The sound patterns that a language admits in morphologically simple words often differ from those in complex words. In English, for example, no simple word ends in /md/ or /vz/, while complex ones, such as roamed, or lives do. The patterns that occur in simple words reflect language-specific phonotactic constraints, and we call them ‘phonotactic patterns’ (PTs). Patterns produced in complex words, on the other hand, are referred to as ‘morphonotactic patterns’ (MPTs). MPTs often violate constraints that PTs obey. Clearly, some patterns may be PTs in some cases and MPTs in others, such as /nd/, which is PT in wind, and MPT in sinned.
In this paper, we focus on a subset of MPTs, namely consonant clusters. They often come about through affixation (e.g. wife+s, ex+change) or compounding (black+board), but also indirectly through morphologically motivated vowel deletion (Lat spØr+ē+vi ‘despise 1sg.perf’ from spern+o ‘despise 1sg.pres’). MPT clusters reflect interactions between phonology and morphology. The traditional Neogrammarian position on such interactions was that phonological change (by sound law) was primary: it usually preceded and triggered morphological change (by analogy), which was therefore considered secondary (Hermann 1931). However, since Neogrammarian heydays many cases of primary morphological change have been found (Dressler 2002), and the ways in which phonology and morphology interact have turned out to be much more complex than envisaged (see Amdamczyk and Versloot (this issue)).
We consider evidence from various languages, and survey and classify the processes that produce and change clusters. Extending extant work in the area (Dressler et al. 2010; Baumann and Kaźmierski 2016), we compare historical developments to language acquisition and adduce evidence from language processing to arrive at explanatory accounts. In Section 2, we outline our theoretical background and our hypotheses, and introduce the concept of ‘net auditory distance’ (see also Dziubalska-Kołaczyk (this issue)), by which we assess the phonological preferability of consonant clusters. Section 3 surveys and classifies diachronic sources of clusters, Section 4 discusses phonological and psycholinguistic constraints on their emergence, and Section 5 provides a summary and an outlook.
2 Theoretical background and hypotheses
Our background is in Natural Linguistics (Dressler 1985, Dressler 1999), and specifically in Natural Phonology (Donegan and Stampe 1979; Donegan and Nathan 2015). Natural Linguistics deduces linguistic preferences from more general semiotic, cognitive or phonetic principles (Dressler 1999). To derive phonetically grounded preferences regarding consonant clusters, we use Dziubalska-Kołaczyk’s Beats-and-Binding Model (2002, 2009, Forthcoming). There, the preferability of clusters is taken to reflect ‘net auditory distances’ (NADs) between the involved consonants and their vocalic neighbours (Dziubalska-Kołaczyk 2002, and Forthcoming; Zydorowicz et al. 2016). This measure is both more comprehensive and more fine-grained than measures derived exclusively from the Sonority Sequencing principle (Jespersen 1904; Ohala and Kawasaki-Fukumori 1997; Ohala 2010; Dziubalska-Kołaczyk 2001).
Typologically, language-specific phonologies can be less or more consonantal. Their position on that scale reflects the size of their consonant inventory and the variety and complexity of the clusters they admit (cf. Maddieson 2013a; Donohue et al. 2013). On phonological grounds, clusters are the less preferred the more complex they are. Since processes that result in MPT clusters can disregard phonological preferences to some extent, MPT clusters are more likely to be phonologically dispreferred than PT clusters.
As far as morphological preference parameters (Dressler and Kilani-Schoch 2017) are concerned, the following ones are relevant for the present discussion:
Morphotactic transparency refers to the ease with which the compositionality of a word form can be inferred from its sound shape. Full morphosemantic transparency means fully compositional meaning. This is generally the case in inflection, whereas in word formation compositional meanings may become opaque when they lexicalize and their morphological patterns become unproductive. Morphological richness refers to the wealth of productive morphological patterns in a language.
In our paper, we discuss the roles that preference parameters play in the development of morphonotactic clusters. Specifically, we discuss following six hypotheses:
H1: Due to the preference for morphotactic transparency more MPT clusters will come about through concatenation than through opacifying processes such as vowel deletion (as in Lat spØr+ē+vi, see above).
H2: The complexity of the MPT clusters that arise in a language will reflect how much complexity that language admits in clusters of any type, i.e. also in PT clusters.
H3: The richer the morphology of a language is, the more MPT clusters will arise. This generalisation is limited by a typological variable (in the sense of Skalička 1979): inflecting-fusional languages allow more morphotactic opacity than agglutinating languages.
H4: Morphological richness will also increase the chance of morphotactically opaque MPT clusters to arise.
H5: MPT clusters may become PT clusters when a word loses an internal morpheme boundary. This may happen in morphosemantically opaque words that are not token frequent, unless the pattern on which they are formed is highly productive.
H6: When MPT clusters become PT clusters, the change is always lexical and diffuses in steps, if at all. An MPT pattern never becomes PT in all forms that display it. This is related to the thesis that phonological rules can become morphonological rules (in the sense of Dressler 1985, Dressler 2002) but not vice versa.
3 Diachronic sources of consonant clusters
3.1 Vowel loss
Most PT clusters come about through vowel loss. Some of them may have MPT homophones that come about in the same way and that involve concatenation (cf. H1).
Good examples are German word final triple clusters ending in /st/. They derive from schwa loss in unstressed final syllables (Werner 1978). Two PT examples are Herbst ‘fall’<Middle High German Herbest (cf. harvest), and Ernst ‘seriousness’<MHG Ernest (cf. earnest). MPT clusters of that type involve the 2sg ending of verbs, as in schimpf+st ‘scold-2sg’<MHG schimpf+est.
Schwa loss is often phonologically restricted. Before German word final sonorants it occurs only in casual speech. There, however, it also results in PT and MPT clusters, as in PT Kolb(e)n ‘club, piston’, Wes(e)n ‘essence’, 1 and MPT geb+(e)n ‘give-inf/1pl.pres/3pl.pres’, les+(e)n ‘read-inf/1pl.pres/3pl.pres’.
There are also morphological restrictions (Stopp 1974; Thoursie 1984). German schwa loss did not occur in the subjunctive present (schimpf+e+st ‘scold-subj.pres-2sg’), or in the preterite (schimpf+te+st ‘scold-pst-2sg’).
None of these schwa-deletions reduce morphotactic transparency, but in colloquial forms such as geb+n ‘give-inf/1pl.pres/3pl.pres’ and leg+n ‘lay-inf/1pl.pres/3pl.pres’ the nasals assimilate to preceding plosives, yielding [geːm] or [leːŋ], and this does reduce transparency.
In Southern German schwa reduction is blocked after nasals, as in Rahmen ‘frame’ or wohn+en ‘live-inf/1pl.pres/3pl.pres’. This prevents cluster emergence and opacification, which both occur in casual colloquial Northern German, where forms such as wohn+en are reduced to [vo:n:] or even [vo:n].
Crucially, these changes affect all forms with the relevant patterns and have no lexical exceptions. When MPT clusters lose a morphological boundary and become PT clusters, on the other hand, this affects only individual words, never whole patterns (H5 and H6). It occurs when a complex word becomes opaque, or loses its base. Examples are Kunst ‘art’, whose relation to könn+en ‘be capable’ (cf. can) is no longer identified; Oberst ‘colonel’ (cf. ober(er) ‘upper’) whose superlative ending is no longer recognized; zu+letzt ‘last’ which has lost its positive and comparative counterparts; or Kraft ‘strength’, Kluft ‘cleft’, Schaft ‘shaft’, which were derived with a -t suffix that has become opaque.
In many German dialects the prefixes be- and ge- also lost their schwa before obstruents, and ge- also lost it before sonorants. So we have Austrian [ksofn] ge+soff+en ‘drunk.pst.ptcp’, and [psofn] be+soff+en ‘drunk.adj’. Again, individual derivations may lose their boundaries yielding PT clusters as in [gmoa] Gemeinde ‘municipality’. Also, the particle and prefix zu- lost its vowel before sonorants, as in [tsruk] zurück ‘back(wards)’, [tsletst] zu+letzt ‘at last’. In [tsniaxtal] (<*zu-nicht-erl [to-naught-dim] ‘unimportant person’) the cluster has become PT through lexicalisation.
The diachrony of MPT clusters in English is similar to that of German (Baumann et al. 2015, Baumann et al. 2016; Baumann and Kaźmierski 2016), e.g. the PT cluster in OE hand (<honed) resulted from schwa loss, as did the MPT clusters in sinn+ed and the plural sin+s. Since English inflection is poorer than German inflection it created fewer MPT clusters. English has hardly any inflectional suffixes, and there is much homophony among the few that it has. For more on English phonotactics see the contributions by Baumann et al. (this issue); Honeybone, Minkova & Lefkowitz (this issue), and Schlüter (this issue).
3.1.3 Slavic languages
Also in Slavic languages, most PT and MPT clusters result from vowel loss. Their main source was the deletion (Walczak 1999: 45–46) of the two Proto-Slavic ultrashort high vowels jer (front) and jor (back) in unstressed position in the eleventh century. This deletion created novel PT clusters with no internal boundary, as in Proto-Slavic *pъtakъ>Polish ptak ‘bird’.
The deletion also created stem alternations that resulted in some opaque MPT clusters. Polish examples of root vowel deletion in oblique inflectional cases (here exemplified with the gen.sg) are, mech – mch+u ‘moss’, len – ln+u ‘linen’, wieś – ws+i ‘village’, wesz – wsz+y ‘louse’, lew – lw+a ‘lion’, bez – bz+u ‘elderberry’, kiep – kp+a ‘fool’, pień – pni+a ‘trunk’, szew – szw+u ‘stitch’, kieł – kł+a ‘fang’ (szw and kł have PT homophones).
Vowel deletion occurred also in adjectives derived from nominal bases, as in mch+owy ‘mossy’, lni+any ‘linen (adj)’, wsz+awy ‘lousy’, lw+i ‘leonine’.
Word-initial /mx/, /ln/, and /lv/ are always MPT. /ws/ and /kw/ are today also PT because some of the morphotactic operations that gave rise to them were lost.
Russian also has clusters resulting from vowel deletion, although the root vowel was retained or restored in many derivatives and compounds. The reason why vowel loss has been preserved in inflections is presumably that it is reflected in all oblique case forms, which may have created a “gang effect”. See the examples in Table 1.
In Standard Ukrainian, vowel deletion was sporadic in word initial syllables. Instead, the vowel that shows up as /o/ in oblique cases changed to /i/ in monosyllabic nominatives (Pugh and Press 2005), giving rise to alternations such as those between nominatives and genitives in riv – rov+u ‘ditch’, vil – vol+a ‘ox’, bik – bok+u ‘side’, mist – most+u ‘bridge’, etc. Thus, MPT clusters arose only in a few token-frequent words, such as pes – ps+u ‘dog’, den’ – dnj+a ‘day’, pen’ – pnj+a ‘stump’, šov – šv+a ‘stitch’. In others, such as lev – lev+a ‘lion’ analogical levelling occurred.
In Slovak, the effects of root vowel deletion have been reversed by analogical levelling, e.g. in mach – mach+u (gen) – mach+ovy (adj) ‘moss’, ľan – ľan+u (gen) – ľan+ovy (adj) ‘linen’, lev – lev+i (gen) – lev+i (adj) ‘lion’. This re-established morphotactic transparency. It has been preserved in pes – ps+y (pl) ‘dog’ – ps+í (adj) – ps+in+ec ‘dog den’ – ps+o+vod ‘dog guide’. This, we think, supports H4 that morphological richness makes morphotactically opaque MPT clusters more likely. 2 – However, most examples of root vowel deletion have been preserved in final unstressed syllables of disyllabic roots, such as mozog – mozg+u (gen) ‘brain’, laket’ – lakt’+a (gen) ‘elbow’, otec – otc+a (gen) ‘father’, list+ok – list+k+a/u (gen/dim) ‘leaf’. Since Slovak has no word-final clusters -zg, -tk, -tc, -stk, morphonotactic root vowel deletion might have been reanalysed as phonotactic vowel insertion (cf. Dressler et al. 2015; Hliničanová et al. 2017). In any case, the stem alternations create some morphotactic opacity.
3.1.4 Baltic languages
In Lithuanian, a conservative Indo-European language, the only word-final MPT clusters that are due to vowel loss are in imperatives. The imperative suffix -k goes back to the particle -ki, and lost its final vowel only during the last centuries (Stang 1966: 219; Kazlauskas 1968: 382). The loss was sporadic and affected only -ki. It occurred when the particle grammaticalised into a suffix, and produced many clusters that are exclusively MPT, such as dirb+k ‘work!’, temp+k ‘bend!’, megz+k ‘knit!’, skris+k ‘fly!’, lauž+k ‘break!’, blokš+k ‘give a blow!’, im+k ‘take!’, kel+k ‘lift!’, ar+k ‘plow!’
Word final /nk/, as in aiškin+k ‘explain!’, sodin+k ‘plant!’, on the other hand, also occurs in the mono-morphemic link ‘towards’. 3
In Latvian, which is less conservative, vowel loss in unstressed final syllables has produced many final clusters, e.g. nom.sg. cilvek+s ‘man’, ak-men+s ‘stone’, rag+s ‘horn’, pil+s ‘castle’, av+s ‘sheep’.
3.1.5 Summary of cluster emergence through vowel loss
Clusters can clearly result from vowel deletion. The German examples in particular show how vowel deletion can create both MPT and PT clusters of the same structure (as in Herbst ‘fall’ and stirb+st ‘die-2sg.pres.ind’). There are, however, two different ways in which vowel deletion can produce MPT clusters. On the one hand, it can interact with concatenation, when a vowel in a -VC(C) suffix (such as German -est ‘2sg.pres.ind’) is deleted (=Scenario 1). In clusters arising that way, a morpheme boundary occurs between stem final consonant and the (first) consonant of the suffix. On the other hand, the effect of vowel deletion can be more indirect, as in the Slavic cases. There, vowels were deleted within roots, but only before certain suffixes or in certain compounds (=Scenario 2). Although the resulting clusters have no boundary inside, they are also morphologically conditioned. What is important is that the clusters arising in scenario 2 are predictably more diverse than the ones created in scenario 1, because suffixes belong to closed classes and are phonologically less diverse than lexical roots. That scenario 2 is attested in the morphologically rich Slavic languages but not in English or German, which are morphologically poorer, supports H3, which predicts a greater diversity of MPT clusters in morphologically richer languages. At the same time, the fact that there is no language in which clusters have emerged only in scenario 2, supports H1, which predicts that on the whole more MPT clusters will reflect concatenation than other morphological operations.
Phonological root vowel deletion must not be confused with morphological root vowel deletion in Indo-European zero grade ablaut, which also created MPT clusters as in (1):
/spr/ and /str/ in the Latin perfects spØrē+vi, stØrā+vi, and perfect participles spØrē+tus, stØrā+tus (<sper+no ‘despise-1sg’, ster+no ‘strew.1sg’)
/tm/ in the Ancient Greek perfect té+tØmē-ka, the passive aorist e+tØmē+thēn, and the verbal adjective
tØmē+tós (from tém+no ‘cut-1sg’), perfect té+tmē-ka
/tl/ in Ancient Greek in the verb=tØlē+nai=talá+ssai ‘suffer, endure’, which does not occur in the present, but has two aorists
/kt/ (a metathesis of /tk/) in the Ancient Greek zero grade reduplicated present tí+kØt+o (<é+tek+on ‘I gave birth to’
Quantitative ablaut has been speculated to derive, ultimately, from phonological vowel deletion in early Proto-Indo-European (Passler 1947; Mayrhofer-Passler 1952). Thus, ablaut patterns, which are purely morphological in attested languages and in reconstructible Proto-Indo-European, might ultimately also have a phonological origin.
The same applies to corresponding Semitic root patterns. While discontinuous tri- or quadri- consonantal roots are basic, they are unpronounceable, e.g. Ar. /ktb/ ‘write’. Thus, the lexical entry has been postulated to be either the 3sg.pret kataba or its stem /katab/. This is warranted because the second vowel of the stem is unpredictable, as in the minimal triple ḥasab- ‘count’, ḥasib- ‘believe’, ḥasub- ‘be highly esteemed’ (for psycholinguistic evidence see Shimron 2002; Ravid et al. 2016). Thus, the MPT clusters in forms such as Ar. a+ktub+u ‘I write’, causative a+ktab+a ‘he dictated’, verbal noun katb, kitb+a+t ‘book’, ma+ktab ‘primary school’ are due to vowel loss. – Examples of diachronically attested vowel loss occur in modern Arabic varieties, notably in those of the Maghreb, such as in Tunisian Arabic, e.g. ktāb<kitāb ‘book’, ma+ktb+a<ma+ktb+a+t ‘library’ (Kilani-Schoch and Dressler 1985).
The synchronic analysis of “lacking vowels” in ablaut alternations may differ from diachronic vowel loss, but this holds also for many other cases of diachronic vowel loss, for which vowel epenthesis may be preferred in synchronic analyses.
Clusters also arise when purely consonantal affixes combine with root-initial or root-final consonants. Although they are related to clusters that emerge when a suffix loses a vowel (as in German schimpf+est>schimpf+st ‘scold-2sg.pres’>schimpf+st), there is a difference in the order of sound change and morphological operation: the MPT clusters in this section come about through the affixation of morphemes that are already fully consonantal at the time of their use.
In Italian, such clusters arise word-initially through prefixation, as in s+leale ‘illoyal’, s+qualificare ‘disqualify’ (Iacobini 2004: 145–146, 159). s- goes back to Latin ex-, which lost its vowel and had the remaining /ks/ simplified to /s/ ([z] with voicing assimilation, see also Pustka’s contribution to this issue). While the latter change was a regular sound law, the vowel loss was limited to the prefix. Some of the MPT clusters arising through s-prefixation have PT homophones, even complex ones like /skw/, which also occurs in words like squadra ‘team’. Others, however, are exclusively MPT, such as [zr-, zdǯ-, zñ-], as in sradicare ‘eradicate’, sgelare ‘defrost’, sgnaccare ‘to punish’. PT sl- occurs only in loan words (e.g. slow food), and the fact that it occurs at all is clearly due to the MPT model.
3.2.2 Slavic languages
In Slavic languages, consonantal prefixes were produced by the Proto-Slavic loss of the ultrashort high back vowel (Walczak 1999: 45–46, see also above). Ultimately, these prefixes may go back to Indo-European “preverbs” (Watkins 1963): preverbs were particles, which Proto-Indo-European had instead of verbal prefixes. Morphological and syntactic elements could be inserted between them and following verb stems. Slavic prefixes arose only later through univerbation, which occurred before the two Proto-Slavic prefixes lost their unstressed vowels. It represents a precondition for the emergence of the word initial MPT-clusters attested in modern Slavic languages.
In Polish, the prefixes w- ([v] or [f] by devoicing) and s-, and their combination w-s- ([fs]) have given rise to complex clusters, many of which are exclusively MPT, such as all word-initial ws- clusters and all quadruple clusters (e.g. ws+tręt [fstr-] ‘disgust’ and w+strzelać ‘to shoot in’).
Cognate Russian prefixes create word-initial MPT clusters beginning with [v] or [f]. They include twelve quadruple clusters: /vzbr/, /vzgl/, /vzgr/, /vzdr/, /fskl/, /vzdv/, /fskr/, /fspl/, /fspr/, /fstr/, /vsxl/, /vsxr/, with only few phonotactic homophones, such as /fstr/ in fstreča ‘meeting’.
Also in Ukrainian, there exist word-initial quadruple clusters such as /vzdr/ in the dialectal perfective verb v+z+driv ‘has seen’, /vpxn/ in the imperative v+pxny ‘shove sth. in!’, /vstr/ in dialectal v+striv ‘has met’, /vškv/ in v+škvaryty ‘to strike’. They are exclusively MPT.
In Slovak, the word-initial clusters /vb/, /vp/, and /vst/ as in v+bit’ ‘to hit on’, v+padnút’ ‘to fall in’, and v+stat’ ‘to get up’ are exclusively MPT, but both in Slovak and Polish the majority of MPT clusters has PT homophones. This seems to support H2, predicting that the complexity of the MPT clusters arising in a language will reflect the complexity that this language admits in general.
3.2.3 Latin, Lithuanian, Latvian
In conservative Indo-European languages word-final clusters ending in /s/ occur almost exclusively in inflected forms, e.g. in the Latin nom.sg in
/n+s/: lauda+n+s ‘praising’, fon+s ‘source’, den+s ‘tooth’, pon+s ‘bridge’ (vs. PT /ns/ in trans ‘beyond’), all with loss of the stem-final dental stop
/k+s/: pax ‘peace’, dux ‘leader’, lex ‘law’ (vs. rare PT /ks/ in sex ‘6’ mox ‘soon’)
/p+s/: inop+s ‘helpless’, pleb+s (vs. isolated PT /ps/ in abs=ab ‘away’)
After sonorants, the resulting triples are exclusively MPT, as in stirps, arx, falx, lanx.
Note that the PT cluster in trans supports H6, which predicts that changes from MPT to PT clusters diffuse lexically and in steps: trans goes back to the present participle of extinct trare (preserved in intrare ‘enter’), but has lost its boundary in lexicalisation. It is the only item in which a PT /ns/ cluster came about in that way.
Lithuanian MPT clusters (except the ones in Section 3.1.4) result from the attachment of consonantal affixes. One example is future formation, as in the 3.fut forms kep+s ‘bake’, dirb+s ‘ work’, kel+s ‘lift’, gin+s ‘defend’, or megz+s [meks] ‘knit’. 4 Other clusters occur in irregular genitive singulars, where -s is added and which tend to be replaced, in colloquial Lithuanian, by productive genitives without clusters, as in the genitives obel+s>obel+ies ‘apple tree’, moter+s>moter+ies ‘woman’, šun+s>šun+io ‘dog’, piemen+s>piemen+io ‘shepherd’ (cf. Ambrazas 2006: 79–80).
3.2.4 The typological variable of morphotactic opacity
The morphologically richer Slavic languages have more MPT clusters than the morphologically poorer Germanic and Romance languages. This supports the first part of H3, which predicts more MPT clusters for morphologically richer languages.
The second part of H3 predicts that more MPT clusters in inflecting-fusional languages than in agglutinative ones. It is supported by the small number of MPT clusters in morphologically rich agglutinating languages such as Finnish. In Finnish no MPT clusters result from affixation (ex+puoliso ‘former husband’) or compounding (syys+myrsky ‘autumn storm’), but MPT clusters are often simplified through consonant deletion or assimilation, as in the partitive last+ta of lapsi ‘child’, or the participle hakan+nut of hakata ‘to beat’ (Klaus Laalo, pers. comm.). Estonian, which is less agglutinating than Finnish (Skalička 1979), has more vowel loss and also more clusters (Skalička 1979: 308).
Agglutinating Hungarian often avoids MPT clusters by vowel insertion, as in pl. biciklis+ek ‘bicycles’, acc. Szék+et ‘chair’, loc. Pécs+ett ‘in Pécs’, iterative het+ente/nap+onta ‘every week/day’. Still, Hungarian has many PT clusters (although word-initially only in loans). It also has word-final and word-medial MPTs (Kenesei et al. 1998: 386–409). It is not clear to what extent it supports H3.
3.2.5 Summary of cluster emergence through affixation
Affixation can create diverse and complex types of MPT clusters. Among the languages we have looked at, the greatest diversity seems to be attested in Slavic languages, which are morphologically rich and allow diverse and complex PT clusters as well. This confirms our expectations. We have also shown, again, that changes of MPT clusters into PT clusters always affect only individual lexical items (H6).
3.3 Compounding (German and Lithuanian)
German is particularly rich in compounds, and compounding is a source of many German MPT clusters (Dressler et al. 2015; Dressler and Kononenko 2018). Many German compounds are formed with the interfix -s- as in König+s+hof ‘royal court, lit. ‘king+interfix+court’. This obviously increases the complexity of the word internal clusters. In this respect, the Germanic family differs from other Indo-European languages, where interfixes are typically vocalic, highly productive, often obligatory, and prevent clusters rather than increasing their complexity. An example would be the interfix -o- in words such as gas+o+meter.
Also in Lithuanian, word-internal MPT clusters arise in compounding, when the stem-final thematic vowel is deleted and the bare root appears as first part of a compound. Of the 80 MPT clusters arising in Lithuanian compounding, 46 have PT homophones and 34 are exclusively MPT (Dressler et al. 2010). Some examples are:
/bg/: sklyp + gal -i -s ‘end of a plot’ ← sklyp-a-s ‘plot’ + gal-a-s ‘end’
/tp/: rud[a] + plauk-i-s ‘brown-haired’ ←rud-a-s ‘brown’ + plauk-ai ‘hair’ (Pl.)
/tk/: led(a) + kaln-i-s ‘iceberg’ ←led-a-s ‘ice’ + kaln-a-s ‘mountain’
/gb/: pilk + balt-i-s ‘grey-white’ ←pilk-a-s ‘grey’ + balt-a-s ‘white’
/čt/: didž + turt-i-s ‘wealthy man’ ←did-i-s ‘big’ + turt-a-s ‘wealth’
/džd/: treč + dal-i-s ‘one third’ ←treči-a ‘third’ + dal-i-s ‘part’
/džg/: plač/t + gal-y-s ‘oar blade’ ←plat-u-s ‘broad’ + gal-a-s ‘end’
/fs/: gyv + sidabr-i-s ‘mercury’ ←gyv-a-s ‘lively’ + sidabr-a-s ‘silver’
/fš/: diev +š auk+i+s ‘praying person’ ←diev-a-s ‘god’ + šauk-ti ‘to call’
/šs/: kryž + snap-i-s ‘crossbill’ ←kryži-u-s ‘cross’ + snap-a-s ‘beacon’
3.4 Other sources
Consonant clusters can also result from metathesis, such as Slavic liquid metathesis (Boryś 2005), as in Polish mleko (<*melko) ‘milk’, or groch (< *gorchъ)>‘pea’.
3.4.2 Final consonant epenthesis
An exceptional case of cluster emergence occurred in the German words Axt ‘axe’ and Obst ‘fruit’, and Palast ‘palace’. The clusters result from an ‘unetymological’ addition of word final /t/, which may have prevented the final -s from being mistaken for the frequent genitive suffix -s. Also, /st/ (as in Mast ‘mast’, or Forst ‘forest’) was a much more common noun ending than /s/. Furthermore, -st was a nominalizing suffix (as in Dien+st ‘service’ from dienen ‘serve’). Thus, the addition of /t/ might have made the nouns more easily recognizable as nouns. However, this does not fully explain similar examples such as Sekt ‘sparkling wine’ (<Fr vin sec), Werft ‘shipyard’ (cf. E wharf), and dialectal Austrian German Senft (<Senf ‘mustard’). All one can observe is that the resulting word-final clusters had PT and MPT models, and that German generally allows more complex clusters at the ends of words than at their beginnings (Dressler and Kononenko 2018). This can be understood as partial support of hypothesis H2, which predicts that the complexity of the MPT clusters a language allows should correlate with the cluster complexity it allows generally i.e. also in PT clusters.
3.4.3 Internal consonant epenthesis
Internal consonant epenthesis is a common transition phenomenon. It occurs in Austrian German dialects in diminutives with the suffix /l/, as in Mandl from Mann ‘man’, or Hendl from Henne ‘hen’. /ndl/ is an MPT cluster, but when the base Henne was lost, Hendl ceased to be a diminutive and the cluster became PT. Dental epenthesis has also created the initial clusters in G Strom and E ‘stream’ (<*sroumo- ) and in Pol strumień ‘creek’. In North Mazovian dialects of Polish epenthesis produced /hɛndrɨk/=Standard /hɛnrɨk/ Henryk ‘Henry’, /tšɛrɛmpka/=Standard /tšɛremxa/ czeremcha ‘bird cherry’, /rustcɛ/=Standard /ruscɛ/ Ruskie ‘Russians (derogatory)’ (Czaplicki 2010).
Often vowel loss and consonant epenthesis combine, as in Proto-Slavic *bъčela>Pol pszczoła ‘bee’, Lat ponere ‘to put’>Fr pondre ‘lay eggs’, Lat cinerem ‘ash (acc)’>Fr cendre, Lat hominem ‘man (acc)’>Sp Hombre.
Labial epenthesis between root-final /m/ and a dental suffix has taken place in G Brunft ‘rutting season’, (An)kunft ‘arrival’, Vernunft ‘reason’, Zunft ‘guild’. When the words became morphosemantically opaque, the MPT clusters became PT.
4 Explaining the diachronic emergence and the historical stability of MPT clusters
The survey in Section 3 has been primarily descriptive, although we have related our findings to hypotheses when appropriate. In this section, we discuss predictions about specific properties to be expected from MPT clusters and about differences between MPT and PT clusters. Some predictions are derived from phonetically grounded preferences, others from psycholinguistic evidence and language acquisition.
4.1 The impact of phonological preferences (Net Auditory Distance)
Since clusters are generally not preferred, one hypothesis is that MPT clusters should be phonologically even less preferred than PT clusters. This is because (a) MPT clusters can signal boundaries, particularly when they have no PT homophones, (b) they often include consonantal morphemes, which are motivated by their morphosemantic functions. This may outweigh the articulatory and perceptual difficulties that MPT clusters create. Thus, MPT clusters ‘can afford’ to be less preferred than PT clusters, and therefore will be.
One way to assess the relative preferability of clusters is in terms of their Net Auditory Distance profile, described in Dziubalska-Kołaczyk’s contribution to this issue. Applying this measure to the MPT clusters in the languages discussed above, it turns out that the majority of MPT clusters are indeed dispreferred in Polish and Russian, but not in German and Austrian German. It is therefore premature to decide on the hypothesis. Russian and Polish may be more suitable as test cases, because their cluster inventory is greater (e.g. more than 100 word initial triple clusters in Polish, only 8 in German), and may yield more significant results. At the same time, the preferability of MPTs clusters does not seem to affect their chance of being ousted in analogical levelling.
Clearly, the complexity that a language allows in clusters is related to the complexity it allows in syllables (Maddieson 2013b; Duanmu 2008). Since clusters may span syllable boundaries, word-internal ones will be complex more often than peripheral ones. Syllable structure can also explain, at least partially, why some languages (e.g. Slavic languages) allow more complexity initially than finally and why the opposite is true of others (e.g. German): the Slavic distribution reflects a strong preference for open syllables attested already in earliest stages.
4.2 Language acquisition and psycholinguistic factors
The status of MPT clusters in language acquisition is relevant for diachrony, as (a) at least some diachronic changes may come about through ‘imperfect’, non-target-like acquisition (Dressler 1997), and (b) early acquisition supports historical stability. Moreover, (c), morphological richness stimulates and facilitates the acquisition of morphology (Xanthos et al. 2011). Here we briefly report some of our studies on the acquisition of MPT versus PT clusters and discuss what they might mean for their histories.
Freiberger (2014) investigated a longitudinal corpus of spontaneous mother-child interaction of three monolingual toddlers aged 1;7 to 3;0, acquiring German. She found a significant effect of position, but the ages at which MPT and PT clusters were acquired did not differ significantly. The same result was obtained by Korecky-Kröll et al. (2015, 2016): longitudinal spontaneous parent-child interaction data of slightly older children (3;0–5;0) showed that socio-economic status was relevant, but not the difference between MPT and PT clusters. In a comparative study, on the other hand (Zydorowicz et al. 2015), we found that Polish and Lithuanian children acquired (i.e. produced) MPT clusters significantly earlier than PT homophones, while English and German children did not do so. This once again suggests that the morphology of a language needs to be sufficiently rich, and its cluster inventory sufficiently large, for differences between MPT and PT patterns to become significant.
As far as cluster emergence through vowel loss is concerned, reduction processes can in principle occur in all age brackets and are usually insensitive to morphological boundaries. Still, acquisition may explain why diachronic vowel loss affected the verbal 2sg only in the present indicative (schimpf+st ‘scold-2sg.pres.ind’), but not in the subjunctive (schimpf+est ‘scold-2sg.pres.subj’) or the verbal past (schimpf+t+est ‘scold-pst.2sg’). Recency effects in early learning diminish over time. Thus, young children abbreviate Ger. Schokolade ‘chocolate’ as Ladi [‘la:dɪ] (recency effect), but adults as Schoko [ˈʃo:ko] (primacy effect). Since the present indicative is acquired earlier than the subjunctive and preterite, this may cause early entrenchment of reduced, vowel-less /st/. At the same time, the fact that already young children try to maintain morphotactic contrasts between morphosemantically contrasting categories may explain why they retained the vowel in Early Modern High German preterites and subjunctives.
Analogy in language acquisition may be responsible for blocking the simplification of MPT clusters in forms like haben ‘have-INF’ [ˈha:bṃ] to [ha:ṃ] . Children are likely to prefer the disyllabic pattern because it is more transparent and makes complexity easier to identify.
Experiments about processing in adolescents and adults have also yielded ambivalent results. In a letter decision task (Korecky-Kröll et al. 2014), the morpheme boundary in MPT clusters turned out to be helpful. In a fragment monitoring task (Celata et al. 2015), on the other hand, both adults and adolescents were significantly faster in detecting sequences containing PT than MPT clusters, and with respect to accuracy there were no significant differences. In a split-cluster task (Celata et al. 2015), finally, adolescents, but not adults, split significantly more MPT than PT clusters in an accurate way.
Studies addressing the lexical level show that complex words with MPT clusters are processed more slowly than simple ones with PT clusters (Freiberger et al. 2015). This supports previous findings that have identified higher processing costs in inflected word forms as opposed to monomorphemic words (e.g. Laine et al. 1999). It says little about the difference between PT and MPT clusters, however. Therefore, we investigated a domain where German morphology is rich, namely compounding (Sommer-Lolei et al. 2018). We conducted a lexical decision task that contrasted German compounds with monomorphemic nouns, both with and without clusters, e.g. Haus+tier ‘domestic animal’ vs. Tee+tasse ‘teacup’ vs. Kastanie ‘chestnut’ vs. Rakete ‘rocket’. We found a significantly higher accuracy for compounds with clusters at the boundary than for all other types of stimuli. As far as reaction time was concerned, the trend was the same, but statistically not significant. Thus, our results suggest that significant differences between MPT and PT clusters show up only in domains where the inventory of clusters that the morphology of a language produces is sufficiently rich. The results from processing experiments in these domains, however, do predict a greater diversity of historically stable MPT clusters in languages that are rich in morphology (Sommer-Lolei et al. 2018; Zydorowicz et al. 2015), and in clusters. 5
5 Summary and outlook
Most MPT clusters arise in one of two ways: phonologically, they arise through vowel loss, and morphologically, they arise through concatenation (H1). These two possibilities intersect orthogonally, especially in affixation of suffixes that have come to be consonantal or that provoke vowel loss. In contrast, the complexity of MPT clusters and their position in the word is primarily due to the phonological complexity of syllable structure (H2) and only secondarily to the richness of morphology (H3, H4). An intervening variable is the degree of morphotactic opacity that the language type allows (H3). MPT clusters may become PT clusters only via lexical change (H5, H6). This reflects the semiotic priority of the lexicon over phonology (Dressler 1985, Dressler 2002; Dressler and Kilani-Schoch 2017).
A question that our observations raise is whether MPT clusters differ from PT clusters as far as their emergence, their historical stability, or their loss are concerned: do their histories reflect, at least partly, their specific status in an area where phonology and morphology overlap and interact, i.e. do they have histories in their own right? Or do their histories represent mere epiphenomena of developments that happen, irrespectively, to sounds, on the one hand, and to morphemes on the other?
Our typological comparisons and our evidence from language acquisition and processing suggest that differences affecting the histories of MPT clusters and PT clusters are slight and difficult to detect. In languages (or subdomains such as German compounding) that are both morphologically rich and rich in consonants, however, it seems that MPT clusters may indeed acquire a status that distinguishes them from PT clusters. In such languages (or domains) evidence from typology (which both emerges from and shapes the historical evolution of languages), from acquisition (Freiberger 2014; Kamandulytė 2006; Kamandulytė-Merfeldienė 2015; Korecky-Kröll et al. 2016; Sommer-Lolei et al. 2018; Zydorowicz 2007, Zydorowicz 2010; Zydorowicz et al. 2015), and from processing converge. It suggests that a sufficiently great diversity of clusters and morphological operations seem to be required for speakers to become sensitive to systematic distinctions between MPT clusters and PT clusters, to make respective generalizations and abstractions, exploit them in learning, processing and use, and to transmit them stably across generations, thereby establishing MPT clusters in their languages.
Although our conclusions need to remain tentative, they suggest in which languages and domains further research on morphonotactic patterns and their histories promises to be particularly productive.
We thank Chiara Celata (Pisa), Klaus Laalo (Tampere), Paula Orzechowska (Poznań), Martin Peters (Vienna) and Anke Sennema (Vienna), two anonymous reviewers and the editors for their help and valuable suggestions.
Ambrazas, Vytautas (ed.). 2006. Dabartinės lietuvių kalbos gramatika [A grammar of modern Lithuanian]. Vilnius: Mokslo ir enciklopedijų leidybos institutas. Google Scholar
Baumann, Andreas & Kamil Kaźmierski. 2016. A dynamical-systems approach to the evolution of morphonotactic and lexical consonant clusters in English and Polish. Yearbook of the Poznan Linguistic Meeting 2. 115–139. CrossrefGoogle Scholar
Baumann, Andreas, Christina Prömer & Nikolaus Ritt. 2015. Identifying therapeutic changes by simulating virtual language stages: A method and its application to Middle English coda phonotactics after schwa deletion. VIEWS 24. 1–31. Google Scholar
Baumann, Andreas, Christina Prömer & Nikolaus Ritt. 2016. Diachronic dynamics of Middle English phonotactics provide evidence for analogy effects among lexical and morphonotactic consonant clusters. Papers in Historical Phonology 1. 50–75. CrossrefGoogle Scholar
Boryś, Wiesław. 2005. Słownik etymologiczny języka polskiego [Polish etymological dictionary]. Kraków: Wydawnictwo Litterackie. Google Scholar
Celata, Chiara, Katharina Korecky-Kröll, Irene Ricci & Wolfgang U. Dressler. 2015. Phonotactic processing and morpheme boundaries: Word-final /Cst/ clusters in German. Italian Journal of Linguistics 27. 85–110. Google Scholar
Czaplicki, Bartłomiej. 2010. Emergent stops in English and in Polish: Against syllable-based accounts. Poznan Studies in Contemporary Linguistics 46. 177–191. Google Scholar
Donegan, Patricia J. & Geoffrey S. Nathan. 2015. Natural phonology and sound change. In Patrick Honeybone & Joseph Salmons (eds.), Oxford handbook of historical phonology, 431–449. Oxford: Oxford University Press. Google Scholar
Donegan, Patricia J. & David Stampe. 1979. The study of natural phonology. In D. A. Dinnsen (ed.), Current approaches to phonological theory, 126–173. Bloomington: Indiana University Press. Google Scholar
Donohue, Mark, Rebecca Hetherington, James Mc Elvenny & Virginia Dawson. 2013. World phonotactics data-base. Canberra: Department of Linguistics, The Australian National University. Google Scholar
Dressler, Wolfgang & Alona Kononenko. 2018. German phonotactic vs. morphonotactic obstruent clusters: A corpuslinguistic analysis. In Wolfgang U. Dressler, Sabine Sommer-Lolei & Basilio Calderone (eds.), Experimental and acquisitional approaches to morphonotactics. Vienna: Verlag der Österreichischen Akademie der Wissenschaften. Google Scholar
Dressler, Wolfgang U. 1985. Morphonology. The dynamics of derivation. Ann Arbor: Karoma. Google Scholar
Dressler, Wolfgang U. 1997. “Scenario” as a concept for the functional explanation of language change. In Jadranka Gvozdanović (ed.), Language change and functional explanations, 109–142. Berlin: Mouton de Gruyter.Google Scholar
Dressler, Wolfgang U. 1999. On a semiotic theory of preferences in language. In Michael Shapiro (ed.), The Peirce seminar papers, vol. 4. 389–415. New York: Bergham Books. Google Scholar
Dressler, Wolfgang U. 2002. Naturalness and morphological change. In Brian D. Joseph & Richard D. Janda (eds.), The handbook of historical linguistics, 461–471. Oxford: Blackwell. Google Scholar
Dressler, Wolfgang U., Katarzyna Dziubalska-Kołaczyk & Lina Pestal. 2010. Change and variation in morphonotactics. Folia Linguistica Historica 31(1). 51–67. Google Scholar
Dressler, Wolfgang U., Miroslava Hliničanová, Matej Ďurčo, Karlheinz Mörth & Katharina Korecky-Kröll. 2015. Phonotactic vs. morphonotactic obstruent clusters in Slovak and German. Italian Journal of Linguistics 27. 45–60. Google Scholar
Dressler, Wolfgang U. & Marianne Kilani-Schoch. 2017. Natural morphology. In Andrew Hippisley & Gregory Stump (eds.), The Cambridge handbook of morphology, 356–389. Cambridge: Cambridge University Press. Google Scholar
Duanmu, San. 2008. Syllable structure: The limits of variation. Oxford: Oxford University Press. Google Scholar
Dziubalska-Kołaczyk, Katarzyna. 2001. Phonotactic constraints are preferences. In Katarzyna Dziubalska-Kołaczyk (ed.), Constraints and preferences, 69–100. Berlin: Mouton de Gruyter. Google Scholar
Dziubalska-Kołaczyk, Katarzyna. 2002. Beats-and-binding phonology. Frankfurt am Main: Peter Lang. Google Scholar
Dziubalska-Kołaczyk, Katarzyna. 2009. NP extension: B&B phonotactics. Poznan Studies in Contemporary Linguistics 45(1). 55–71. Google Scholar
Freiberger, Eva Maria. 2014. Der Erwerb von Konsonantenclustern im Deutschen: Eine Untersuchung longitudinaler Spontansprachkorpora [The acquisition of consonant clusters in German; an investigation of longitudinal spontaneous corpora]. Slowakische Zeitschrift Für Germanistik 6. 5–23. Google Scholar
Freiberger, Eva Maria, Katharina Korecky-Kröll, Basilio Calderone & Wolfgang U. Dressler. 2015. Morpheme boundaries inside consonant clusters delay visual word recognition: Evidence from a lexical decision and a progressive demasking experiment. Paper presented at the 3rd International Workshop on Phonotactics and Phonotactic Modeling (PPM 2015), Vienna, 26 November. Google Scholar
Hermann, Eduard. 1931. Lautgesetz und Analogie. Berlin: Weidmann. Google Scholar
Hliničanová, Miroslava, Matej Ďurčo, Karlheinz Mörth & Wolfgang U. Dressler. 2017. Phonotaktische versus morphonotaktische Konsonantengruppen im Slowakischen und Deutschen: Eine kontrastive korpuslinguistische Untersuchung. In Claudia Resch & Wolfgang U. Dressler (eds.), Digitale Methoden der Korpusforschung in Österreich (Veröffentlichungen zur Linguistik und Kommunikationsforschung Nr. 30), 159–178. Vienna: Verlag der Österreichischen Akademie der Wissenschaften. Google Scholar
Iacobini, Claudio. 2004. Prefissazione [Prefixation]. In Maria Grossmann & Franz Rainer (eds.), La formazione delle parole in Italiano, 69–163. Tübingen: Niemeyer. Google Scholar
Issatschenko, Alexander. 1974. Das “Schwa mobile” und “Schwa constans” im Deutschen [„Mobile schwa“ and „constant schwa“ in German]. In Ulrich Engel & Paul Grebe (eds.), Sprachsystem und Sprachgebrauch, Festschrift für Hugo Moser, vol. 1, 142–171. Düsseldorf: Pädagogischer Verlag Schwann. Google Scholar
Jespersen, Otto. 1904. Lehrbuch der Phonetik. Leipzig: Teubner. Google Scholar
Kamandulyte, Laura. 2006. The acquisition of morphonotactics in Lithuanian. Wiener Linguistische Gazette 73. 88–96. Google Scholar
Kamandulytė-Merfeldienė, Laura. 2015. Morphonotactics in L1 acquisition of Lithuanian: TD vs SLI. Eesti Rakenduslingvistika Ühingu Aastaraamat [Estonian Papers in Applied Linguistics] 11. 95–109. CrossrefGoogle Scholar
Kazlauskas, Jonas. 1968. Lietuvių kalbos istorinė gramatika [Historical grammar of the Lithuanian language]. Vilnius: Mintis. Google Scholar
Kenesei, István, Robert M. Vago & Anna Fenyvesi. 1998. Hungarian. London: Routledge. Google Scholar
Kilani-Schoch, Marianne & Wolfgang U. Dressler. 1985. Natural morphology and classical vs. Tunisian Arabic. Studia Gramatyczne 7. 27–47. Google Scholar
Korecky-Kröll, Katharina & Wolfgang U. Dressler. 2015. (Mor)phonotactics in high vs. low SES children. In Katarzyna Dziubalska-Kołaczyk & Jarosław Weckwerth (eds.), Rajendra Singh: In memoriam. Papers from a special commemorative session at the 44th Poznań Linguistic Meeting, 25–40. Poznań: Adam Mickiewicz University Press.Google Scholar
Korecky-Kröll, Katharina, Wolfgang U. Dressler, Eva Maria Freiberger, Eva Reinisch, Karlheinz Mörth & Gary Libben. 2014. Morphonotactic and phonotactic processing in German-speaking adults. Language Sciences 46. 48–58. CrossrefGoogle Scholar
Korecky-Kröll, Katharina, Eva M. Freiberger, Wolfgang U. Dressler & Basilio Calderone. 2016. Acquiring and processing German words of different (mor)phonotactic structure: Morpheme boundaries inside consonant clusters delay processing, but not acquisition. Poster presented at the 10th International Conference on the Mental Lexicon. Ottawa, 19 October. Google Scholar
Maddieson, Ian. 2013a. Consonant inventories. In Matthew S. Dryer & Martin Haspelmath (eds.), The world atlas of language structures online. Leipzig: Max Planck Institute for Evolutionary Anthropology. Available online at http://wals.info/chapter/1 accessed on 23/4/18.
Maddieson, Ian. 2013b. Syllable structure. In Matthew S. Dryer & Martin Haspelmath (eds.), The world atlas of language structures online. Leipzig: Max Planck Institute for Evolutionary Anthropology. Available online at http://wals.info/chapter/1 accessed on 1/11/18.
Mayrhofer-Passler, Edeltraut. 1952. Der Quantitätsablaut in den indogermanischen Sprachen [Quantitative apophony in the Indo-European languages]. In Wilhelm Brandenstein (ed.), Studien zur indogermanischen Grundsprache, 15–22. Wien: Gerold. Google Scholar
Ohala, John. 2010. The emergent syllable. In Barbara L. David & Krisztina Zajdó (eds.), The syllables in speech production, 179–186. New York: Lawrence Erlbaum Associates. Google Scholar
Ohala, John & Haruko Kawasaki-Fukumori. 1997. Alternatives to the sonority hierarchy for explaining segmental sequential constraints. In Stig Eliasson & Ernst Håkon Jahr (eds.), Language and its ecology: Essays in memory of Einar Haugen. Trends in linguistics. Studies and monographs, vol. 100. 343–365. Berlin: Mouton de Gruyter. Google Scholar
Passler, Edeltraut. 1947. Der indogermanische Ablaut als funktionelles Element [The Indo-European apophony as functional element]. Revue Des Études Indo-Européennes 4. 365–404. Google Scholar
Pugh, Stephan & Ian Press. 2005. Ukrainian: A comprehensive grammar. London: Routledge. Google Scholar
Ravid, Dorit, Orit Ashkenazi, Ronit Levie, G. Ben-Zadok, T. Grunwald & Steven Gillis. 2016. Foundations of the early root category: Analyses of linguistic input to Hebrew-speaking children. In Ruth A. Berman (ed.), Acquisition and development of Hebrew: From infancy to adolescence, 95–134. Amsterdam: Benjamins. Google Scholar
Shimron, Joseph (ed.). 2002. Language processing and acquisition in languages of Semitic, root-based, morphology. Amsterdam: Benjamins. Google Scholar
Skalička, Vladimír. 1979. Typologische Studien. Braunschweig: Viewegh. Google Scholar
Sommer-Lolei, Sabine, Katharina Korecky-Kröll & Wolfgang U. Dressler. 2018. Acquisition and processing of (mor)phonotactic patterns of German consonant clusters. In Wolfgang U. Dressler, Sabine Sommer-Lolei & Basilio Calderone (eds.), Experimental and acquisitional approaches to Morphonotactics. Vienna: Verlag der Österreichischen Akademie der Wissenschaften. Google Scholar
Stang, Christian. 1966. Vergleichende Grammatik der Baltischen Sprachen [Comparative grammar of the Baltic languages]. Oslo: Universitetsforlaget. Google Scholar
Stopp, Hugo. 1974. Veränderungen im System der Substantivflexion vom Althochdeutschen bis zum Neuhochdeutschen [Changes in the system of noun inflection from old high German to new high German]. In Ulrich Engel & Paul Grebe (eds.), Sprachsystem und Sprachgebrauch, Festschrift für Hugo Moser, vol. 2, 324–344. Düsseldorf: Pädagogischer Verlag Schwann. Google Scholar
Thoursie, Stig A. O. 1984. Die Verbalflexion eines südbairischen Autographs aus dem Jahr 1464, Ein Beitrag zur frühneuhochdeutschen Morphologie [Verbal inflection of a Southern Bavarian manuscript of 1464. A contribution to early New High German morphology]. Göteborg: Acta Universitatis Gothoburgensis. Google Scholar
Wagner, Monica, Valerie L. Shafer, Brett Martin & Mitchell Steinschneider. 2012. The phonotactic influence on the perception of a consonant cluster /pt/ by native English and native Polish listeners: A behavioural and event related potential (ERP) study. Brain and Language 123. 30–41. CrossrefGoogle Scholar
Walczak, Bogdan. 1999. Zarys dziejów języka polskiego. Wrocław: Wydawnictwo Uniwersytetu Wrocławskiego. Google Scholar
Watkins, Calvert. 1963. Preliminaries to a historical and comparative analysis of the syntax of the old Irish verb. Celtica 6. 1–49.Google Scholar
Werner, Otmar. 1978. Schwa-Schwund und Phonotaktik im Deutschen. In H.D. Pohl (ed.), Opuscula Slavica et linguistica: Festschrift für Alexander Issatschenko, 471–486. Klagenfurt: Heyn. Google Scholar
Xanthos, Aris, Sabine Laaha, Steven Gillis, Ursula Stephany, Ayhan Aksu-Koç, Anastasia Christofidou, Natalia Gagarina, Gordana Hrzica, F. Nihan Ketrez, Marianne Kilani-Schoch, Katharina Korecky-Kröll, Melita Kovačević, Klaus Laalo, Marijan Palmović, Barbara Pfeiler, Maria D. Voeikova & Wolfgang U. Dressler. 2011. On the role of morphological richness in the early development of noun and verb inflection. First Language 31. 461–479. CrossrefGoogle Scholar
Zydorowicz, Paulina. 2007. Polish morphonotactics in first language acquisition. Wiener Linguistische Gazette 74. 24–44. Google Scholar
Zydorowicz, Paulina. 2010. Consonant clusters across morpheme boundaries: Polish morphonotactic inventory and its acquisition. Papers and Studies in Contemporary Linguistics 46. 565–588. Google Scholar
Zydorowicz, Paulina, Laura Kamandulyte, Eva Freiberger, Katharina Korecky-Kröll & Wolfgang U. Dressler. 2015. Morphological richness facilitates first language acquisition of morphonotactic consonant clusters: Polish and Lithuanian vs. Austrian German and English. Paper presented at the 3rd International Workshop on Phonotactics and Phonotactic Modeling (PPM 2015). Vienna, 26th November. Google Scholar
Zydorowicz, Paulina, Paula Orzechowska, Michał Jankowski, Katarzyna Dziubalska-Kołaczyk, Piotr Wierzchoń & Dawid Pietrala. 2016. Phonotactics and morphonotactics of Polish and English: Description, tools and applications. Poznań: Wydawnictwo Naukowe UAM. Google Scholar
At least Wagner et al. (2012) have found that an existing cluster can be better perceived in a language rich in clusters [Polish] than in a language [English] poorer in clusters.
About the article
Published Online: 2019-07-28
Published in Print: 2019-07-26