1. Introduction.

It is clear from historical sources that the coastal area of the Netherlands north of the River Maas was known as Frisia in the Middle Ages until about 1100. Whether Zeeland and Flanders were also at any moment Frisian territory is a matter of debate (a summary of scholarly opinions can be found in Krogh 1997:22–23). Since for this early period the only meaningful definition of “Frisian” is “frisophone”, the discussion has to rely on linguistic argumentation.

Toponymical evidence from the Old Dutch period and dialectal variation in the Middle and Modern Dutch periods suggest that the coastal zone of Holland was frisophone before it became Franconian-speaking after 1100. For references see Bremmer 2008, who concludes:

The part of North Holland called West Friesland remained frisophone for a longer period, possibly until the seventeenth century (Versloot 2003).

Bremmer (2008) tries to determine whether the coastal dialects of Holland, Zeeland, and Flanders were Frisian or Ingvaeonic before they switched to Franconian. He draws up a relative chronology of sound changes that the coastal dialects shared with Frisian, but that were absent from inland Franconian speech.Footnote ¹ The earlier changes he refers to as Ingvaeonic, the more recent ones (typical of West Frisian) as Frisian. Since some typically Frisian changes (WGm. ⁺e and ⁺i > iu, initial ⁺þ- > t-, medial -þþ-, -dþ- > -tt-) are occasionally found in the Old Rhine area but hardly ever further south, Bremmer (p. 293) concludes that Holland was the southernmost frisophone area. Zeeland and Flanders, in his view, did not go beyond the Ingvaeonic stage.

While I agree with Bremmer's approach to the subject, I would point out that: (1) some of the typically Frisian reflexes are also found in Flanders, such as medial -tt-; (2) some Frisian features may have been lost earlier or more completely from Flanders and Zeeland than from Holland. Both points are illustrated below.

There is no generally accepted terminology to describe the dialects spoken along the North Sea coast before the emergence of Old Saxon, Old Frisian, and Old and Middle Dutch. This is due partly to the scant linguistic data, and partly to the long period of time involved. The terms North Sea Germanic and Ingvaeonic have been used synchronically to refer to the attested dialects sharing a number of linguistic features, and diachronically to refer to a common pre-stage between West Germanic and the attested dialects (Nielsen 1994:104–109, Stiles 1995). Bremmer (2009:22, note 2) still applies these terms in both of these meanings. I prefer the term “Anglo-Frisian,” which refers to the reconstructible common ancestor of English and Frisian that underwent a number of sound changes with a relative chronology (Kortlandt 2008:270–276; partly different views in Bremmer 2009:23–35).

The coastal dialects from Flanders to Holland that became franconized may be labeled “Proto-Frisian”—a term borrowed from Bremmer 2009:22ff.Footnote ² In spite of this label, I neither claim that the whole of the coastal area was frisophone during the entire early Middle Ages nor exclude that possibility. I refer to the traces of Proto-Frisian in the later Flemish, Zeeland, and Hollandish dialects as FRISIANISMS.

2. Language Contact: Frisian and Franconian.

The Modern Dutch standard language is largely based on the (urban) dialects of Holland. Several scholars have tentatively concluded from this observation that a number of features of standard Dutch, in which it differs from the more easterly Franconian dialects of Brabant, Limburg, Gelderland, and Overijssel, are the result of language contact between coastal Frisian and inland Franconian (see, among others, Heeroma 1951, Buccini 1992, 1995, 2003; van Bree 1997, Bremmer 1997, 2008: 295–299). During the shift from Proto-Frisian to Franconian the two languages would have influenced one anther, and the resulting standard Dutch language is thought to show traces of this process.

Buccini, van Bree, and Bremmer apply the language contact theory developed by van Coetsem (1988, 2000), who distinguishes between source language (L1, usually the native language) and recipient/target language (L2), between borrowing (the transfer of a linguistic feature from L2 to L1) and imposition (the transfer of a linguistic feature from L1 to L2), and between stable and unstable elements in language. According to van Coetsem's theory, borrowing from L2 into L1 favors the adoption of unstable features, such as vocabulary and word-formational morphemes. Conversely, when imposition from L1 onto L2 takes place, stable elements, such as phonetics and syntax, are transferred more easily. Some of the data from the Frisian/Franconian contact zone seem to fit this theory quite nicely by showing stable elements of Frisian character in non-Frisian dialects. Consider, for instance, the palatal reflexes of PGm. ⁺ǣ in coastal dialects of Zeeland and Holland (compare the vocalism of sheep, summarized in Markey 1981:23), or syntactic features in North Holland which point to a Frisian substrate (Hoekstra 1993, 1994). In van Coetsem's theory, these data would suggest the imposition of features from L1 onto L2, thus supporting the alleged shift of L1 Frisian to L2 Franconian.

The most detailed application of this language contact theory to date is Buccini's (1995:45–51, 2003) explanation of the absence of i-mutation in western Dutch dialects (compare Goossens 1980:178ff, 2008:63–85). Buccini assumes that Proto-Frisian had already phonologized i-mutation by 1000 and reduced vowels in unstressed syllables (that is, the conditioning factors) to schwa. In Franconian, however, the fronted vowels were still only allophones conditioned by front vowels in the next syllable. Frisophone speakers then adopted the Franconian words but without i-mutated vowels and without the relevant vowel distinctions in unstressed syllables. In other words, native speakers of L1 Frisian imposed their own vowel system on the pronunciation of their newly acquired L2 Franconian. This is the most attractive solution for the problem of i-mutation in coastal Dutch proposed so far.

Additional research will have to establish whether other cases of linguistic variation between coastal and inland Dutch can be explained in the same way. It will be necessary to collect all relevant data on variation between the coastal dialects and their eastern neighbors, and to interpret each phenomenon in terms of language contact theory. In the present contribution, I have the modest objective of adding one lexeme to the collection of possible Frisianisms in the coastal dialects.

3. Dutch ie, ee instead of aa.

One of the Anglo-Frisian phenomena in coastal Dutch is the reflex of Proto-Germanic ⁺ǣ, traditionally called ⁺ē₁. It yields aa /a./ in standard Dutch, but some words in coastal Dutch show a high front vowel ie /i(.)/, in some sources also ee /e./.Footnote ³ A small collection of words containing ie instead of aa is given in van Loey 1970:§80:

• • North-Hollandish deek, diek, South-Hollandish deek versus Utrecht and Overijssel daak, daok ‘washed-up rubbish, driftwood’ (Opprel 1936) < ⁺þǣka- ‘cover’; compare Modern West Frisian toponyms De Tike, Tiekelân (Versloot 2001b), East-Frisian teek, North-Frisian Föhr tiak, Amrum siak, Sylt teeki ‘washed-up seaweed’.

• • Dutch gier (North-Hollandish ier, Groningen ier, jier, and so on) versus Middle Low German gâre ‘manuring’ (see TNZN vol. 1, map 7) < ⁺jǣzō-; compare MoWF jarre ‘liquid manure’.

• • Middle Dutch lieke, leke, Kiliaan lijcklaecke, West-Flemish lieklake, East-Flemish lieke, lake, versus Middle Dutch lake, Modern Dutch laecke (Kiliaan), laak ‘leech’ < ⁺lǣkja-; see van Sterkenburg 1975: 224–227 with dialect map, de Tollenaere 1997:209f; compare OFri. lē(t)za, Gothic lēkeis ‘physician’.

• • Coastal Dutch mede (meet) ‘mowing field, meadow’ in toponyms versus standard Dutch made, maat < ⁺mǣdwō-; compare Old Frisian mēde, MoWF miede, Old English mǣd.

• • Middle Dutch mielen ‘to paint’, mielre ‘painter’ versus Early Modern Dutch malen. MoWF miel(j)e, Gothic mēljan.

• • Standard Dutch schriel ‘lean’ versus schraal ‘thin, frugal’ < ⁺skrǣla-. The standard language has adopted both words but with differenttiated meanings; compare MoWF skriel; Middle Dutch, Middle Low German schrā ‘thin, lean’, Old Icelandic skrá ‘piece of dry skin or parchment, scroll’.

• • Standard Dutch wiel, Flemish-Zeelandish-Hollandish weel, versus standard Dutch waal ‘pool (after a dike has collapsed)’; compare OF wēl, MoWF wiel, Old English wœ̄l.

Words absent from standard Dutch usually have a restricted distribution in the coastal dialects of Flanders, Zeeland, and Holland, but it is not necessarily the case that forms in -ee- or -ie- are found more frequently in the north than in the south. The toponym mead, for instance, is found well into Flanders (Rentenaar 1972, van Loon 1986:41). The thirteenth-century attestations of this word mapped in Mooijaart 1992:114 show 68% of e(e)-spellings in West Flanders, 33% in East Flanders, 20% at the count's court in Holland, and zero in the remainder of Holland. Even if these numbers are partly due to the unequal distribution of written sources from the period (the majority of them being Flemish), they still show that this particular Frisianism was found all along the coastal zone in the Middle Ages.

The raised front vowel of coastal Dutch is similar to the Frisian reflex ie /iə/, with monophthongization in certain environments also ii /i./ or y, i /i/ (Hoekstra 2001:723). The West Germanic vowel is traditionally reconstructed as ⁺ā, but I follow van Wijk (1911:161ff), Buccini (1992: 407), and Kortlandt (2008:266), who assume a West Germanic realization ⁺ǣ which was backed in inland Franconian and Saxon but raised in Anglo-Frisian, as shown in 1.

1. (1)

The coastal Dutch reflex ie /i(.)/ must reflect a similar raising of Anglo-Frisian ⁺ē to that witnessed in Modern West Frisian. It could be maintained that the similarity in vocalic development between Frisian and coastal Dutch was accidental or caused by systemic pressures independently at work in both areas. Alternatively, one could argue that these coastal Dutch words in /i./ actually reflect Frisian words, which remained in use when the speakers of coastal dialects switched from Proto-Frisian to Franconian. Whether the final stage of the raising (from ⁺/e./ to /iə/ or /i./) took place before or after the franconization remains to be determined.

4. Dutch Traam, Treem, Triem: Dialect Geography.

I would like to propose another member of this group of words in WGm. ⁺ǣ, namely Dutch traam, treem or triem ‘beam used for joining or supporting’. It has all but disappeared from the vocabulary of present-day speakers, but it is found with a variety of technical meanings in the traditional dialects and in older literature. WNT actually has two entries, traam and treem, but the meanings of these words largely overlap: ‘1. Rung (of a ladder). 2. Crossbar (of a chair). 3. Shaft of a wheelbarrow, a stretcher or a bier. 4. Stile of a ladder. 5. Wooden ledges between which the flour trough in a gristmill is suspended. 6. Tooth of a comb or rake. 7. Beam (in a loft).’ The semantic overlap suggests that we are dealing with one original word, and this is confirmed by dialect geography. The latter also points to a possibility that triem is a Frisianism in coastal Dutch.

4.1. The ‘Shaft of a Wheelbarrow’.

Detailed dialect-geographical data are available for the meanings ‘rung of a ladder’ and ‘shaft of a wheelbarrow’ from TNZN and from RND. I have drawn three maps on the basis of the transcriptions found in RND, using the program Kloeketabel (available at the website of the Meertens Institute, www.meertens.knaw.nl).

Map 1.

The vocalism of trVm(ə)(n) ‘shaft of a wheelbarrow’ according to RND.

Map 2.

Zooming in on the West-Flemish vocalism.

The word trVm(ə)(n) with the meaning ‘shaft of a wheelbarrow’ occurs in sentence 129 of the RND. Relevant variants of the form trVm- are found in Flanders, Zeeland, South Holland, and a few adjacent areas.Footnote ⁴ Flanders shows a tripartite division. French Flanders and the province of West Flanders have a back vowel, rendered by /a./ on map 1. More exact vowel transcriptions for this area are given on map 2: /a./, /ɔ./ (the notation /ò./ on the map is due to cartographic restrictions) and /o./ (actually /ʊ./ in many dialects). These vowels may be reflexes of lengthened short ⁺a (as in vader) or old long ⁺ā (<WGm. ⁺ǣ), see Taeldeman 1979:62. The south and east of East Flanders mainly have /e./ from ⁺e or ⁺ä in open syllable. Map 1 also shows a wedge-shaped area with /i(.)/ in the western part of East Flanders, as well as in large parts of Zeeland Flanders. This variant is sporadically found further north up to the islands of South Holland and in western North Brabant.

A deviating form in initial p- /prim/ is dominant in South Holland, and stretches to adjacent parts of North Holland and Utrecht. In Zeeland, /pri(.)m/ competes with /tri(.)m/, and is found down to Walcheren. A few isolated occurrences of pr- appear at the other end of the Netherlands—once as /praimn/ (with regular diphthongization of /i./ in this dialect) in eastern Groningen, and as /pri.mən/ ‘shafts of a peat cutter's wheelbarrow’ in two towns in southeastern Brabant (Crompvoets 1991:331).

Since trVm occurs both north (not on map 1, but see sections 4.2 and 4.3 below) and south of the /prim/-area, it is likely that /prim/ replaced /trim/ in South Holland relatively recently, and from there spread to Zeeland. This replacement seems surprising at first sight, given the meaning of standard Dutch priem ‘awl, piercer’. However, WNT provides for priem the submeanings ‘small stick or rod, peg’ (already found in Kiliaan) and ‘kind of beam’. In a nineteenth-century dictionary of engineering (Pijtak 1848), priemen refers to two long, parallel beams in the construction of a lift bridge. This usage comes quite close to the two shafts of a wheelbarrow or the two stiles of a ladder. We may, therefore, assume that the motivation for the replacement was a semantic association of priem with triem. Thus, at an earlier stage, there must have been a continuous /trim/-area from Flanders in the south to North Holland and Friesland in the north.

4.2. The ‘Rung of a Ladder’.

The meaning ‘rung of a ladder’ was investigated in 1943 for TNZN (issue 3, map nr. 8). The map shows a plural /trimən/ in all of Friesland and also in adjacent parts of Groningen (Westerkwartier) and Overijssel, and /tre.m(ə)n/ in the southwestern part of Drenthe. The singular rung was also elicited in sentence 83 of RND, see map 3. It shows several vowel distinctions within Friesland which are ignored on the map in TNZN, but which matter for the etymology of this word. We find trime /trimə/ in the southern and western parts of Friesland (as in standard Frisian) against northern trym /trim/ with apocope, a distribution which is also found in other words in Old Frisian -e (see Versloot 2008:136, 172, maps 2.8 and 2.12). A variant triem /triəm/ occurs in some northeastern dialects (on Ameland, in Foudgum, Birdaard, Zwaagwesteinde, Surhuisterveen) and is probably an archaism (see section 6.1). In the MoWF standard, triem(e) not only means ‘rung’ but also ‘row, long list’, which can easily be connected with ‘ladder’—compare Dutch scala, German Skala ‘range, gamut’.

The same noun trVm was also used for ‘rung’ in formerly frisophone North Holland, as is shown by four relic attestations captured in TNZN (twice /trimə/, twice /tremə/), but twenty years later, by the time of the RND inquiry (around 1960), it had apparently disappeared from that area.

Map 3.

The meaning rung (of a ladder) in and around Friesland according to RND.

4.3. Dialect Data from Other Sources.

Some additional evidence for the distribution of trVm can be found in other dialect sources. In North Holland, the meaning crossbar of a chair is reported as triem in Zaankant (Boekenoogen & Woudt 2004:537), Volendam and Monnikendam (van Ginneken 1954 II:173), and West-Friesland (Karsten 1931:157, Pannekeet 1984:364), and as treem in Marken (van Ginneken 1954 II:173) and in Enkhuizen (Spoelstra 1981:112). According to WNT, some dialects have a short vowel: trem (Marken, Middelie, Beemster) or trim (Middelie, Warder). In Oud-Beijerland in South Holland, trem may mean ‘crossbar of a chair’, ‘spoke of a wheel’, or ‘rung’.

The meaning tooth of a rake is less common, but can already be found in the Late Middle Dutch dictionary Teuthonista (1477): gavel van vyer of vyf tramen ‘a rake with four or five teeth’. This dictionary reflects the speech of the Lower Rhine area between Venlo, Xanten, and Cleves, and the Cleves dialect still had [tra:m] ‘tooth of a rake’ around 1900 (Rheinisches Wörterbuch). In the Lower Betuwe, treem was known in the twentieth century in the same meaning.

The side ledges of a flour trough (in a gristmill) consist of two parallel beams, which explains why we find them referred to with the same etymon trVm in modern Flemish dialects. WVD provides a distribution of variants identical to that found for the shafts-of-a-wheelbarrow meaning: West-Flemish traam, pl. tramen, East- and Zeeland-Flemish tremen and triemen.

4.4. Older Literature.

The rare occurrences of trVm in the older literature show, by and large, the same distribution of forms as in the twentieth century. In 1566, Marcus van Vaernewijck from Gent uses treme with the meaning ‘stretcher’, with the same vowel still found in Gent in the twentieth century. The meaning rung of a ladder is attested as trame in the fifteenth century in Antheunis de Roovere from Brugge, with the same vowel /a./ found in twentieth-century West Flemish.

The modern distribution of dialect forms seems to differ only in the wider spread of /a./ in the earlier period. The dictionaries of Plantin (1573) and Kiliaan (1599) have trame as ‘rung of a ladder’, and Kiliaan even lists trame as a Frisian form. In 1681, the Zeeland author Goeree uses tramen as ‘side boards of a staircase’. The modern WZD mentions /træ.bɔ.m/ (Zuid-Beveland), /træ.mby.m/ (Westdorpe) ‘pair of shafts’, in which /æ./ matches standard Dutch /a./.

5. West Germanic Etymology.

In order to establish the possible etymologies for western Dutch traam, treem, triem, we must take a look at their immediate cognates in Dutch and German. As it turns out, all of them may go back to WGm. ⁺tram-i/u- or ⁺trǣm-i/an-:

• • ⁺tramu- in Middle Dutch tram ‘wooden beam’, modern Limburgian tram ‘id.’ (Genk: Jan Goossens, p.c.; Hasselt: Staelens 1989:464); northern Ripuarian trām, Eifel drām (Rheinisches Wörterbuch), with ā from lengthening in open syllable. The Middle Dutch (see Onions 1966, following Llewellyn 1936:129) or Middle Flemish (Barnhart 1988) form was borrowed into English and ultimately yielded tram(way).

• • ⁺trami (or, theoretically, ⁺tremV) in Sittard tr ɛ :m, pl. tr ɛ :mǝ ‘beam’ (Dols 1953). Tone accent 1 in the singular tr ɛ :m proves the original disyllabicity of this form.

• • ⁺trǣmi in Limburgian tri.əm ‘crossbar’ (Bree and surroundings; WLD), ‘supporting beam’ (near Roermond: Mertens 1885:230); Ripuarian tr ɛ :m (Geilenkirchen, Heinsberg), tri.əm (Wassenberg and others) ‘supporting beam’ (Rheinisches Wörterbuch).

• • ⁺trǣm-an-/ ⁺trǣm-i- in Middle Low German trame, treme ‘crossbeam, rung’, Mecklenburg tram, pl. tramen ‘rung’, ‘tooth of a fork’ (Wossidlo/Teuchert).

• • ⁺trǣm-an-/ ⁺trǣm-i- in Modern High German Tram, for example, in Austrian (WBMÖ), Swiss German (WSS), Alsatian (WEM); Hessian Drehme, Träm ‘crossbeam’ <⁺ǟ.

In Old and Middle High German, we find the probable pre-forms of these dialectal forms, but they have initial d- <WGm. ⁺þ-:

• • ⁺þrǣm-i- or ⁺þrǣm-an-/-ōn- in the OHG glosses drâma/trâma (f. n-st.?), drâmo/trâmo (m. n-st.) ‘beam’ (length uncertain); MHG drâm, pl. drǣme (m.) and drâme, pl. drâmen (m.); Lloyd/Lühr/Springer II:753f s.v. drâma.

Indeed, the closest cognate of Dutch and German trVm is the word family of ⁺þram- discussed in Lloyd/Lühr/Springer II:819ff: Old Icelandic þrǫmr ‘beam’ < PGm. ⁺þram-u- [m.] ‘end piece’, OHG dremil ‘bolt’, MHG dremel ‘beam, bolt’, Swabian trämel, Bavarian tremel ‘beam, slender pole’ < PGm. ⁺þramila.

The semantics leaves little doubt that PGm. ⁺þram-, ⁺þrǣm- ‘crossbeam’ and Dutch-Frisian-German ⁺tram-, ⁺trǣm- ‘crossbeam’ represent variants of the same words, but Dutch and Low German tr- is an irregular reflex of PGm. ⁺þr-. Lloyd/Lühr/Springer assume that tr- was adapted analogically from other words in the same semantic sphere, such as OIc. tróÐ ‘rafter, chevron’, MHG truodel, truoder ‘slat, rod’ < PGm. ⁺trōþa. Other candidates for such an analogical model could be PGm. ⁺trewa- ‘tree, wood’, and, for southern Dutch and western German, the Latin loanword trabs, accusative -em ‘beam’ (> Dutch and Ripuarian dialects /tre.f/, /trɔ.f/, /tri.f/, MoHG Träbe).Footnote ⁵ If we accept a secondary origin of tr-, Dutch and German trVm ‘beam’ may ultimately go back to four different stems with two different root vowels, WGm. ⁺þram-i/u- and ⁺þrǣm-i/an-.

There is no discernible difference in meaning between the two ablaut variants ⁺-a- and ⁺-ǣ-.Footnote ⁶ Lühr (1988:319) points to a small number of PGm. nouns with ⁺a/ǣ-ablaut and a stem-final geminate: ⁺χakkan- versus ⁺χǣggan- ‘hook’, ⁺krakkan- versus ⁺krǣggan-? ‘hook’, ⁺krappan- versus ⁺krǣppan- versus ⁺krabban- versus ⁺krǣbban- ‘hook’. Kroonen (2009: 204–210) adds the reconstructions nom. ⁺dǣbō versus gen. ⁺dappaz ‘paw’, and ⁺snǣgō versus ⁺snakkaz ‘snake’. As Lühr rightly explains, the semantics offers no support for assuming a denominal derivative in -ǣ-. Kroonen (p. 204) proposes to explain the ⁺a/ǣ-ablaut as a secondary extension of the quantitative ablaut of the type ⁺ī (< ⁺ei) versus ⁺i and ⁺ū versus ⁺u, which arose in n-stem paradigms. Thus, for example, on the model of an original n-stem nominative ⁺kībō, gen. ⁺kippaz ‘basket’ (Kroonen 2009:75), speakers would have provided the genitive ⁺krappaz ‘hook’ with a new nominative ⁺krǣbō. In Kroonen's view, the rise of this ⁺a/ǣ-pattern goes back to Proto-North West Germanic. Note that ⁺þrǣm- versus ⁺þram-, if it arose in this way, would show an extension of this ablaut to nouns without a stem-final geminate stop in the paradigm.

6. Dutch and Frisian Developments.

The comparison with German and with other Germanic languages suggests that western Dutch traam, treem, and triem, as well as Modern West Frisian trym, trime also contain WGm. ⁺a or ⁺ǣ. This suggestion restricts the number of possible pre-forms. Whereas treem could, in theory, reflect ⁺trem- (with lengthening in open syllable), and Dutch triem could continue a WGm. form in ⁺-eo- (< ⁺-eu-), such reconstructions would conflict with the remaining Germanic evidence, and are, therefore, less likely.

6.1. Dutch Triem.

The geographical distribution of /tri(.)m/ in East Flanders (map 1) is unusual. It cannot be explained in terms of any known urban focus, such as Gent or Brugge. Nor can we assume that the form spread from Zeeland Flanders southward, since such a phenomenon is otherwise unknown. One might regard /tri(.)m/ as a dialectal buffer zone between the two variants—/tra.m/ to the west and /tre.m/ to its east. However, the /tri(.)m/-area seems too homogeneous and too broad for such an explanation. It is, therefore, more likely that East-Flemish /tri(.)m/ is chronologically a relic form, left over from an originally larger area stretching all along the coast from Flanders to Friesland. Apparently, it was ousted from the larger part of Flanders by the forms with /a./ and /e./.

As was shown in section 3, coastal Dutch ie /i(.)/ may reflect WGm. ⁺ǣ. The same applies to the vowel of Modern West Frisian /triǝm/, /trim(ǝ)/: the regular reflex of ⁺ǣ is /iǝ/ (compare Hoekstra 2001, Versloot 2001a), but in various cases, this diphthong was monoph-thongized to /i:/, and, especially before nasals, shortened. Arjen Versloot (p.c.) adduces the examples of stieme and stymje ‘to steam’ < ⁺staumjan which occur in Frisian side by side with the same variation of /iǝ/ versus /i/ before m. One of the reviewers of this paper adds tiemje/tymje ‘to draw hay together in a heap’.

Thus, I would like to add Dutch triem versus traam to those words of the type schriel : schraal, which show a Frisian development of WGm. ⁺ǣ in the coastal dialects. In fact, triem would—especially, if priem is factored out as a recent intruder—be one of the most frequently attested words of this type. The larger number of variants in North Holland (triem, treem, trem, trim) may plausibly be attributed to the recent date of the shift from West Frisian to Hollandish in this area (between 1200 and 1600), and the subsequent incorporation of the vowel in different phonemes.

An additional argument for the Proto-Frisian origin of triem may be seen in the sequence tr-, since WGm. ⁺þr- regularly yields dr- in Franconian but tr- in Frisian. This was acknowledged by Vercoullie (1925: 352): “Ndl. t berust wellicht op Friso-Saks[ische] invloed” [Dutch t may be based on a Friso-Saxon influence]. However, since eastern Dutch tram and treem also have tr-, which cannot be due to Frisian, this remains uncertain.

The related North-Frisian forms, Sylt trimk /trɪmk/ ‘rung’ (with diminutive -k) and Föhr/Amrum trem ‘rung, step’, can also reflect West-Germanic ⁺þrǣm-. The vowel ⁺ǣ mostly yields Sylt /i:/ or, with shortening in closed syllable, /ɪ/, and Föhr/Amrum /ia/; yet after r, l, w, the result is often /e:/ or /ɛ/ (Ârhammar 2001:752). Thus, North Frisian and West Frisian still have the same pre-form.Footnote ⁷

6.2. Flemish Traam and Treem.

The vowel of West Flemish /tra.m/ may theoretically reflect WGm. ⁺a in open syllable or WGm. ⁺ǣ. In the latter case, /tra.m/ would continue the same pre-form as coastal Dutch /tri(.)m/, but with the Franconian outcome /a./. This would mean that West Flanders completely replaced the earlier coastal form with /i./ < ⁺ǣ by the form with /a./. Such a scenario seems unlikely. None of the more easterly Flemish dialects show /a./, which makes it difficult to understand where West Flemish /a./ would have come from. Thus, /a./ in traam probably contains short ⁺a in an open syllable.

The Flemish variants /a./ and /e./ are in complementary distribution: West Flanders has traam, while East Flanders has treem (except for the area with triem). This pattern makes it attractive to regard the West Flemish form as the umlautless counterpart of treem < ⁺trami-. Both variants may then continue a single original noun, with regular i-mutation in some case-forms but not in others. This points to an i-stem (the type Dutch stad, pl. steden ‘town’) or an u-stem (the type Old High German hant, pl. henti ‘hand’). Note that the plural use of ⁺tram- must have been frequent, when referring to, for example, the two shafts of a wheelbarrow or the rungs of a ladder. With a view to PGm. ⁺þram-u- [m.] reflected in Old Icelandic þrǫmr and Modern High German Tram ‘beam’, I prefer to reconstruct an u-stem for Dutch.

6.3. Summary.

I arrive at the following survey of pre-forms reflected in both coastal Dutch and Frisian:

1. (2)

Thus, the same two pre-forms found for German are represented here. Every region has generalized one of these variants: West Flanders chose traam, a large part of East Flanders—treem, whereas the remaining coastal Dutch area retained triem. In view of sporadic ⁺trami in Limburgian, one might assume that treem originally occurred across the board in East Flanders, Brabant and Limburg, and was ousted as “shaft of a wheelbarrow” from Brabant by the words berrie and boom. In support of the view that ⁺tram- and ⁺trǣm- originally occurred side by side in all dialects of Dutch, note the evidence provided above that trame ‘rung’ used to occur in a much wider area in previous centuries.

7. Language Contact Scenario.

A final question concerns the sociolinguistic situation that caused the adoption of (the Proto-Frisian predecessor of) triem into Franconian Dutch. Unlike the absence of i-mutation, on which Buccini bases his theory of language contact between Proto-Frisian and Franconian, the survival of triem implies the imposition of an unstable element of L1 onto L2.

At first sight, this conclusion would seem to contradict a scenario in which Franconian is learned as L2 by sociolinguistically inferior speakers of L1 Frisian. Yet, the tendencies that van Coetsem assumes in his theory of language contact are only tendencies, and different contact situations can lead to different results. In Modern English, for instance, based on L2 Anglo-Saxon, which was learned by large groups of L1 British Celtic speakers, we only sparsely find lexemes of Celtic origin—though recent studies indicate that traditional estimates in this field are too pessimistic (see Laker 2008:21–23). The contact situation between Frisian and Franconian must have differed from that between Celtic and Anglo-Saxon in at least one respect: one may assume a large degree of mutual intelligibility between these two West Germanic dialects. In addition, just as High German has adopted a large number of Low German loanwords during its spread as a means of communication over Low German territory since the sixteenth century, one might assume that Frisian words were imposed by L1 Frisians on their newly acquired Franconian. It is only natural to assume a period of frequent bi-directional bilingualism: Frisians learning Franconian (this must have been the prevalent direction), but also Franconians learning to speak or at least understand Frisian. In this way, Proto-Frisian lexemes would have been able to enter coastal Franconian.

An additional factor in the fate of loanwords in a situation of language contact is the kind of lexemes involved: specialized technical terminology has a greater chance to be imposed by L1 speakers on L2 than more general vocabulary. Van Bree & Versloot (2008:71f) refer to such more stable vocabulary as the “secondary lexicon,” which in their view comprises at least small plants and animals, domestic and children's terminology, (old) agricultural words, local toponyms, and emotional words. A relevant example has been given by Bremmer (1993:18–20) when dealing with PGm. ⁺munþa- ‘mouth’ > Old Dutch mund, Old Frisian mūth. Except in place names such as Arnemuiden, the nasal-less Anglo-Frisian reflex has only been preserved in coastal Dutch mui ‘depth between two sandbanks just off the coast through which the tidal streams run’ < ⁺mūde. Quite obviously, this is a specialized technical term, for which speakers of inland Franconian probably had no replacement. A similar case can be made for some of the words discussed in section 3, such as wiel ‘pool left after a flood caused by a broken dike’. In the case of schriel versus schraal, the former was able to survive by means of semantic differentiation.

It is likely that triem, too, was preserved by Proto-Frisian speakers when switching to Franconian. This word always has a very precise technical meaning, such as ‘shaft of a wheelbarrow’, ‘rung of a ladder’, ‘crossbar of a chair’, or ‘tooth of a rake’. Future research will have to show whether the other Frisianisms in coastal Dutch can also be reconciled with the language contact scenario outlined above.