3

Reiß and reis are two words that have the same pronunciation in standard German. So why is it that the final phoneme in each word is different? In reiß it is /s/, and in reis it is /z/.

Is there some system whereby words retain the same phonemes when they are inflected? Does it have to do with the fact that the words are not pronounced the same in different varieties of German? Or am I just entirely misunderstanding the whole thing?

ˈvʀ̩ʦl̩ˌpʀm̩ft
  • 519
  • 1
  • 4
  • 15
Jacob Lee-Hart
  • 139
  • 1
  • Your question is somewhat mis-stated, in that the phonemes in the two stems are the same in the imperative, but the underlying forms are different. "Phoneme" is not the same as "underlying form". You can tell that the underlying forms are rais and raiz from looking at the consonant as it appears before a vowel. The voiced obstruents all change to the corresponding voiceless phoneme in syllable-final position. The phoneme value is whatever the allophonic value is in the particular word, minus any allophonic changes like aspiration, release, etc. – user6726 Oct 23 '22 at 00:36
  • @user6726 How do you decide that aspiration and release are allophonic changes, but coda fortition is not? It seems rather perverse to me to assume that all syllables ending in an underlyingly lenis consonant change their phonemes when that consonant ends up in coda position. – Janus Bahs Jacquet Oct 23 '22 at 08:40
  • The criterion of complementary distribution establishes that in English. I don't know what you mean by "consonant fortition", something that we don't have in English. It seems perverse to me to deny this change at least in German. Are you opposed to phonological neutralization rules in principle? – user6726 Oct 23 '22 at 15:29
  • @user6726 Consonant fortition is the ‘devoicing’ that occurs in German in syllable coda, where lenis consonants become fortis (and voiced consonants become unvoiced). I’m not opposed to phonological neutralisation rules, but I am opposed to changing root phonemes based on syllable structure unless there’s a very good reason for it (e.g., phonemicised syncope in alternating unstressed syllables in Old Irish), and I can’t see any good reason for it here. You would say, then, that fortis /t/ ([tʰ]) becoming lenis [t] based on its position (after /s/) is an allophone of the same phoneme, but → – Janus Bahs Jacquet Oct 23 '22 at 20:18
  • → lenis /t/ ([t]) becoming fortis [tʰ] based on its position (syllable coda) is a different phoneme. I don’t see any reason for that. – Janus Bahs Jacquet Oct 23 '22 at 20:19
  • … Sorry, lenis /d/ ([t]), of course. – Janus Bahs Jacquet Oct 23 '22 at 21:27
  • @JanusBahsJacquet this specific instance of coda fortition causes the opposite of allophony: it is not a single phoneme getting multiple sounds, but multiple phonemes losing their contrast because they sound the same. If German did not have /s/, and all instances of [s] could be backtracked to /z/ being devoiced in some environment, then you could claim allophony. – lvxferre Oct 23 '22 at 21:42
  • @lvxferre But fortis /t/ becoming lenis [t] syllable-initially after /ʃ/ is generally considered allophony, despite the fact that that too is a neutralisation of multiple phonemes: contrast with /d/ is lost there too, but we don’t transcribe dein as /taɪ̯n/ just because it comes after a /ʃ/, but [wa]sch deine [Hände] is pronounced like Steine. (Obviously harder to come up with direct parallels there, since syllable-initial [ʃt] can always assumed to have /t/ because there are no circumstances where it clearly shows up as /d/.) – Janus Bahs Jacquet Oct 23 '22 at 21:57
  • @JanusBahsJacquet that's also neutralisation, not just "plain" allophony. The reason that it won't pop up often in phonemic transcription is twofold.: 1) most of the time, it's more important to represent a word in a consistent way across utterances, and this kind of process can be inferred from the differences between phonetic and phonemic transcriptions. Even then, if we are to be strict in this regard, I personally would transcribe that as /vaʃ Ta͜inə hɛndə/, with that /T/ highlighting that that environment doesn't distinguish between /t/ and /d/. 2) some linguists are sloppy with notation. – lvxferre Oct 23 '22 at 22:20
  • In terms of the difference between a phoneme and an underlying form, what is the difference in notation between these two, if there is one? And what is the difference in use? The book I am using, which transcribes reis and reiß, has not mentioned underlying form. – Jacob Lee-Hart Oct 23 '22 at 22:35
  • @lvxferre My point was that neutralisation and allophony are not mutually exclusive – allophony can cause neutralisation. Representing words/roots/forms in a consistent way across utterances is arguably the most useful feature of phonemic writing, and altering phonemes in every context that neutralises them greatly reduces the point of phonemic notation. I disagree with user6726’s statement above that phoneme and underlying form are not the same: phonemes are inherently underlying forms – they do not appear on the surface at all. Archephonemes belong two levels under the surface, and -> – Janus Bahs Jacquet Oct 23 '22 at 22:43
  • -> we don’t need three levels to explain coda fortition; two is enough. – Janus Bahs Jacquet Oct 23 '22 at 22:44
  • @JanusBahsJacquet they are mutually exclusive because allophones are associated with a single phoneme, by definition. Either [t] is an allophone for /t/, or for /d/. Phonemes and underlying representations are different things; phonemes are abstract units grouping sounds, while the underlying representation works on a morphological level, and this distinction is cross-linguistically useful (see e.g. plural //z// surfacing as /s/ in English "cats", but /z/ in "dogs"). – lvxferre Oct 23 '22 at 23:52
  • -> Coda devoicing isn't explained by those three levels (phonetic, phonemic, underlying), as it's a phenomenon on an even more basic level (the articulatory one). Those levels however are required to explain how the language handles that coda devoicing: the sound output is the same, it prevents a phonemic contrast, but there's still a morphemic distinction. (Remember, the underlying representation is mostly morphemic - that's why we need to use alternations like reis/reisen vs. reiß/reißen, to detect it). – lvxferre Oct 24 '22 at 00:01

2 Answers2

8

The idea is, humans don't memorize every individual form of reisen separately. Instead, they memorize the rules for deriving reise, reist, and all the other forms from a single stem. (Look into "the wug test" for more on how we know this.)

And for a speaker to know that reisen is pronounced with a [z] and reißen is pronounced with a [s], that information must somehow be included in the stem.

So the standard explanation is, your internal, unconscious knowledge of how the language works includes:

  • One word has the stem /rais/ and the other has the stem /raiz/
  • To form the imperative, you just use the stem on its own without any endings
  • When /z/ comes at the end of the word, you pronounce it as [s]
Draconis
  • 65,972
  • 3
  • 141
  • 215
  • 1
    The third point becomes all the clearer when you notice that this happens with all words that end in voiced sounds if they have an unvoiced counterpart: /b d g v z dʒ/ all get devoiced at the end of words to /p t k f s tʃ/ (not sure if there are actually any German roots that end in /dʒ/). – Janus Bahs Jacquet Oct 22 '22 at 23:06
  • So from a practical point of view, do the phonemes for a word stem always stay the same throughout the various inflections (as long as the root itself does not change)? Including verbs, nouns adjectives. – Jacob Lee-Hart Oct 22 '22 at 23:34
  • @JacobLee-Hart Pretty much, yeah. It's not quite that simple because there are processes like umlaut that change the phonemes inside the stem, but even then you're taking the stem and applying a predictable, regular rule to it. – Draconis Oct 22 '22 at 23:36
  • I seem to remember a study where they found an actual difference in pronunciation in (some?) final consonants in German, where a contrast like say /t/ and /d/ wasn't completely removed phonetically. – LjL Oct 23 '22 at 00:08
  • 1
    @LjL: Terminal devoicing is a regional phenomenon in Germany. It’s arguably complete in Northern German, but fades the more you go South. – ˈvʀ̩ʦl̩ˌpʀm̩ft Oct 23 '22 at 05:58
  • 1
    @LjL: It is called "incomplete neutralization". I have also seen papers about it, which I mentioned in my answer to "Does assimilation of voice produce different phonemes, or just allophones?" – brass tacks Oct 23 '22 at 07:06
4

Phonemes are defined by their contrast with other phonemes. Since German doesn't allow /z/ and /s/ contrast in word final position, we can't claim that both phonemes exist in that position; or, in fancy words, "their distinction is underspecified in that environment". So on a phonemic level both words need to be transcribed the same, as /ra͜is/ (see note)*.

The actual difference between both words in on another level - the underlying representation of their morphemes. That difference is not visible in their imperatives...

⟨reis⟩ "travel!" is //ra͜iz// → /ra͜is/. In word final position, //z// appears as the phoneme /s/.

⟨reiß⟩ "tear!", "break!" //ra͜is// → /ra͜is/. In word final position, //s// appears as the phoneme /s/.

...but it is in their infinitives:

⟨reisen⟩ "to travel" is //ra͜iz// + //ən// = //ra͜izən// → /ra͜izən/. In intervocalic position, //z// appears as /z/.

⟨reißen⟩ "to tear", "to break" is //ra͜is// + //ən// = //ra͜isən// → /ra͜isən/. In intervocalic position, //s// appears as /s/.

So it's a bit like languages pile up abstraction over abstraction: you have the raw sounds, then the phonemes, then the underlying representation.

*NOTE: how to transcribe it might change from author to author. A few authors would use /S/, an "archiphoneme", to highlight that the distinction between /s/ or /z/ doesn't exist in that position. Other authors would simply plop /s/ (as it's closer to the raw sound) and call it a day. Either way, the important bit is: if there's no phonemic distinction, you can't transcribe both differently.

lvxferre
  • 195
  • 1
  • 6
  • Is my book wrong then for transcribing the imperative 'reis!' as /ra͜iz/? Or have I misunderstood? – Jacob Lee-Hart Oct 23 '22 at 22:30
  • @JacobLee-Hart It's slightly incorrect, indeed. Probably for didactic reasons, since this difference between the underlying form vs. phoneme adds quite a bit of complexity, that might be not relevant for what the book is talking about. – lvxferre Oct 23 '22 at 23:49