Mekelle University

College of Social Sciences and Languages

Department of Foreign Languages and Literature
Post Graduate Program - MA in Linguistics

Course Title: Historical and Comparative Linguistics

Course Code: Ling 513
Instructor: Abinet Sime (PhD), abinetsimeg@[Link]
Credit hours: 2
Academic Year: 2019/20
Semester: Second

An assignment on: linguistic reconstruction

Submitted by: Misgina Teklemariam, mistek2006@[Link]

Id : cssl/pr/0002/12
Term paper three

Unit four: RECONSTRUCTION

Linguistic reconstruction is the practice of establishing the features of an unattested ancestor

language of one or more given languages. There are two kinds of reconstruction:

 Internal reconstruction uses irregularities in a single language to make inferences about an

earlier stage of that language – that is, it is based on evidence from that language alone.
 Comparative reconstruction, usually referred to just as reconstruction, establishes
features of the ancestor of two or more related languages, belonging to the same language
family, by means of the comparative method. A language reconstructed in this way is often
referred to as a proto-language (the common ancestor of all the languages in a given family);
examples include Proto-Indo-European, Proto-Dravidian.

Texts discussing linguistic reconstruction commonly preface reconstructed forms with

an asterisk (*) to distinguish them from attested forms.

An attested word from which a root in the proto-language is reconstructed is a reflex. More
generally, a reflex is the known derivative of an earlier form, which may be either attested or
reconstructed. Reflexes of the same source are cognates.

4.1 comparative reconstructions

It is one of the most important methods and techniques that we use to recover linguistic history
of a language. In this section we will look at the basic assumptions and limitations of the
comparative method.

In addition, the sole focus is to understand how to apply this method to reconstruct a linguistic
item. During our session of linguistic classification we say that languages which belong to the
same language family are genetically related each other.
This refers that these related languages are derived from a single original language called proto-
language. Through time, Various dialects from the proto language develop through linguistic
changes in different regions where they are spoken. The dialects keep changing and later become
distinctive languages with regard to the proto language.

The aim of reconstruction by the comparative method is to recover the ancestor language (the
proto language by doing a comparison of the descendant languages). The work reconstruction
usually begins with phonology as an attempt to reconstruct the sound system. This also leads to
reconstruction the vocabulary and the grammar of the proto language.

Every proto language was once a real language regardless of whether we are successful at
reconstructing it or not. Therefore, comparative method is a way to examine the ways which help
us to find out the language that have been derived from the proto language.

In order to have a basic sense of comparative reconstruction, let’s have these concepts.

 Proto language; it’s once spoken ancestral language from which daughter languages
come up or the languages that are reconstructed by the comparative method represent the
ancestral language from which the compared languages came out.
 Sister language: languages which are related to one another by virtue of having
descended from the common ancestor are sister languages.
 Cognate: a word or morpheme which related to a word morpheme in sister languages by
virtue of being inherited by the sister languages from a common word (morpheme) of the
proto language from which the sister languages have descended.
 Cognate set; the set of words (morphemes) which are related one another across the
sister languages because they are inherited and descended from a single word
(morpheme) of a proto language.
 Comparative method; a method or set of procedures which compares forms from related
languages and cognates which have descended from a common ancestral language or
proto language.
  Sound correspondent: a set of cognate sounds, the sounds found in the related words of
cognate sets which correspond from one related language to another because they have
descended from a common ancestral sound.
 Reflex; this a term used for the descendant in a daughter language for a sound of the
proto language that is said to be a reflex of that original sound.

Steps of comparative method

Note that, these steps are not used in a uniform way in every case, but they are really important
to understand the concept of comparative method.

Step one: assemble cognates

Step two: establish sound correspondences

Step three: reconstruct the proto sound

Step four: determine the status of similar correspondence sets

Step five: check the plausibility of the reconstructed sound from the perspective of the overall
phonological inventories of the proto language.

Step six: check the plausibility of the reconstruction sound from the perspective of linguistic
universals and typological expectations.

Step seven: reconstruct individual morphemes

Now let us briefly examine the above steps.

Step one; assemble cognates

We look for potential cognates among related languages and list them in some orderly manner or
form. See the table below.

Some cognate sets

 Italian Spanish Portuguese French Latin English
gloss

1. Capra Cabra Cabra Chẻvre capra goat

/kapra/ /kabra/ /kabra/ /ʃƐvrə/

2. Caro Caro Caro Cher caru dear

/karo/ /karo/ /karu/ /ʃƐr/

3. Capo Cabo Cabo Chef caput Head,top

/kapo/ /kabo/ /kabu/ /ʃƐf/

4. Carne Carne Carne Chair carό meat

/karne/ /karne/ /karne/ /ʃƐr/

5. Cane Can cảo Chein canis dog

/kane/ /kan/ /kảẁ/ /ʃjέ/

Step two: establish sound correspondence

In the words for goat in cognate 1 the first sound in each language corresponds in the way as
indicated in sound correspondence 1. We have to focus on the phonemic representation of the
sound than the conventional spelling.

Sound correspondence 1

Italian k-; Spanish k-; Portuguese k-; French ʃ-

Remember that historical linguistics often use the convention of a hyphen after a sound to
indicate initial position, before a sound to indicate a final and at both end to show a middle
sound. It is important to avoid potential sound correspondences which are merely due to chance.
To order to determine whether a sound correspondence such as that of (1) is real, in a sense that
it reflects sounds which are inherited in words from the proto language and are not accidental
similarity, and we need to determine whether the correspondence appears in other cognates or
not. Given that sound correspondence 1 appears frequently in the romance languages, as seen in
the forms compared in the table, we deduce that this correspondence is real.

Step three: reconstruct the proto sound

In order to get the correct proto sound we go on and on to set up other sound correspondences
and check to see that they reappear; that is, we repeat step to over and over until we will find all
the correspondence in the language being compared.

We could also go on step 3 and try to reconstruct the proto sound from which the sounds in each
of the daughter languages are descended.

We reconstruct the proto sound by postulating what the sound in the proto language most likely
was on the basis of phonetic properties of the descendant sounds in the various languages in the
correspondence set.

Linguists have put these general guide lines which they thought are most reliable for
reconstruction.

a. Directionality
b. Majority wins
c. Factoring in features held in common
d. Economy

Step four: determine the status of similar correspondence sets

Some sound changes, particularly conditioned sound changes, can result in a proto sound being
associated with more than one correspondence set. This is to be deal with to success an accurate
reconstruction.
Step five: : check the plausibility of the reconstructed sound from the perspective of the
overall phonological inventories of the proto language.

Steps 5 and 6 are related. The rule of thumb instep 5 takes advantage of the fact that languages
tend to be well behaved. That is they to have symmetrical sound systems with congruent
patterns. In step five when we consider the broader view of these sounds in the context of overall
inventories, we refine and correct our earlier proposals.

Step six: check the plausibility of the reconstruction sound from the perspective of
linguistic universals and typological expectations.

When we check our postulated reconstructions for the sounds of a proto language, we must make
sure that we are not proposing a set of sounds which is never or only rarely found in human
languages. For example, we do not find any language which has no vowels whatsoever.

Hence, a proposed reconstructed language lacking vowels would be ruled out by step 6.

There are no language with only glottalised consonants and no plain counter parts and therefore a
reconstruction which claimed that some proto language had only glottalised consonants and no
non-glottalised counterparts would be false.

Languages do not have only nasalized vowels and no non-nasalized vowels, and so we never
propose a reconstruction which would result in a proto language in which there are only
nasalized vowels.

Step seven: reconstruct individual morphemes

When we have reconstructed the proto sound from which we assume that the sounds in the sound
sets have descend, it is possible to reconstruct lexical items and grammatical morphemes.

Example,
From the cognate set for ‘goat’ in in the table above, the first sound (in sound set 1) was
reconstructed as ‘k’ based on the k: k: k: ʃ set. The second sound in the cognate for goat is ‘a’ as
in sound set 3 ( a: a: a: e). and the third sound is represented by sound set 2 (p: b: b:v) for which
we reconstructed ‘p’ and the last sound in the ‘goat’ cognates was reconstructed as ‘a’.

Putting these reconstructed sounds together based on the order they appear in the cognates for
‘goat’ in set 1 we arrive at *kapra. That is we have reconstructed a word in proto romance *kapra
‘goat’.

In this way we can continue reconstructing proto romance words for all the cognate sets based on
the sequence of sound correspondences that they reflect, building a proto romance lexicon.
Further reading on this part is available at( Campbell 1999:p 108-147)

4.2 Internal reconstruction

Unlike comparative reconstruction, which relies on the comparison of forms across related
languages, internal reconstruction relies on analysis of forms within a single language. Some
morphophonemic changes in the history of a language leave noticeable traces in its current
forms, so that by analyzing these traces we can reconstruct earlier forms of some morphemes and
the changes which led to their current forms. Comparative and internal evidence are often
compared for corroboration or refinement of a posited reconstruction of a protolanguages.

“The things that are compared in internal reconstruction, which correspond to the cognates of the
comparative method, are the forms in the language which have more than one phonological
shape in different circumstances, that is, the different allomorphs of a given morpheme, such as
those found in alternations in paradigms, derivations, stylistic variants and the like. Internal
reconstruction is frequently applied in the following situations where it can recover valuable
information: (1) to isolates (languages without known relatives); (2) to reconstructed proto-
languages; and (3) to individual languages to arrive at an earlier stage to which the comparative
method can then be applied to compare this with related languages in the family.” Campbell
1999.
The mystery of internal reconstruction is that the thought the variants (allomorphs) of a
morpheme are not all original, but that at some time in the past each morpheme had but one form
(shape) and that the variants known today have come about as the result of changes that the
language has undergone in its past.

As indicated in Campbell 1999, the steps to follow in internal reconstruction as

provided below.

Step 1: Identify alternations, that is, forms which have more than one phonological shape
(different allomorphs) in paradigms, derivations, different styles and so on.

Step 2: Postulate a single, non-alternating original form.

Step 3: Postulate the changes (usually conditioned sound changes) which must have taken place
to produce the alternating forms. (Where relevant, determine the relative chronology - the
sequence in which these changes took place.) As in the comparative method, we use all the
information at our disposal concerning directionality of change and how natural or likely (or
unexpected and unlikely) the changes we postulate are in order to evaluate the reconstruction and
the changes we propose.

Step 4: Check the results to make certain that the changes we postulated do not imply changes
for other forms that they do not in fact undergo; that is, we must guard against proposing changes
which might seem to work for certain morphemes but which, if allowed to take place, would
produce non-existent forms of other morphemes. We must also check to make certain that the
postulated reconstructions are typologically plausible and do not imply things that are impossible
or highly unlikely in human languages.

This steps are practically used together with some instants to distinguish one step from the other.
To have good stand in these steps, let’s have a look the examples and notice the variants for the
morpheme ‘I’.

(1) H-man I buy man to buy

(2) H-lap I dress lap to dress

(3) H- k’an I want k’an to want

(4) K-il I see il to see

(5) K-uʔ I drink uʔ to drink

(6) K-al I say al to say

From step 1, we realize that h-and k- are the alternant of the morpheme meaning ‘I’ h- is the
variant which occurs before consonants, and k- is the form which appears before vowels.

In step 2, we attempt to postulate the original form of the morpheme for 'I' in Pre-Tojolabal.
Three hypothesis (1) h- to change to k- before vowels to drive k- allomorph (2) k- to h- before
consonants to drive the h-variant and (3) h- before consonants and k- before vowels. Hypothesis
(3) is less likely exist. Thus, we have 1 and 2. There is no particular phonetic motivation for h- to
change into k- before vowels, as presupposed by hypothesis (1) (and if we had more data, we
would see that there are plenty of words with initial h- before a vowel). However, a change of k-
to h- before consonants is not phonetically unusual. Therefore, we assume that hypothesis (2)
with *k- is more plausible.

In step 3, we postulate that the *k which we reconstruct for 'I' in Pre-Tojolabal undergoes the
change *k- to h- before consonants and that this accounts for the h- variant of this morpheme.
So, for example, we would reconstruct *k-man 'I buy', and then the change of *k- to h- before
consonants would give modern h-man; for 'I see', however, we reconstruct *k-il, and since this k-
'I' is before a vowel, it does not change, leaving modern Tojolabal with k-il.
The internal reconstruction of words containing velar nasals in English involves a somewhat more
complicated example. In many words with a velar nasal, English has two variant forms (allomorphs), for
example:

lɔŋ/ lɔŋg:- lɔŋ ‘long’ lɔŋg-ər ‘longer’ lɔŋg-əst ‘longest’

strɔŋ/ strɔŋg strɔŋ ‘strong’ strɔŋg-ər ‘stronger’ strɔŋg-əst

English requires an assimilation rule which produces velar nasals before velar stops; the relevant portion
this rule for velars can be represented as in Rule (1):

Rule (1) (Nasal assimilation): n >ŋ /__k, g

This fact of English suggests that perhaps 'long' and 'strong' formerly ended in g and that their ŋ may also
have resulted from this nasal assimilation rule, and that later, after the assimilation to ŋ had taken place,
the final g of these morphemes was lost; this can be represented as in
Rule 2:

Rule (2) (Loss of final g): g > Ø / ŋ__#

we suggest that from the changes we made in rule 1 and 2 that can help us derive modern English.

Internal reconstruction and derivation of long and strong

Pre-English lɔŋg lɔŋg-ər strɔŋg strɔŋg-ər

Rule 1 nasal assimilation lɔŋg lɔŋgər strɔŋg strɔŋgər
Rule 2 loss of final g lɔŋ - strɔŋ -
Modern English lɔŋ lɔŋgər strɔŋ strɔŋgər

The two changes represented in Rules (1) and (2) took place in this sequence (relative
chronology, see below): first the change in Rule (1) occurred (producing ŋ before velars), and
then at some later time the change in Rule (2) occurred (deleting final g after ŋ ). It is not
possible to assume that the two changes could have taken place in the reverse chronological
order, first Rule (2) and later Rule (1), since this would give the wrong results, as illustrated in
the hypothetical but inaccurate derivation. That is, if the final g were lost first (in Rule (2)), there
would be no g left to condition the assimilation of final n to ŋ (in Rule (1)), resulting in the
wrong forms, ᵡlɔŋ, ᵡstrɔŋ.

In the example of English velar nasals, we dealt with a reconstruction which requires attention to
the sequence (or order) in which two changes took place, nasal assimilation before loss of final -
g. The identification of the sequence (temporal order) of different changes in a language is called
relative chronology. When more than one change is involved in the reconstruction, sometimes
they can each affect a form, and in such situations it may be necessary to figure out which
change or changes took place earlier and which later. There is no hard-and-fast procedure for
working out the relative chronology of the changes. However, the criterion of predictability is the
most useful - determining a chronological sequence of changes which, when applied in order to
the words of that language, does not produce any non-occurring forms. See illustrations in the
examples of ( Campbell 1999;p 207)

Campbell has mentioned limitations of internal reconstruction as follows,

First,the strongest limitation is that, while internal reconstruction is often able to recover conditioned
changes, internal reconstruction cannot recover unconditioned changes. For example, in the
unconditioned merger of *e, *o, *a to a in Sanskrit (seen in Chapter 2), these original vowels ended up as
a. If we attempt to reconstruct internally the Pre Sanskrit forms of dánta 'tooth' or dva- 'two', we find no
alternations in these vowels which would provide clues to the fact that danta originally had *e (Proto-
Indo-European *dent-, compare Latin dent-) and that
dva had *o (Proto-Indo-European *dwo-, compare Latin duo-). It is simply impossible to recover via
internal reconstruction the unconditioned change which these Sanskrit vowels underwent: if a is all we
ever see, there is no basis in Sanskrit itself for seeing anything else in the past of the a which occurs in
these words.
Second, internal reconstruction can be difficult or impossible if later changes have severely altered the
contexts which conditioned the variants that we attempt to reconstruct. For example, some splits are
impossible to recover due to subsequent changes, as illustrated by the case of voiced fricatives in English.
We observe in English such forms as breath/breathe ([br ɛθ]/ [brið]), bath/bathe ([bæθ]/ [beið]),
wreath/wreathe ([[riθ] / [rið]) which suggest an alternation between θ and ð (voiceless and voiced dental
fricative). Because we can identify alternations, we would like to be able to reconstruct a single original
form, but since in these forms both alternant can occur in exactly the same phonetic environment, we
have no basis for reconstruction. From other sources of information, however, we know that the voiced
fricatives in Old English were allophones of the voiceless fricatives in intervocalic position.

4.3 Glottochronology

Glottochronology is thus only one lexicostatistic method among many, although it is perhaps the
most controversial. It refers to the study of the rate of change in languages as measured from the
rate of replacement of "basic vocabulary" items over time.

This rate may, in turn, be used to calculate the approximate time at which two related languages
began to diverge in development. The validity of the glottochronological method rests on several
assumptions. First, that certain parts of the vocabulary are less subject to change than other parts.
The rate of retention of these basic vocabulary items is held to be constant, not only through
time, but across languages. If the percentage of true cognates within the core vocabulary is
known for any two languages, then the length of time that has elapsed since the two languages
began to diverge from a single parent language can be computed. Alternatively, one can compare
the basic vocabulary of a single language at two different stages of its development in order to
determine the rate of loss or retention.

The mechanics of the technique are relatively simple. The researcher collects word lists of the
core vocabulary, determines probable cognates within the two lists under comparison and counts
the totals. The time depth is then computed according to the formula t=logC/2log r, where t is the
divergence date in ages, C is the percentage of cognates, and r is the "glottochronological the
percentage of cognates assumed to remain after a thousand years of independent development.
The value of this constant varies according to the list being used. For Swadesh's 200—word list,
the constant is claimed to be 81%; for the 100-word list (the one used in the present study), 86%
(Swadesh 1955).

The glottochronological method which Swadesh 1955 developed has tested many times to be
sure its validity on different languages with different known linguists and found various results.
The most respected result found is its calculation of time depths as validated by
historic documentation.

In applying the glottochronological method to language data, two distinct situations are possible.
On the one hand, the languages being compared may be different historical stages on a single
line of development. This is a so—called "control case" for which the time separating the two
stages is known, as in the percentage of core vocabulary that the later stage has retained from the
earlier stage. The formula used in this case merely allows the researcher to translate the retention
percentage into a figure per one thousand years to allow for comparison with retention figures
arrived at in other control cases. On the other hand, the languages compared may be two
contemporaneous languages known to be related and to have diverged at some time in the past.
This situation constitutes an "application" of the glottochronological method. The time-depth
separating the two languages from their common source is unknown, but can be ascertained
given the percentage of cognate pairs retained in the two languages and the assumption that
languages lose morphemes in their basic vocabulary at a constant rate as determined by the
control cases mentioned above. Clearly, the validity of the application of the method depends on
the validity of the assumption of a constant rate of retention in core vocabulary, and on the extent
to which the control cases accurately measure that constant.

Consider the data collected from nine Arabic dialects first, in terms of "control cases'T and
second, as an "application" of the method to establish internal relationships among these dialects.
Arabic data have been used before, with somewhat conflicting results, in control cases to
establish the value of the retention constant. We will attempt to compare the results of our
research with previous studies and demonstrate how different methodological decisions can lead
to widely disparate results. Arabic data have not been used to compare contemporaneous
dialects, so we have no other research for comparison, only historical facts about the migration
patterns of Arabic speaking communities.

To calculate the percent of retention, in each of the dialects of Arabic, the earlier stage used for
comparison was the Classical written language. As has been noted, a methodological problem
arises since for eighteen of the words on the test—list, there are two or three synonyms in the
Classical language. We chose to consider an occurrence of any single synonym as a full retention
and ignore the fact that one or more of the synonyms had not survived as such in the dialect.
Furthermore, if a dialect retained a word cognate with one in Classical Arabic but the consultant
indicated the existence of another equally common non—cognate form used for the item, this
also was considered as a full retention. Table 1 shows the retention rate per millennium (r)
obtained by solving for r in the formula log C/log r = t, where t = 1.25 millennia and C is the
percent of cognates retained in the dialect from the Classical language.

Dialect C r

Moroccan 86 88.6
Tunisian 86 88.6
Libyan 93 94.3
Egyptian 93 94.3
Palestinian 92 93.6
Jordanian 93 94.3
Iraqi 94 95.0
Saudi 93 94.3
Yemeni 94 95.0

At first glance, it is apparent that these retention rates are, on the average, higher than those
reported by Swadesh for the 100—word list. This seems to confirm the results obtained based
on investigation of the Egyptian, Iraqi and Jordanian dialects using the 200—word list that the
rate of retention of core vocabulary in the Semitic language family is generally higher than for
those thirteen control languages originally studied by Swadesh. It should be noted that eleven of
the original thirteen control languages belong to the Indo European language family, a sample
which can hardly be considered representative of the languages of the world.

At this point, it may be useful to look more closely at how the "constant" of retention was first
calculated by Swadesh (1955). Swadesh himself was the first to point out that "what has been
called the tconstant' , but might be better called the 'index' of lexical retention" (Swadesh
1955:122) is based on extremely limited data. The original thirteen languages used as control
cases to test the 200—word list had to be reduced to seven for the 100-word list because of a
methodological problem of "overlapping histories". Thus, the final "constant" was calculated as
the arithmetic mean of the retention rates exhibited by Swedish (94.4%). German (90.0%),
English (86.2%), Rumanian (85.6%), French (65.1%), Athenian (84.8%), and Chinese (81.5%).
The range of retention rates within the control cases spans a full thirteen percentage points.
Swadesh suggests, however, that the rate for Swedish (94.4%), which is four percentage points
higher than the next highest rate, should perhaps be discounted because of evidence that it was
overestimated. He calculates a mean retention rate for the control languages including Swedish
(86.4%) and without it (85.4%). This dual calculation gave rise to a rather confusing situation,
since both 85% and 86% were reported in subsequent literature as the constant to be used for r
with the 100—word list.

To sum up about glottochronology ,it was been ased to truck the time frames of many languages
and families, from Native American languages, the languages of the Pacific, to Indo-European.
Thus, for at least the more modest goals of obtaining relative dates of separation or for looking at
languages within specific families, glottochronology seems to have proved it usefulness.

However, there are at least five important criticisms—or limitations—that might be leveled at
glottochronology, both at the level of assumptions and practice. First, the list, whether the 100-
or 200-item variety, invariably is not culturally neutral. Words for sun, moon, or certain colors,
for example, no doubt carry varying symbolic weight across languages. Second, the method
looks only at cognate lexical items and does not include an examination of pronunciation or
grammar or changes in meaning over time. Third, it ignores borrowing and language contact, and
this could easily skew results. For example, anywhere from 5% to 10% of the daily vocabulary
items used in Japanese may come from English, including words for numbers and colors. Thus, a
lexical statistical analysis might suggest that Japanese and English are more closely related than
they actually are.

Fourth, it is not clear that the rates of retention are the same for all languages. Nor are we sure
that even a single language will change at the same rate throughout its history. For example, the
rate of change of English on a 200-word list is about 76% per 1,000 years, as is Coptic.
However, German is 85%, Chinese 79%, and Greek 83%. But some languages are more
retentive, such as rural Icelandic’s 97% and Armenian’s 94%. However, there are also less
retentive languages, such as East Greenlandic, which appears to only hold 34% of its vocabulary
over 1,000 years.
References

Campbell, L. 1999. Historical linguistics. An introduction. Edinburgh university press.

Swadesh, George M. 1955 "Towards Greater Accuracy in Lexicostatistical Dating".