Final Devoicing In German And Its Implications For Syntax In Related Languages


Lately, I was thinking about a problem regarding final devoicing in German and its implications for syntax in related languages. The problem that this sheds light on is the real interesting thing: it seems to be popular in many syntactic theories to refer to “determiner phrases” rather than “noun phrases,” although this analysis largely only affects West Germanic languages and some other languages with similar separate-word determiners. I think noun phrases are what languages have and not determiner phrases. I also think it’s possible there’s something even more basic than the noun but it’s not the determiner, because the “core meaning” of the phrase is not whether or not it’s determinate, however this must be returned to later.

You see, German is infamous for having final devoicing. The words Rad (wheel) and Rat (advice, counsel) cannot be distinguished at all when spoken in isolation. But that’s exactly the thing: they cannot be distinguished at all when spoken in isolation. A “true” allophone would not merge two words. As silly as it feels to even say this, speaking German does not give you some kind of ESP where if someone says [ʁaːt] you know if it’s supposed to be Rad or Rat, just like in English if someone says [lɛˑd̥] without context no one knows if it’s lead or led as they are phonetically identical. Experiments have been done that show this as well, though this is not an academic paper site and I don’t think it would be easy to link them in a Facebook post.

However, in German as opposed to English with these homophones, the difference is not only pragmatic. No, the German words are differentiated by an article: das Rad, der Rat. And when there are no distinguishing articles, they are in the plural: die Räder, die Räte, which nullifies the effect of the final devoicing because they have different plural forms. So I think this is not an allophonic process at all, even if it is a phonological rule in German that no word may ever end with a voiced consonant.

So now we get to, what does this have to do with syntax? Well, the article is the feature differentiating between the words that would otherwise be pronounced the same in the singular. So, the article is holding a lot of grammatical information despite being a different word. And the exact same thing is occurring in all West Germanic languages. I am no expert in historical linguistics, but historically all West Germanic languages have dropped sounds at the end of words and then added an article in front of the word, even if these weren’t the same sounds in all cases. And even though English only has one definite article now, “the,” in Old English this was not the case. All the articles in West Germanic languages are widely known to either be derived from pronouns, or to actually be pronouns in the current form of the language. Even English can still use this and that as either pronouns or proximal and distal determiners, though these aren’t the default like das is for neuter words in German. And in German der/die/das etc. can be used as stand-alone pronouns the same as this/that in English but not the same as the. So it appears that in West Germanic languages nominal phrases have largely assimilated to pronominal phrases due to sound changes that historically would’ve made many words ambiguous, and that’s why I don’t believe they’re really “determiner” phrases.

However, what the head of those phrases really is relies on another question in the syntax of West Germanic languages that I think is related, and that’s the so-called “helping verbs” more formally known as auxiliary verbs, which are mostly forms of to be or to have, although in Germanic languages in particular these verbal phrases can often use verbs with semantic content as the auxiliary much more often than most languages and this is also a well-known fact about them. For example, if someone steals something and it’s an actual literal crime, they don’t thieve, they commit theft. In most other language families they just thieve. One of the main things that’s actually debated in many cases is whether the main semantic verb is the head of the phrase, or whether the helping verb is the head of the phrase. I think this is typologically related to whether the “determiner” (which I think is much better described as some type of pronoun as it serves more functions than simply describing things as definite or indefinite even in the more analytic West Germanic languages such as English) is the head or whether the noun is. Is the head something with semantic content, or is it the largely-semantically-empty “helper” word which tends to bear most of the marking even in relatively analytic English and also bear relation to some phonological patterns found in all West Germanic languages? And if the head tends to be semantically empty or default rather than the word with more specific semantic content, what does this mean in general?



Question: How about “dem Rad”? Also, the phonological rule doesn’t just affect nouns, but all words.

Another thing: I’m not sure the final devoicing rule cares about generating homophones. It simply affects the realisation of a phoneme.


Answer: I do think it’s a phonological rule in German but I don’t think it’s an example of allophony which it seems to be commonly treated as because you can’t “recover” the lost information. If all you have is the word Rat/Rad in isolation you can’t generate anything because you don’t even know what the phonemes are without having more grammatical information. Of course the final devoicing rule doesn’t care about homophones, but I do think that’s why there’s the articles in German, because otherwise there would be so many homophones no one would have any idea what they’re talking about. And yes the same rule applies to other parts of speech but you also see the same kind of syntactic patterns there with all the Germanic helping verbs.


Question: I don’t understand why you feel semantics has an influence over what the head of the phrase is. Perhaps you should look into the labelling approaches developed by Chomsky and later followed up in by Rizzi this last decade.


Answer: Semantics does have an influence over what the head of the phrase is in some analyses of German and Chomsky and Rizzi don’t even study German. They pretty much just study English, not even languages English-speakers usually learn in school like German, French, and Spanish. Come on. For example, a lot of people consider German phrases with auxiliary verbs SOV.


This idea that Generative Grammar only deals with English is nonsensical and it is used by many Chomsky detractors as a straw man argument. For an historical incident (Chomsky being American), Generative Grammar started with English, but there are a lot of research within GG dealing with many different languages, not even just from the Indoeuropean family, and many key assumptions of the theory have arised from observations made in languages other than English. Sure, there are a lot of things that seem “silly” at first sight (I myself was very puzzled by stuff like “covert movement” or “abstract case”), but we have to admit that there are many very intelligent people working within that particular framework, so we might as well see what they have to say and why, and this goes the same for any other theoretical framework, be it formal or functional.


Question: Well, this is why in generative grammar the DP hypothesis is so widely accepted EVEN for languages that don’t have explicit determiners. In this theory, it is considered that many semantic and syntactic features of phrases are related to the so called functional projections. In the case of noun phrases, is the D projection what allows the NP to be an argument rather than a predicate, and this is regardless of the actual phonological realization of an explicit determiner. This is analogous to the idea that above the VP there is a Time projection that that in some languages and constructions is realized as an actual explicit T head (and auxiliary verb, for example) whereas in other languages and constructions it is not (it might be in the verb morphology, for example). So, your question is more or less what generative linguists asked themselves when they posited that the actual head of constructions are in the functional projections rather in the more “semantically full” nouns or verbs. So you might not want to call it a DP because, as you said, it is not about being definite or not, but the idea is the same: there is a functional projection that allows the the phrase to occupy some structural part of the syntax.


Answer: Yes, but it seems weird people are trying to make the heads things like “determinacy” and “time” when I think those are decidedly not what all those words are conveying. For example, in the case of “time,” most auxiliary verbs in English are not about tense. “Will” is usually about tense, but even that isn’t always about tense. “Have” and “are” are usually not about tense but aspect in English, and lots of other auxiliary verbs have nothing to do with tense such as all the modal verbs (must, should, would, could, can, etc.,) verbs in all the long Germanic auxiliary verb constructions (“He committed theft,” “I stand corrected,” etc.) Honestly English is almost tenseless compared to say Romance languages. We can inflect words for past tense but even that’s not the universal way to show they occured in the past. Whereas in Spanish you definitely don’t have constructions like “I’m going walking in a few minutes.” In Romance languages I don’t think you even “go walking,” much less use the present continuous form for any future events at all.


Next Post