SpecGram Dictionary of the Linguistics of Mythological Beasts—Volume Minus 4: The Minotaur—May Doop-Storeys SpecGram Vol CXCII, No 2 Contents A Collection of Linguistics Themed Clues—Jonathan Downie

How to Use the Comparative Method for Fun and Profit
Al Tayo-Nostradamus, Esq.

The comparative method is one of the most powerful tools ever developed by historical linguists. With the comparative method, you can take any two languages, determine whether they are related, and reconstruct their common ancestor, thus incontrovertibly cementing your reputation as the discoverer of the Italo-Turkic language family.

But enough about me.

The point is, the comparative method canbesides helping you further your scientific goals, as well as your academic and professional goals (which may or may not overlap with your scientific goals)elevate you above the masses and make you one of the linguist elite. This goal may sound lofty and difficult to achieve, but in fact, it can be done in five easy steps.

Step 1: Pick two or more languages to analyze

As the name may suggest, the comparative method involves comparing languages. The most obvious way of doing so involves selecting at least two different languages, such as English and German, as a basis for comparison. One could also compare a language to itself by comparing its modern form to one of its past varieties. These are both perfectly legitimate ways of going about doing comparative work.

However, approaching the comparative method in this way is not likely to help you win friends or influence your archaeologist colleagues into hiring you to help them find the location of the fabled Indo-Uralic Wordhoard. Is anyone really going to be impressed that you’ve been able to determine that “the hand” is cognate with die Hand or that “psychology” can be turned into psicología? Not particularly.

Instead, you should pick languages that are as different from each other as possible. This way, your reconstructions will be so unintuitive that everyone will be convinced that only a genius could have come up with them. A synchronic analysis along these lines could involve Greenlandic and Shanghainese or Albanian and Hausa. Another possibility would be to do a diachronic analysis involving different stages of the same language, such as an analysis comparing Finnish to its ancestor, Sanskrit.

Step 2: Find cognates using a Swedish list

Once you have selected the languages you want to compare, you now need to find words that are cognates in the two languages. The conventional way to do this is to use a Swedish list.1

As an example, take the first 20 words of the Swedish lists for Albanian and Hausa:

tikaiyou (singular)
jukuyou (plural)
i ciliwancanthat
shumëda yawamany

Even in this small list, we can observe several sound correspondences:

AlbanianHausaSound corre­spon­dences
Alb. në <> H. ni
Alb. #V > H. #∅
Alb. ti <> H. kai
Alb. ne <> H. mu
Alb. #jV <> H. #kV
Alb. #VC > H. #∅
Alb. to <> H. tsu
Alb. #kë <> H. #na
Alb. tu# <> H. n#
H. #iC > Alb. #∅C
Alb. #ku <> H. #na
Alb. #g <> H. #d
Alb. ii <> H. u
Alb. thë# <> H. k#
H. #wa > Alb. #∅
H. #ɗa > Alb. #di
H. nsu > Alb. sa
Alb. #p <> H. #k
H. an# (in a multi-syllable word) > Alb. ∅#
H. ɗ# > Alb. k#

There are multiple points that could be made based on this data.

First, we can see that of the 20 words originally listed, at least 10 are cognates that exhibit sound correspondences. As a result, we can see that Albanian and Hausa are 50% related and therefore closely related languages.

Second, Albanian and Hausa have roughly two sound correspondences per word, as the 10 words in the list collectively have a total of 20 correspondences. This means that Albanian and Hausa have a rating of 2 on the Cognate Divergence Index, which measures how different the cognates in two languages are based on the number of sound correspondences they exhibit. (There is, incidentally, no upper limit on how high the Cognate Divergence Index goes.)

Step 4: Reconstruct the proto-forms

Now that we have established the relatedness of Albanian and Hausa and the nature of at least some of its correspondences, we can now reconstruct the proto-forms for the cognates listed above.

Let’s start with the forms for ‘I’ and ‘where’:


An obvious initial step is to see if any of the words could function as an adequate proto-form. For ‘I’, if we assume that the proto-form was *ni and became *unë in Albanian, this would entail two processes: (1) The insertion of an epenthetic word-initial vowel /u/, and (2) a word-final vowel change from /ë/ to /i/.

If we assume that the proto-form was *unë, this would also require two processes: (1) deletion (rather than insertion) of a vowel, and (2) a word-final vowel change from /ë/ to /i/ (as with the other proto-form).

The two posited avenues of sound change (i.e. whether the proto-form is *unë or *ni) differ only in whether the word-initial vowel is deleted or inserted. How can we resolve this? Well, if we look at the forms for ‘where’, we see that there is similar data that could shed light on the matter. For the forms for ‘where’, either (1) the word-initial vowel /a/ is deleted in Hausa, or (2) the vowel /a/ is inserted word-initially in Albanian. If we assume that the word-initial vowels in both Albanian forms are the result of vowel insertion, it is unclear why the inserted vowel would be /u/ in the case of unë but /a/ in the case of ato. We could assume that, word-initially, /u/ is inserted before a nasal consonant while /t/ is inserted before a non-nasal consonant, but this assumption would require two separate rules, whereas deletion would only require one (a word-initial vowel in the proto-form is deleted in Hausa).

Thus, it is most parsimonious to assume proto-forms entailing deletion: *unë and *ato (with the difference in the second syllable explainable by a to>su sound change).

An analogous process can be applied to the other cognates to obtain the following proto-forms:


Most of the proto-forms are straightforward and reflect either the Albanian or the Hausa form, but there are a few reconstructions that require some explanation.

*Ne is assumed as the proto-form of ne and mu because assuming *mu to be the proto-form would suggest that the Albanian cognate of ku should be je, which is not the case.

*ŋkëntu is assumed as the proto-form of këtu and nan because this form would be able to explain the nasals in the Hausa form. If we assume a proto-form like *këtu, this requires positing that the obstruents /k/ and /t/ became /n/ in Hausa without any nasal conditioning environment. However, by positing a proto-form *ŋkentu, there is a more straightforward explanation: Albanian deleted the nasals in complex nasal+obstruent clusters, while Hausa deleted the obstruents (with the word-initial /n/ being the result of assimilation to the /n/ in the following syllable).

Similarly, *inka is assumed as the proto-form for ku and ina because this explains the difference in the consonants: as with ŋkëntu, Albanian deleted the nasal in the cluster, while Hausa deleted the obstruent. The rest of the derivation from the proto-form is straightforward: in Albanian, the /i/ in the first syllable triggered raising of the low /a/, resulting in the attested phoneme /u/, while in Hausa, the /i/ was deleted (as with ni and su) without causing umlaut.

Step 5: Publish your findings in a peer-reviewed journal

The final step is somewhat tedious, but it is necessary if you want to disseminate the knowledge that you have discovered with the world. Specifically, you need to find a peer-reviewed journal that is dedicated to the type of research you have done and then submit your research to the journal for publication.

You may be thinking that finding a journal that would publish an article about something as specific as Albanian-Hausa sound correspondences would be very difficult. However, one useful tip to keep in mind if all the journals you have submitted to have rejected your article is to widen your scope beyond journals about linguistics, as the reviewers in non-linguistic journals are less likely to be linguists and thus less likely to be part of cliques prejudiced against your work.

One suggestion that you may have read about is to publish your work on a blog. This option may seem attractive, as it would enable members of the general public to read your article for free, and thus enable your work to achieve wider distribution. However, this approach has a number of issues.

Blogs are often perceived as questionable avenues for publication because their lower barrier to entry means they are often used by cranks like Francis Bopp or Jacob Grim. As a result, publishing on a blog would likely impact your professional reputation. Peer-reviewed journals do not have this issue because they have a vetting process that makes it extremely unlikely for crank work (like articles promoting the Indo-Germanic Theory) to make it in, such that publishing in a journal keeps your reputation intact.

Furthermore, while wider distribution may seem like an advantage, this is, counter-intuitively, not the case. Historical linguistics is a politically contentious subject in many parts of the world, and the facts you uncover as part of your work may have the potential to upset a significant number of apple-carts. This means that it is extremely important to publish your findings in paywalled journals no one will read; if your work is inaccessible to the general public, it is virtually certain that you will be able to do research free from political interference by prejudiced non-academics.

Step 6: Celebrate

Once you have published your article, celebrate! You are well on your way to becoming a star in the booming and lucrative field of historical linguistics. Excelsior!

Step 7: Consider your next steps

And now that you have established your bona fides as a researcher, the only thing left to do is consider what to work on next. For example, what other languages are part of the language family that contains Albanian and Hausa? Clearly, further research is needed.

1 It is important to note that a Swedish list is not (as one might expect) a list of Swedes or objects related to Sweden. Rather, it is a list named after linguist Norris Swedish that contains 207 words referring to culturally universal conceptssuch as “fuchsia,” “zeitgeist”, “mother’s brother’s daughter-in-law”, “you (dual, used to refer to the two oldest people in a group of three addressees)” and “Pluto”that are unlikely to be replaced by loanwords.

SpecGram Dictionary of the Linguistics of Mythological BeastsVolume Minus 4: The MinotaurMay Doop-Storeys
A Collection of Linguistics Themed CluesJonathan Downie
SpecGram Vol CXCII, No 2 Contents