TOM S JUZEK'S BLOG


<< Tost Calculator <<     –
HOME
–     >> Consent Form >>


cow.jpg
Rinderkennzeichnungs- und Rindfleischetikettierungsüberwachungsaufgabenübertragungsgesetz: A rather amusing example of German compounding. For more on this law, see [html link]. Image from Wikimedia under CC.



E-drop in German compounds*

Below are some thoughts on German e-dropping. It's a neat phenomenon, because 1) it really is underresearched and 2) it might help us understand how morpho-syntactic knowledge is organised in the mind.


It is well know that German likes its compounds. Staub-sauger ("hoover") and Staat-s-trauer ("national mourning") are almost trivial examples - it's easy to come up with more complex examples like Grund-gebuehr-erlass ("base fee waiver") or Feuer-wehr-zufahrt-s-beleuchtung ("lighting of a designated entrance for fire fighters"; something like that, in German it makes perfect sense).

You might have noticed that the examples use different compound strategies. In Staat-s-trauer, an -s has slipped in, a so called fugenlaut. Staub-sauger and Grund-gebuehr-erlass make use of the null strategy and in Feuer-wehr-zufahrt-s-beleuchtung, both strategies can be found. Other strategies include n-insertion, en-insertion, e-drop, etc. (see the references below for more).

Which strategy is chosen depends on the first constituent of a compound. For instance, for compounds with Staat as the first constiuent, it's Staats-s-versagen ("state failure"), Staat-s-krise ("state crisis"), Staat-s-trauer, etc. However, sometimes the same root uses different strategies: there is a Land-es-parlament ("federal state parliament") and a Laender-parlament (probably "federal states parliament"); arguably, there is a semantic difference. And sometimes it's not clear whether we are dealing with a fugenlaut or some other morpheme. Is the -e in Hund-e-leine ("dog leash") a fugenlaut for compounding or is it the plural morpheme plus the null strategy? N.B.: Here's an interesting asymmetry: While the second constituent is more dominant semantically (Staat-s-trauer is a more specific form of mourning), the first constituent determines a morphological question, viz. the choice of compound strategy.

Crucially, though, predicting a nouns compound strategy is non-trivial. Consider Sicht ("view") and Ansicht ("opinion", which is derived from Sicht): Sicht 's compound strategy is the null strategy, Ansicht goes with s-insertion. It is not immediately clear, why this should be the case. There might be phonological factors at play, as well as other factors like gender, genitive form, animacy, etc. At first glance, I think semantics and etymology do play a vital role here. One piece of evidence comes from this: When I thought about examples, I found it really hard to come up with a frequent word that is both a simple concrete concept (like finger, hand, tree, sun, etc.) and uses s-insertion.

Of the different strategies, there is one that caught my eye in particular: the e-drop. Some nouns ending in -e drop their final -e when being the first constituent of a compound. Sache ("question") is the prime example. Sache and Frage ("question") make Sach-_-frage ("factual issue"). However, not all words ending in -e do this. Drache ("dragon") + Kopf ("head") make Drache-n-kopf ("dragon head").^1  In fact, it is really hard to figure out which words do this and which don't: Sache drops its -e, Drache takes an -n. Bitte ("request") drops its -e, Sitte ("custom") takes the -n. Sippe ("tribe") drops its -e, Wippe ("seesaw") takes the -n, etc.

Not only is it hard to predict whether a word drops its -e or not – to my knowledge, there isn't even a resource in which this can be looked up (Duden or go to Wiktionary do not help, really). Moreover, I am not aware of a deeper analysis of this phenomenon in the academic literature. And while it seems as if native speakers have rather robust intuitions on a word's compound strategy (this needs to be tested, though), few non-linguists have this phenomenon on their radar at all. So, as with many other constructions, the e-drop is a nice reminder that morpho-syntactic knowledge is tacit.

However, e-drop is important enough to matter to industry and academia. As to industry, it matters e.g. to text-to-speech systems. Knowing that Sachfrage comes from Sache and Frage can be useful information. As to academia, the phenomenon might help us to understand more about the organisation of morphological and syntactic knowledge. Is this distributed arbitrarily and just convention? Or are there factors that drive such a choice, potentially interacting with each other in complex ways?

It is hard to formulate rules that predict e-dropping (a few of my colleagues tried - to no avail). If there are rules, they are apparently too complex to be forumlated, even after some enquiry. And probabilistic approaches are not very appealing, either: Even if we were able to formulate something along the lines of "there's a 65% that a noun drops its -e drops if the penultimate phone is +labial". This has no predictive power though: If your model gives a 60% chance of Trump winning the 2016 election, then you cannot claim that you've called it if Trump does win. Ideally, we'd have a model that can predict +/- e-drop reliably. My intuition is that to get to such a model, we have to move on from rule-based and probabilistic approaches to something else, probably machine learning.


*:  Many thanks to Jana Häussler for her feedback and for pointing me to the references.
^1:  Jana has pointed out that for her, it's fine to have ein Drachen instead of ein Drache (both Nom. sg. m.), which might or might not have some impact on its compound strategy.



REFERENCES

Donalies, E., 2011. Grammatische Variation im Deutschen - Tagtraum, Tageslicht, Tagedieb: Ein korpuslinguistisches Experiment zu variierenden Wortformen und Fugenelementen in zusammengesetzten Substantiven. Mit einem Exkurs und zahlreichen Statistiken von Noah Bubenhofer. Mannheim: IDS. (https://ids-pub.bsz-bw.de/frontdoor/index/index/docId/570)

Eisenberg, P., 2004. Das Wort.  Grundriss der deutschen Grammatik. Stuttgart: Metzler.

Fleischer, W., Barz, I., 2012. Wortbildung der deutschen Gegenwartssprache. Berlin: De Gruyter.

Fuhrhop, N., 1998. Grenzfälle morphologischer Einheiten. Tübingen: Stauffenburg.

Neef, M., 2009. IE, Germanic: German. In Rochelle Lieber & Pavol Stekauer (eds.), The Oxford handbook of compounding. Oxford: OUP, 386-399.

Neef, M., Scherer, C. (eds.), 2013 Die Schnittstelle von Morphologie und geschriebener Sprache. Berlin: De Gruyter.

Nübling, D., Szczepaniak, R., 2009. In Peter O. Müller (ed.), Religion+s+freiheit, Stabilität+s+pakt und Subjekt(+s+)pronomen: Fugenelemente als Marker phonologischer Wortgrenzen. In Studien zur Fremdwortbildung. Hildesheim: Georg Olms Verlag.



tsj; originally posted on 18 Sep 2016
last modified: 8 Mar 2017