Students understanding the complexity of language

Chapter 15.2 Linguistic Reconstruction


The problem with studying a language like PIE is that it was spoken before the development of writing. Thus, we must “reconstruct” the language. This is done through what is called the comparative method, a process developed by early linguists such as Rask, Bopp, and Jakob Grimm (one of the famous brothers who gathered together Grimm's Fairy Tales). The comparative method, as its name suggests, is simply based on comparing different languages to find their similarities and differences. Through this comparison, we can reconstruct what their ancestor language must have looked like. Linguistic reconstruction can be an incredibly complicated task, since it requires knowledge of the phonology, morphology, and lexicon of all Indo-European languages, but the principle behind it is simple enough: the comparison of the phonology and morphology of languages.

As an example, look at the words for “father” in English and some other Germanic languages like German, Swedish, Norwegian, and Icelandic.

English father
German Vater
Swedish far
Dutch vader
Icelandic faðir
Norwegian far
Faroese faðir
Danish far
North Frisian faaðer

It should be clear immediately that all these words are very similar to each other. These are the contemporary words for “father” in the Germanic family (Scaliger’s Godt group). We call them Germanic because they are all descended from a common language spoken by the Germanic tribal peoples who lived in northern Germany and southern Scandinavia about 2500 to 2000 years ago. Although the Germanic tribes did begin writing a few inscriptions in the Runic alphabet around the first or second century CE, much of the language that they spoke was never written down. Thus, if we want to know what the Germanic language was like before the Germanic tribes broke up, we must reconstruct it from the surviving words. First, we must gather together all the words spoken in all these languages. We could use contemporary versions, like those for “father” listed above, but it’s much easier if we use the oldest versions of the words possible. In fact, we have examples that were written down only a few centuries after the Germanic tribes broke up.

Gothic faðar
Old Icelandic faðir
Old English fæder
Old Saxon fadar
Old High German fater

Note the similarities in these words. Each word begins with an [f] and ends with an [r], making it very likely that the original Proto-Germanic word also began and ended like this. Thus we can begin our hypothetical construction like this:

*f _ _ _ r

Next, the first vowel in all the words is either [a] or [æ]. Only Old English has [æ], and if we compare other words in these languages we will notice that Old English always has [æ] where the others have [a]. It is therefore most likely that [a] was the original vowel and that the [æ] is the result of a later sound change that affected English only. So we can reconstruct the first vowel as [a], giving us:

*fa _ _ r

For the medial consonant two languages have [d], two have [ð], and one has [t]. The details of reconstructing which of these sounds was the original is complex, and so I will not go through all the details here, but if we compare these sounds with what we find in other words in these languages we can see that the original sound was [ð] and that the [d] and the [t] were the products of later sound changes in English and German:


The next vowel is also difficult to explain without a much longer and more complex discussion than is necessary here, but it seems to have been [ē] (that is, a long [e]; we could also write it [eː]). Thus, the reconstructed form of the word for “father” in Proto-Germanic is


This is in fact almost exactly how we still pronounce the word today in Modern English, showing how stable languages can be even over such a long period of time.

(Note: the symbol * is used before a word or phrase that is not attested, meaning that there is no witness to it in written records. In contemporary English, this refers to word forms or phrases that we consider unacceptable, but it is also used for all reconstructed forms to signify that these forms are hypothetical and not found in any written text, i.e., there are no witnesses to them. Since Proto-Indo-European was not written down, every single PIE word should be preceded by *).

This reconstruction allows us to formulate rules of certain sound changes. Consider the [d] in the Old English word fæder and the [t] in the Old High German word fater. From this one example we can extrapolate that whenever we have a [d] in Old English, we should expect to find a [t] in German – of course this will not be true every time because of other sound changes that interfere, but it is often true. Consider the following words in Modern English and Modern High German:

English German
good gut
door Tür
day Tag
middle mittel

Observations like this are what allow linguists to figure out the sound changes that have occurred so that they can reconstruct languages. They are also what allow us to discuss how historical languages like English have changed over the years.

The last point to make about these Germanic words is that they are not different words. They are cognates, related words that spring from a common source. It is somewhat misleading to speak of the English word father as a different word from the German Vater or from Icelandic faðir. They are not different words, but the same word that now happens to be pronounced slightly differently depending on whether one lives in Iceland or Germany or the US. In fact, the pronunciation differences between these languages are probably no more varied than the differences across the English speaking world.

Now that we have reconstructed the Proto-Germanic word for father, let’s go even further back, to about 2500 B.C. when the Indo-Europeans were still living as a linguistically unified group of peoples. We can start with other words for father that you may know: Spanish padre, French père, and Italian padre. All these words descend from the Latin word pater. Latin is a well-known and well-attested language; it was spoken around the same time as Proto-Germanic, but the speakers of Latin were literate, and so we have many attestations of the word pater (notice we do not have to write it with an *).

But if you look closely at the Latin word pater and compare it with the English word father or the Proto-Germanic word *faðēr, you should see how similar the words are. They both begin with labial sounds [p] and [f], they both have a low back vowel [a], they both have a medial dental consonant, [t] or [ð], and they both end with [r]. Immediately we recognize that father and pater are cognates; they both descend from a common original. That original is Proto-Indo-European. So what was the PIE word for “father”? We can reconstruct it in the same way. First, we gather up all the earliest attested forms in the various Indo-European languages:

Latin pater
Greek patēr
Sanskrit pitar
Old Irish athair
Gothic faðar

As with the Germanic words, the similarities here should be noticeable, though less obvious. Three of the languages, the three oldest by far, have an initial [p], where the Germanic has an initial [f] and the Old Irish has no initial consonant at all. All the languages end in [r] and have a medial dental sound. Vowels are harder to reconstruct and those in this word are difficult, but linguists who study PIE have reconstructed the original root morpheme for father in PIE as *ph2tēr- (where h2 represents a sound called a laryngeal that became an [a] like vowel in many later languages).

(Note: Because the reconstruction of all PIE words are root morphemes, we usually add a hyphen (-) at the end of them, showing that any derivational or inflectional suffixes may be added to them and that we should not take these reconstructed root morphemes as actual “words”).

The examples above give a very simplified illustration of how linguistic reconstruction is done. To do this thoroughly and correctly, all the evidence must be compared, that is, all the words in all the earliest surviving Indo-European languages. For further reading, you should look at the wikipedia page on the Comparative Method.