1. Introduction
Like the representation of affricates, the representation of s (sibilant) + consonant (sC) clusters is a topic that has seen much debate in the cross-linguistic literature (see Lin 2011 and Goad 2011 for critiques of various types of proposals). The status of affricates and sC clusters has also been a topic of discussion in accounts of the phonology of German. Mono- as well as bisegmental analyses have been proposed for the affricates [pf], [ts], and [tʃ] (see Ramers & Vater 1992:85–91 for a survey of arguments that have been put forth for both positions). The analyses of sC clusters are varied as well. Goad & Rose (2004) argue that all sC clusters in German are appendix-initial (s is extrasyllabic, an appendix to the syllable). Hall (1992) takes the position that only s + obstruent clusters are appendix initial; sC clusters with rising sonority are structured as branching onsets.Footnote 1 Wiese (1996:42–43; 265–268) argues for a different analysis of s + stop clusters, treating them not as clusters, but as complex segments.
In this squib, I provide evidence in support of Wiese’s position that s + stop sequences in German behave as complex segments. Wiese offers several arguments in favor of this analysis, only two of which deal directly with German. Reduplication in Gothic and Germanic alliteration demonstrate only that sp, st, and sk behave as single units in “historical precursors of German” (Wiese 1996:43). The evidence I provide comes from the German superlative and involves one s + stop sequence, namely, sk. I first discuss the distribution of sk in German, since this forms the basis for Wiese’s arguments and is relevant for assessing the importance of sk in the formation of the superlative. I then discuss the superlative and the evidence it provides for treating sk as a complex segment in German.Footnote 2
2. The Distribution of sk in German
The sequence /sk/, like the affricate /pf/, can appear in word-initial position before /l/, a position otherwise limited to a single consonant.
(1)
(2)
The ability of /pf/ to occur word-initially before /l/ is explained by treating it as a unit equivalent to a single consonant. Because /sk/ behaves like /pf/ in this respect, Wiese argues that it too should be treated as a single unit, a complex segment (Wiese 1996:43). This is a specific case of an s + stop sequence with characteristics similar to those of an affricate. More generally, s + stops behave like affricates in that both can typically occur in syllable-initial as well as syllable-final position.Footnote 3
(3)
(4)
If affricates are treated as single units on the basis of these distributional facts, then s + stop sequences should be as well.
Wiese also argues that an account of the phonology of German would be complicated if stops and fricatives were allowed to occupy a single prosodic position, but only in the order stop-fricative, not fricative-stop. The existence of affricates, stop-fricative combinations, leads one to expect analogous fricative-stop combinations.
Although /sk/ is found in syllable-final as well as syllable-initial position, in syllable-final position it is limited to words formed with the suffix -esk and a handful of other words that have been borrowed into German from various languages (Muthmann 2001:418).
(5)
Nevertheless, /sk/ cannot be treated as a peripheral sequence of sounds in the phonology of German.
The sequence /sk/ is much more common than its counterpart with /∫/, which is extremely rare. /∫k/ occurs in only three words, in word-initial position in the place names Schkeuditz [ˈ∫kɔɪ̯dɪts] and Schkopau [ˈ∫koːpaʊ̯], and in word-final position in the surname Waschk [va∫k] (Wiese 1996:266–267). The sequence /sk/, by contrast, is one of three s + stop sequences in German that “occur with high frequency,” the other two being /∫p/ and /∫t/ (Wiese 1996:42). Furthermore, its membership in this set is not accidental. Hall (1992:76–80) and Wiese (1996:267–268), for example, explain the unmarked combinations as governed by a rule of dissimilation. Finally, the lower frequency of /sk/ in comparison to /∫p/ and ∫t/ can be explained by considering the history of /sk/.Footnote 4 Old High German (OHG) /sk/ merged into /∫/ (<sch>) in Middle High German (MHG): OHG skif ‘ship’ > MHG schif; OHG fisc ‘fish’ > MHG visch. The /sk/ sequence in Modern German is thus relatively young, occurring in words borrowed into the language after the MHG period.
3. The Superlative Suffix
The superlative in German is formed by adding the suffix -st [st] to the stem of an adjective; in some cases a longer suffix, -est [əst], is required: kleinst- [klaɪ̯nst] ‘smallest’, lautest- [ˈlaʊ̯təst] ‘loudest’.Footnote 5 Although it is clear that the stem-final sounds of an adjective play a role in determining if -e- is required before -st, a characterization of these sounds that is both accurate and concise is not to be found in standard grammars of German. Eisenberg 2013, Durrell & Hammer 2017, and Wöllstein 2016 are representative of standard accounts. All three provide information on variation, describing environments in which -e- is optional before the superlative ending as well as environments in which it is required. In the following discussion, I consider only those environments in which -e- is required.
According to Eisenberg (2013:176), adjectives that end in a schwa syllable do not add -e- before the -st suffix (heiterst- /ˈhaɪ̯tərst/ [ˈhaɪ̯tɐst] ‘brightest’, ebenst- [ˈeːbənst] ‘flattest’, dunkelst- [ˈdʊŋkəlst] ‘darkest’); those that end in a coronal obstruent do, although -e- is optional following [∫] (zartest- [ˈtsaɐ̯təst] ‘tenderest’, blindest- [ˈblɪndəst] ‘blindest’, süßest- [ˈzyːsəst] ‘sweetest’; but raschst- [ra∫st]/raschest- [ˈra∫əst] ‘quickest’, frischst- [frɪ∫st]/frischest- [ˈfrɪ∫əst] ‘freshest’). Eisenberg’s account predicts that an adjective like brüsk [brʏsk] ‘abrupt’ will not require -e-, since it does not end in a coronal obstruent. However, brüsk requires -e-, as do all adjectives ending with the suffix -esk: brüskest- [ˈbrʏskəst] ‘most abrupt’, groteskest- [groˈtɛskəst] ‘most grotesque’. Eisenberg’s analysis also fails to account for adjectives that end in -isch, which never occur with -e-: komischst- [ˈkoːmɪ∫st] ‘strangest’ (Wöllstein 2016:374). Although he claims that -e- is optional after [∫] in monosyllabic adjectives such as rasch [ra∫] ‘quick’ and frisch [frɪ∫] ‘fresh’, I take the position that -e- is required in this environment. According to Wöllstein (2016:374), adjectives ending in [∫] (with the exception of those ending in -isch) predominantly exhibit the superlative variant with -e-.Footnote 6
Durrell & Hammer’s account is accurate, but not succinct (2017:162): Adjectives whose stem ends in -haft, -s, -sk, -ß, -x, and -z always have -est; those with a stem ending in -d, -t, and -sch usually add -est, but longer words ending in these consonants have the ending -st if the last syllable is unstressed.Footnote 7 By considering stress, Durrell & Hammer are able to account for superlatives of adjectives that end in a schwa syllable as well as those that end in -isch. Both types of adjectives end in unstressed syllables and therefore do not add schwa before -st: ˈspannendst- [ˈ∫panəntst] ‘most exciting’, ˈkomischst- [ˈkoːmɪ∫st] ‘strangest’. By including -sk in their list of stem endings that require -e-, they can also account for superlatives such as brüskest- [ˈbrʏskəst] ‘most abrupt’. However, they provide no generalization about the stem-final sounds that require the epenthesis of -e-, listing instead the orthographic symbols that represent these sounds. In addition, they must include the suffix -haft in what is otherwise a list of consonant symbols in order to account for the presence of -e- in an adjective like boshaftest- [ˈboːshaftəst] ‘nastiest’ but its absence in one such as gefürchtetst- [gəˈfʏɐ̯çtətst] ‘most dreaded’.
Wöllstein’s (2016:373–374) account is similar to Durrell & Hammer’s, but differs in several important details. According to Wöllstein, an adjective requires the superlative suffix -est if its final syllable contains a full vowel and it ends in -d, -t, -sk, or an s-sound; adjectives that end in -sch also require -est, exceptions being those formed with the suffix -isch (frischest- [ˈfrɪ∫əst] ‘freshest’, but fantastischst- [fanˈtastɪ∫st] ‘most fantastic’). By considering vowels as well as consonants, Wöllstein is able to account for adjectives formed with the suffix -haft without having to mention the suffix explicitly: -haft requires -est because it contains a full vowel and ends in -t. However, Wöllstein is forced to treat the suffix -isch as an exception, since it contains a full vowel and ends in -sch, yet does not require -e-. Like Durrell & Hammer, Wöllstein provides no generalization that unites the consonant sounds that require the epenthesis of -e-.
There are two crucial questions that these grammars fail to address: i) What is the difference between the suffixes -haft and -isch? Both contain a full vowel and end in a consonant that typically requires epenthesis of -e-, but only -haft, not -isch, occurs with -e-, and ii) What feature unites all the consonant sounds that require epenthesis of -e-?
The difference between -haft and -isch is one of stress. The suffix -haft bears secondary stress (Giegerich 1985:109); the suffix -isch is unstressed. Thus, an analysis of the superlative must take secondary stress into account: Adjectives that end in an unstressed syllable never require the epenthesis of -e-.
(6)
Those that end in a stressed syllable, with either primary or secondary stress, require -e- if they also end in a particular consonant or combination of consonants.
(7)
This leads to the second question posed above: What feature do all of these consonants have in common?
The difficulty in characterizing the consonants that require epenthesis of -e- is determining how to account for the sequence /sk/. This is the one sequence that poses a problem for Eisenberg’s position that adjectives ending in a coronal obstruent require -e-. As I argue in the preceding section, one cannot simply ignore this sequence on the grounds that it does not belong to the phonology of German. The sequence ends in /k/, which is [dorsal], yet it behaves like coronal obstruents in that it requires -e- in the superlative. In other words, adjectives with final -sk unexpectedly pattern with nass [nas] ‘wet’, which ends in [s], not with dick [dɪk] ‘fat’, which ends in [k]. The sound /k/, whether on its own at the end of an adjective or in a cluster with a preceding sonorant, does not require -e-: dickst- [dɪkst] ‘fattest’, jeckst- [jɛkst] ‘most foolish’, ungelenkst- [ˈʊngəˌlɛŋkst] ‘stiffest’, kränkst- [krɛŋkst] ‘sickest’, stärkst- [∫tɛɐ̯kst] ‘strongest’.Footnote 8 Why should it be that /sk/ requires -e-, yet /k/ by itself or in sonorant + /k/ clusters does not? If one assumes that /sk/ is a single, complex segment, the formation of the superlative can be accounted for quite simply.
The sequence /sk/ can be represented as in 8, where the manner features, [continuant] (marked as [coronal]) and [stop] (specified as [dorsal]), are attached to a single root node with the feature [−sonorant].
(8)
Because /sk/ has the features [coronal] and [−sonorant], it functions as a coronal obstruent in the formation of the superlative.
One can now account for the formation of the superlative with a simple rule of schwa epenthesis, assuming that the basic form of the superlative suffix is -st. Note that this rule is not intended as a formal analysis; it is an informal account that identifies the environments in which schwa appears in the superlative.
(9)
This rule accounts for the absence of -e- in the superlatives in 10, all of which are formed with adjectives that end in a coronal obstruent but whose final syllables are unstressed.
(10)
The rule also accounts for the presence of -e- in the superlatives in 11. These are formed with adjectives that end in a coronal obstruent and have final syllables that are stressed.
(11)
In particular, the rule accounts for the presence of -e- in superlatives of adjectives that end in /sk/—if one assumes that /sk/ is a single, complex segment.
Why should both stress and the feature [coronal] play a role in the formation of the superlative? Because the conditions on the formation of the superlative are very likely ones that hold in the language in general, a formal account in an optimality-theoretic framework, for example, could address these issues. While I do not attempt such a formal analysis here, I do offer some observations.
The superlative suffix, /st/, is itself a coronal obstruent. Thus, the epenthesis of schwa prevents the occurrence of two adjacent coronal obstruents and could be viewed as a strategy to avoid a violation of the obligatory contour principle (OCP): Adjacent identical elements are prohibited (McCarthy 1988:88).Footnote 9 However, while the OCP elsewhere in German involves the place feature [coronal], it also involves the manner feature [stop]. In present tense verb forms, for example, schwa epenthesis applies after stems ending in coronal stops (before the suffixes /st/ and /t/).
(12)
However, it does not apply after coronal fricatives.Footnote 10
(13)
As the examples in 10 demonstrate, two adjacent coronal obstruents are allowed in superlatives—schwa epenthesis is blocked—if an adjective ends in an unstressed syllable. A constraint that may play a role here is the three-syllable window restriction, which could be ranked higher than the OCP: main stress must fall on one of the last three syllables in a word (Jessen 1999:519). Because superlatives typically receive further inflection (adjective endings determined by the properties of the NP in which they occur; either /ə/, /əm/, /ən/, /ər/, or /əs/), schwa insertion in the superlative of adjectives ending in an unstressed syllable would yield forms with a stressed syllable followed by three unstressed syllables, a violation of the three-syllable window restriction.
(14)
The blocking of schwa epenthesis in these superlatives could therefore be driven by this restriction. However, superlatives are polymorphemic, and the restriction typically governs the placement of stress in mono-morphemic words. In addition, inflected forms with three consecutive unstressed syllables are acceptable elsewhere in German, as the following inflected comparative adjective forms demonstrate (Wiese 1996:276):
(15)
Although the formation of the superlative appears to be constrained by a resistance to sequences of obstruents with the same place of articulation as well as sequences of more than two unstressed syllables, it remains to be seen whether a fully articulated formal account of the superlative will involve the OCP or the three-syllable window restriction.
4. Conclusion
The argument presented here for treating /sk/ as a complex segment in German is very similar to the arguments from reduplication in Gothic and Germanic alliteration for treating s + stop sequences as complex segments. In Gothic, for example, s + stop sequences pattern like one-member onsets in the class of verbs that form the preterit through reduplication, not like other two-member onsets. The entire s + stop sequence is copied, not just s, the first member of the sequence (Wright 1910:147–148).
(16)
If one assumes that s + stop sequences in Gothic are single, complex segments, then the process of reduplication can be accounted for in a straightforward manner: The consonantal portion of a word that is copied is simply the first C (Wiese 1996:43, van de Weijer 1996:181).
A difference between the evidence offered here and that provided by Gothic reduplication and Germanic alliteration is that it involves only one of the three common s + stop sequences in German, /sk/. However, the evidence from the superlative necessarily involves consonants in coda position, and only /sk/ can provide this evidence. There are no /∫p/ sequences in coda position in German (see note 3), and the /∫t/ sequence is of no help because the stop as well as the continuant is coronal. Evidence is provided by the sequence /sp/ in the adjective krisp [krɪsp] ‘crisp’, with superlative form krispest [ˈkrɪspəst] ‘crispest’, but krisp is a loan adjective and /sp/ is rare in German, in onset as well as coda position. Thus, the evidence provided by /sk/ is crucial. It is also compelling. If /sk/ were not treated as a complex segment, with the feature [coronal] as well as [dorsal], its behavior in the formation of the superlative would be left unexplained.
