Campbell chapter 6
Subgrouping: the classification of which languages are more closely related to one another in a
language family tree
6.2 The world’s language families
420 distinctive language families, but not all languages are already in a tree.
24% of the world’s linguistic diversity is gone.
6.2.1 Language isolates and their history
Language isolate = has no known relatives, no demonstrable genetic relationship with any other
language.
They are not very different from language families with multiple members.
- Sometimes their relatives have disappeared, which we do not know about (otherwise they would
not be isolates, but just a language that survived as the only one of many)
- Language isolates and trees are not different from one another, because sometimes languages were
isolate before it was discovered that they have unknown relatives
There are also unclassified languages
- Languages which are extinct and not grouped with other languages yet
- Languages with data so unknown that they cannot yet be classified
They could be language isolates, but this cannot be said because of the lack of data
32% of the 420 languages families are language isolates [1/3th]
The general attitude is that language isolates are highly unusual, and treated with suspicion and it is
hard to accept that some languages do not have relatives. Lack of knowledge
6.2.2 How can we advance knowledge of the history of language isolates?
How can we learn about the history of a language without relatives?
1. Internal reconstruction.
This is best known and most used tool to investigate language.
Example: “Wine” is “ardo” in isolation and “ardon” in compounds. Internal reconstruction would then
show *Ardano [or something in which the N is present]. Internal reconstruction postulates a change
of –n– ø (loss of intervocalic N)
2. Comparative reconstruction.
A less known but extremely valuable tool for investigating the history of language isolates. It does not
separate related languages, but forms from regional dialects.
3. Loanwords.
They can provide excellent evidence of the history of isolates. Basque has a lot of Latin words so it
shows that the Romans had much influence there. Furthermore, the age of the loanwords says
something too.
4. Wörter und Sachen.
This can also help with information.
- Words which can be analysed into multiple morphemes were created more recently in the history
than words which do not have this internal composition.
Example: Beer is more recent than wine in Basque (garagar (barley) + ardo (wine) = garagardo (beer))
- Place names that can be analysed into component parts are also more recent than place names
,which cannot.
- Words with non-productive (irregular) morphemes are possibly older than words with productive
morphemes.
6.3 Terminology
Terms are not always used consistently.
Subgrouping: Internal classification of the languages within language families: Branches of a family
tree and which sister languages are most closely related to one another
Dialect: A variety [regional/social] of a language which is mutually intelligible with other dialects of
the same language. Not used: little-known (exotic) or minority language/daughter language of a
language family
Language: any distinct linguistic entity (variety) which is mutually unintelligible with other such
entities.
Language family: Group of genetically related languages, which share a linguistic kinship by virtue of
having developed from a common ancestor. They have [-an] as a suffix in the family name.
Isolates: Language without known relatives; a family with one single member.
Subgroup: A group of languages within a language family which are more closely related to each
other than to other languages of that family.
6.4 How to draw family trees: subgrouping
Languages are not only part of one family, but they are also closely related to one another within the
family [sister-languages].
Language A splits up in Language B and C, Language B splits up in B1 and B2, and Language C splits up
in language C1 and C2:
Proto-Language A
Language B Language C
Language B1 Language B2 Language C1 Language C2
The only accepted criterion for subgrouping is shared innovation: linguistic change which shows a
departure (innovation) from some traits of the proto-language and is shared by a subset of the
daughter languages.
, Chapter 17: Glottochronology
17.2.1 Basic assumptions
1) Basic vocabulary.
There is a basic vocabulary which is universal and relatively culture free and thus is less subject to
replacement than other kinds of vocabulary.
2) Constant rate of retention through time.
The rate of retention of items of core vocabulary is relatively constant through time, that a language
will retain about 86% of the words of the 100-word list reach 1000 years.
3) Constant rate of loss cross-linguistically.
Claims that the rate of loss of basic vocabulary is approximately the same for all languages (14% of
the 100-word list)
4) Calculation of the date of divergence.
When the number of ‘cognates’ in the basic vocabulary list shared by related languages is known, the
number of centuries since the languages split from an earlier ancestor can be computed formula
logC
t = ------
2Logr
t = time depth in millennia [1000 years]
C = percentage of cognates
r = retention rate [% of cognates assumed to remain after 100 years 86% of 100-word list]
log = logarithm of
17.2.3 Criticism on Glottochronology
1) Criticism on basic vocabulary
- Many of the words on the list are not culture-free
- There is not always a one-to-one matching between each word on the Swadesh list
- Some of the 100 words do not have an equivalent on the list
- Some basic vocabulary changes rapidly for cultural reasons
- Taboo has resulted in the replacement of considerable vocabulary
2) Criticism on Constant rate of retention through time and 3) Constant rate of loss cross-linguistically
- Common sense will make you doubt this, since there is no logical reason for a pattern
- Some languages extremely deviate
- There are not a lot language that have 1000 year database
4) Criticism on the calculation of the date of divergence.
- The split-up of language families (or subgroups) is not sudden, so one date is unrealistic
- Sometimes languages still have things in common that stem from the time they were already split
- The statistical model is criticised
Subgrouping: the classification of which languages are more closely related to one another in a
language family tree
6.2 The world’s language families
420 distinctive language families, but not all languages are already in a tree.
24% of the world’s linguistic diversity is gone.
6.2.1 Language isolates and their history
Language isolate = has no known relatives, no demonstrable genetic relationship with any other
language.
They are not very different from language families with multiple members.
- Sometimes their relatives have disappeared, which we do not know about (otherwise they would
not be isolates, but just a language that survived as the only one of many)
- Language isolates and trees are not different from one another, because sometimes languages were
isolate before it was discovered that they have unknown relatives
There are also unclassified languages
- Languages which are extinct and not grouped with other languages yet
- Languages with data so unknown that they cannot yet be classified
They could be language isolates, but this cannot be said because of the lack of data
32% of the 420 languages families are language isolates [1/3th]
The general attitude is that language isolates are highly unusual, and treated with suspicion and it is
hard to accept that some languages do not have relatives. Lack of knowledge
6.2.2 How can we advance knowledge of the history of language isolates?
How can we learn about the history of a language without relatives?
1. Internal reconstruction.
This is best known and most used tool to investigate language.
Example: “Wine” is “ardo” in isolation and “ardon” in compounds. Internal reconstruction would then
show *Ardano [or something in which the N is present]. Internal reconstruction postulates a change
of –n– ø (loss of intervocalic N)
2. Comparative reconstruction.
A less known but extremely valuable tool for investigating the history of language isolates. It does not
separate related languages, but forms from regional dialects.
3. Loanwords.
They can provide excellent evidence of the history of isolates. Basque has a lot of Latin words so it
shows that the Romans had much influence there. Furthermore, the age of the loanwords says
something too.
4. Wörter und Sachen.
This can also help with information.
- Words which can be analysed into multiple morphemes were created more recently in the history
than words which do not have this internal composition.
Example: Beer is more recent than wine in Basque (garagar (barley) + ardo (wine) = garagardo (beer))
- Place names that can be analysed into component parts are also more recent than place names
,which cannot.
- Words with non-productive (irregular) morphemes are possibly older than words with productive
morphemes.
6.3 Terminology
Terms are not always used consistently.
Subgrouping: Internal classification of the languages within language families: Branches of a family
tree and which sister languages are most closely related to one another
Dialect: A variety [regional/social] of a language which is mutually intelligible with other dialects of
the same language. Not used: little-known (exotic) or minority language/daughter language of a
language family
Language: any distinct linguistic entity (variety) which is mutually unintelligible with other such
entities.
Language family: Group of genetically related languages, which share a linguistic kinship by virtue of
having developed from a common ancestor. They have [-an] as a suffix in the family name.
Isolates: Language without known relatives; a family with one single member.
Subgroup: A group of languages within a language family which are more closely related to each
other than to other languages of that family.
6.4 How to draw family trees: subgrouping
Languages are not only part of one family, but they are also closely related to one another within the
family [sister-languages].
Language A splits up in Language B and C, Language B splits up in B1 and B2, and Language C splits up
in language C1 and C2:
Proto-Language A
Language B Language C
Language B1 Language B2 Language C1 Language C2
The only accepted criterion for subgrouping is shared innovation: linguistic change which shows a
departure (innovation) from some traits of the proto-language and is shared by a subset of the
daughter languages.
, Chapter 17: Glottochronology
17.2.1 Basic assumptions
1) Basic vocabulary.
There is a basic vocabulary which is universal and relatively culture free and thus is less subject to
replacement than other kinds of vocabulary.
2) Constant rate of retention through time.
The rate of retention of items of core vocabulary is relatively constant through time, that a language
will retain about 86% of the words of the 100-word list reach 1000 years.
3) Constant rate of loss cross-linguistically.
Claims that the rate of loss of basic vocabulary is approximately the same for all languages (14% of
the 100-word list)
4) Calculation of the date of divergence.
When the number of ‘cognates’ in the basic vocabulary list shared by related languages is known, the
number of centuries since the languages split from an earlier ancestor can be computed formula
logC
t = ------
2Logr
t = time depth in millennia [1000 years]
C = percentage of cognates
r = retention rate [% of cognates assumed to remain after 100 years 86% of 100-word list]
log = logarithm of
17.2.3 Criticism on Glottochronology
1) Criticism on basic vocabulary
- Many of the words on the list are not culture-free
- There is not always a one-to-one matching between each word on the Swadesh list
- Some of the 100 words do not have an equivalent on the list
- Some basic vocabulary changes rapidly for cultural reasons
- Taboo has resulted in the replacement of considerable vocabulary
2) Criticism on Constant rate of retention through time and 3) Constant rate of loss cross-linguistically
- Common sense will make you doubt this, since there is no logical reason for a pattern
- Some languages extremely deviate
- There are not a lot language that have 1000 year database
4) Criticism on the calculation of the date of divergence.
- The split-up of language families (or subgroups) is not sudden, so one date is unrealistic
- Sometimes languages still have things in common that stem from the time they were already split
- The statistical model is criticised
