative e.g. 'laugh', [lɑx]→[lɑf]).
Many of the features are defined loosely in phonetic terms. This is perhaps to be expected. Phonology has established highly abstract representations to explain sound alternations (i.e. to factor out what are considered redundant or predictable aspects of a word's pronunciation) and this abstraction is partly opposed to the principle in phonetics of describing in articulatory and acoustic terms the characteristics of speech sound production that are shared by linguistic communities. Nevertheless, if phonology is to be related to how words are actually pronounced, the features are required to have at least some phonetic basis to them.
According to Jakobson (1941), the distinctive features should have definable articulatory and acoustic correlates. For example, [+nasal] implies a lowering of the soft-palate and also an increase in the ratio of energy in the low to the high part of the spectrum. Chomsky & Halle (1968) abandoned the acoustic definitions of phonological features (inappropriately, as Ladefoged, 1971 and many others have argued: for example [f] and [x] are related acoustically but not articulatorily and they participated in the sound change by which the pronunciation of 'gh' spellings in English changed from a velar to a labiodental fricative e.g. 'laugh', [lɑx]→[lɑf]).
Many of the features are defined loosely in phonetic terms. This is perhaps to be expected. Phonology has established highly abstract representations to explain sound alternations (i.e. to factor out what are considered redundant or predictable aspects of a word's pronunciation) and this abstraction is partly opposed to the principle in phonetics of describing in articulatory and acoustic terms the characteristics of speech sound production that are shared by linguistic communities. Nevertheless, if phonology is to be related to how words are actually pronounced, the features are required to have at least some phonetic basis to them.
Chomsky and Halle (1968-1983)
c) Binarity
We have assumed that features are binary (a segment is either nasal or it is not) following Jakobson's (1941) original formulation of distinctive feature theory and this premise was adopted in Chomsky & Halle's (1968) Sound Pattern of English. There were many reasons why Jakobson (1941) advocated a binary approach. Firstly, as we have seen, this is the most efficient way of reducing the phoneme inventory of a language. Secondly, he argued that most phonological oppositions are binary in nature (e.g. sounds either are or are not produced with a lowered soft-palate and nasalisation) and he even proposed that it has a physiological basis i.e. that nerve fibers have an 'all-or-none' response. But the binary principal is certainly not adopted by all linguists, and many phoneticians in particular have argued that some features should be n-ary (where "n" is any relevant number of degrees or levels - see for example, Ladefoged's 1993 treatment of vowel height which is 4-valued to reflect the distinction between close, half-close, half-open, and open vowels).
We have assumed that features are binary (a segment is either nasal or it is not) following Jakobson's (1941) original formulation of distinctive feature theory and this premise was adopted in Chomsky & Halle's (1968) Sound Pattern of English. There were many reasons why Jakobson (1941) advocated a binary approach. Firstly, as we have seen, this is the most efficient way of reducing the phoneme inventory of a language. Secondly, he argued that most phonological oppositions are binary in nature (e.g. sounds either are or are not produced with a lowered soft-palate and nasalisation) and he even proposed that it has a physiological basis i.e. that nerve fibers have an 'all-or-none' response. But the binary principal is certainly not adopted by all linguists, and many phoneticians in particular have argued that some features should be n-ary (where "n" is any relevant number of degrees or levels - see for example, Ladefoged's 1993 treatment of vowel height which is 4-valued to reflect the distinction between close, half-close, half-open, and open vowels).
Subscribe to:
Posts (Atom)
Chomsky and Halle (1968-1983)
Source features
These are related to the source (vocal fold vibration that sustains voiced sounds or a turbulent airstream that sustains many voiceless sounds).
The feature [voice] is self-explanatory (with or without vocal fold vibration). The feature [spread glottis] is used to distinguish aspirated from unaspirated stops (aspirated stops are initially produced with the vocal folds drawn apart). We can therefore make the following distinctions:
voiced
spread glottis
p
-
-
b
+
-
pʰ
-
+
The [strident] feature is used by Halle and Clements for those fricatives produced with high-intensity fricative noise: supposedly labiodentals, alveolars, palato-alveolars, and uvulars are [+strident]. There seems to be little acoustic phonetic basis to the claim that labiodentals and alveolars pattern acoustically (as opposed to dentals). In this course, we will use Ladefoged's feature [sibilant] which is defined by Ladefoged (1971) in acoustic terms as including those fricatives with 'large amounts of acoustic energy at high frequencies' i.e. [s ʃ z ʒ]. The English affricates would therefore also be [+sibilant]:
cont
sibilant
oral stops
-
-
affricates
-
+
sibilant fricatives
+
+
non-sibilant fricatives
+
-
This may be an oversimplification, however, as the alveolar oral stops might also be described as sibilant, so sibilant isn't sufficient to separate oral stops and affricates (and we still need spread glottis).
These are related to the source (vocal fold vibration that sustains voiced sounds or a turbulent airstream that sustains many voiceless sounds).
The feature [voice] is self-explanatory (with or without vocal fold vibration). The feature [spread glottis] is used to distinguish aspirated from unaspirated stops (aspirated stops are initially produced with the vocal folds drawn apart). We can therefore make the following distinctions:
voiced
spread glottis
p
-
-
b
+
-
pʰ
-
+
The [strident] feature is used by Halle and Clements for those fricatives produced with high-intensity fricative noise: supposedly labiodentals, alveolars, palato-alveolars, and uvulars are [+strident]. There seems to be little acoustic phonetic basis to the claim that labiodentals and alveolars pattern acoustically (as opposed to dentals). In this course, we will use Ladefoged's feature [sibilant] which is defined by Ladefoged (1971) in acoustic terms as including those fricatives with 'large amounts of acoustic energy at high frequencies' i.e. [s ʃ z ʒ]. The English affricates would therefore also be [+sibilant]:
cont
sibilant
oral stops
-
-
affricates
-
+
sibilant fricatives
+
+
non-sibilant fricatives
+
-
This may be an oversimplification, however, as the alveolar oral stops might also be described as sibilant, so sibilant isn't sufficient to separate oral stops and affricates (and we still need spread glottis).
Chomsky and Halle (1968-1983)
An overview of commonly used distinctive features
The features described in Halle & Clements (1983) have been commonly used in the phonology literature in their analyses of the sound patterns of various languages. They incorporate many insights of the original features devised by Jakobson (1941) but are mostly based on those of the Sound Pattern of English, taking into account some modifications suggested by Halle & Stevens (1971). Most of these are also discussed below.
i. Major class features
Four features [syll], [cons], [son], [cont] (syllabic, consonantal, sonorant, continuant) are used to divide up speech sounds into major classes, as follows. Note that [syll] means "syllabic" (syllable nucleus), [cons] means "consonantal", [son] means "sonorant" (periodic low frequency energy), [cont] means "continuant" (continuous airflow through oral cavity), and [delrel] means delayed release (release is not "delayed", but there is a longer aspiration phase than oral stops - nb. voice onset is what's actually delayed).
syll
cons
son
cont
delrel
vowels
+
-
+
+
0
oral stops
-
+
-
-
-
affricates
-
+
-
-
+
nasal stops
-
+
+
-
0
fricatives
-
+
-
+
0
liquids
-
+
+
+
0
semi-vowels
-
-
+
+
0
Note that the approximants have been divided into liquids (eg. in English /r, l/) and semi-vowels (eg. in English /w, j/). In this, and most other distinctive feature sets derived from Chomsky and Halle. Semi-vowels (being [-syll, -cons]) form a class of sounds intermediate between vowels ([+syll]) and consonants ([+cons]). The approximants can be defined as a class by the features [-syll, +son, +cont] and can be further sub-divided into liquids and semi-vowels using the [cons] feature. Note that "0" means irrelevant feature for these classes of sounds (there's nothing to release).
We also have a feature [nasal] which, as its name suggests, separates nasal from oral sounds. In the above table, [nasal] would have been redundant as the nasal stops are already defined uniquely as [-syll, +cons, +son, -cont] (ie. as sonorant stops). However. the feature [nasal] is required to define nasal stops, nasalised vowels and nasalised approximants as a single natural class.
The features described in Halle & Clements (1983) have been commonly used in the phonology literature in their analyses of the sound patterns of various languages. They incorporate many insights of the original features devised by Jakobson (1941) but are mostly based on those of the Sound Pattern of English, taking into account some modifications suggested by Halle & Stevens (1971). Most of these are also discussed below.
i. Major class features
Four features [syll], [cons], [son], [cont] (syllabic, consonantal, sonorant, continuant) are used to divide up speech sounds into major classes, as follows. Note that [syll] means "syllabic" (syllable nucleus), [cons] means "consonantal", [son] means "sonorant" (periodic low frequency energy), [cont] means "continuant" (continuous airflow through oral cavity), and [delrel] means delayed release (release is not "delayed", but there is a longer aspiration phase than oral stops - nb. voice onset is what's actually delayed).
syll
cons
son
cont
delrel
vowels
+
-
+
+
0
oral stops
-
+
-
-
-
affricates
-
+
-
-
+
nasal stops
-
+
+
-
0
fricatives
-
+
-
+
0
liquids
-
+
+
+
0
semi-vowels
-
-
+
+
0
Note that the approximants have been divided into liquids (eg. in English /r, l/) and semi-vowels (eg. in English /w, j/). In this, and most other distinctive feature sets derived from Chomsky and Halle. Semi-vowels (being [-syll, -cons]) form a class of sounds intermediate between vowels ([+syll]) and consonants ([+cons]). The approximants can be defined as a class by the features [-syll, +son, +cont] and can be further sub-divided into liquids and semi-vowels using the [cons] feature. Note that "0" means irrelevant feature for these classes of sounds (there's nothing to release).
We also have a feature [nasal] which, as its name suggests, separates nasal from oral sounds. In the above table, [nasal] would have been redundant as the nasal stops are already defined uniquely as [-syll, +cons, +son, -cont] (ie. as sonorant stops). However. the feature [nasal] is required to define nasal stops, nasalised vowels and nasalised approximants as a single natural class.
Chomsky and Halle (1968-1983)
Phonetic interpretation
According to Jakobson (1941), the distinctive features should have definable articulatory and acoustic correlates. For example, [+nasal] implies a lowering of the soft-palate and also an increase in the ratio of energy in the low to the high part of the spectrum. Chomsky & Halle (1968) abandoned the acoustic definitions of phonological features (inappropriately, as Ladefoged, 1971 and many others have argued: for example [f] and [x] are related acoustically but not articulatorily and they participated in the sound change by which the pronunciation of 'gh' spellings in English changed from a velar to a labiodental fricative e.g. 'laugh', [lɑx]→[lɑf]).
Many of the features are defined loosely in phonetic terms. This is perhaps to be expected. Phonology has established highly abstract representations to explain sound alternations (i.e. to factor out what are considered redundant or predictable aspects of a word's pronunciation) and this abstraction is partly opposed to the principle in phonetics of describing in articulatory and acoustic terms the characteristics of speech sound production that are shared by linguistic communities. Nevertheless, if phonology is to be related to how words are actually pronounced, the features are required to have at least some phonetic basis to them.
According to Jakobson (1941), the distinctive features should have definable articulatory and acoustic correlates. For example, [+nasal] implies a lowering of the soft-palate and also an increase in the ratio of energy in the low to the high part of the spectrum. Chomsky & Halle (1968) abandoned the acoustic definitions of phonological features (inappropriately, as Ladefoged, 1971 and many others have argued: for example [f] and [x] are related acoustically but not articulatorily and they participated in the sound change by which the pronunciation of 'gh' spellings in English changed from a velar to a labiodental fricative e.g. 'laugh', [lɑx]→[lɑf]).
Many of the features are defined loosely in phonetic terms. This is perhaps to be expected. Phonology has established highly abstract representations to explain sound alternations (i.e. to factor out what are considered redundant or predictable aspects of a word's pronunciation) and this abstraction is partly opposed to the principle in phonetics of describing in articulatory and acoustic terms the characteristics of speech sound production that are shared by linguistic communities. Nevertheless, if phonology is to be related to how words are actually pronounced, the features are required to have at least some phonetic basis to them.
Chomsky and Halle (1968-1983)
c) Binarity
We have assumed that features are binary (a segment is either nasal or it is not) following Jakobson's (1941) original formulation of distinctive feature theory and this premise was adopted in Chomsky & Halle's (1968) Sound Pattern of English. There were many reasons why Jakobson (1941) advocated a binary approach. Firstly, as we have seen, this is the most efficient way of reducing the phoneme inventory of a language. Secondly, he argued that most phonological oppositions are binary in nature (e.g. sounds either are or are not produced with a lowered soft-palate and nasalisation) and he even proposed that it has a physiological basis i.e. that nerve fibers have an 'all-or-none' response. But the binary principal is certainly not adopted by all linguists, and many phoneticians in particular have argued that some features should be n-ary (where "n" is any relevant number of degrees or levels - see for example, Ladefoged's 1993 treatment of vowel height which is 4-valued to reflect the distinction between close, half-close, half-open, and open vowels).
We have assumed that features are binary (a segment is either nasal or it is not) following Jakobson's (1941) original formulation of distinctive feature theory and this premise was adopted in Chomsky & Halle's (1968) Sound Pattern of English. There were many reasons why Jakobson (1941) advocated a binary approach. Firstly, as we have seen, this is the most efficient way of reducing the phoneme inventory of a language. Secondly, he argued that most phonological oppositions are binary in nature (e.g. sounds either are or are not produced with a lowered soft-palate and nasalisation) and he even proposed that it has a physiological basis i.e. that nerve fibers have an 'all-or-none' response. But the binary principal is certainly not adopted by all linguists, and many phoneticians in particular have argued that some features should be n-ary (where "n" is any relevant number of degrees or levels - see for example, Ladefoged's 1993 treatment of vowel height which is 4-valued to reflect the distinction between close, half-close, half-open, and open vowels).
Subscribe to:
Posts (Atom)