Motion Encoding in Russian and English: Moving Beyond · PDF fileKeywords: Russian; motion...

Motion Encoding in Russian andEnglish: Moving Beyond Talmy’sTypologyANETA PAVLENKOTemple UniversityCollege of EducationDepartment of Teaching and Learning451 Ritter HallPhiladelphia, PA 19122Email: [email protected]

MARIA VOLYNSKYDrexel UniversityDepartment of English and Philosophy5059 MacAlister HallPhiladelphia, PA 19104Email: [email protected]

The aim of the present study is twofold. One, we will show that Talmy’s (1985, 1991, 2000) motiontypology that groups Russian and English together as satellite-framed languages may be justified onlinguistic grounds but is inadequate from a psycholinguistic point of view. Two, we will argue that theshortcomings of the typology may account for inconclusive findings in research on language effects inmotion cognition. The study examined lexicalization of motion in narratives elicited with the use of apicture book Frog, where are you? (Mayer, 1969) from L1 speakers of Russian (n¼ 31), L1 speakers ofEnglish (n¼ 38), andRussian–English bilinguals (n¼ 30). All bilinguals told the story twice, once in eachlanguage, andwere subdivided into early, childhood, and late bilinguals in order to control for combinedeffects of the Age of Arrival (AoA) and Length of Residence (LoR) in the L2 context on L2 performance.Quantitative and qualitative analyses of the four motion verb corpora (L1 Russian, L1 English, BilingualL1 Russian, Bilingual L2 English) revealed that L1 Russian speakers segment motion events in a morefine-grained way and encode the manner, directionality, and spatiotemporal contours of motion eventssignificantly more frequently than speakers of L1 English. Bilinguals followed language-specificlexicalization patterns in both languages but late bilinguals displayed reduced lexical diversity in L2English. These findings were linked to differences in obligatoriness, boundedness, and complexity ofencoding of motion components in the two languages. We argue that these dimensions of motionencoding can be productively explored in instructional contexts and in future studies of language andmotion cognition that go beyond Talmy’s typology.

Keywords: Russian; motion lexicon; manner encoding

WE LIVE IN A DYNAMIC WORLD WHERE WEconstantly attend to our own motion trajectories(Do I turn right here or keep going straight?) and toother animate and inanimate entities movingthrough time and space (When is my package fromAmazon going to arrive?). But do we all talk aboutmotion in the same way and attend to the sameaspects of motion events? If not, are some of thedifferences linked to linguistic encoding of

motion? To answer these questions, studies ofmotion in language and cognition commonlydraw on the work of Talmy (1985, 1991, 2000) andSlobin (1996a, 1996b, 2004a, 2004b, 2006) thatdifferentiate between languages based on twoaspects of encoding of motion components: thelocus of encoding (main verb vs. satellite) and thefrequency of encoding (high vs. low).1 In whatfollows, we will compare two languages that havetraditionally been grouped together in thistypology, English and Russian, and show thatthe typology does not adequately account forcross-linguistic differences in motion encodingbetween speakers of L1 Russian and L1 English.Then, we will examine the implications of these

The Modern Language Journal, 99, Supplement, (2015)DOI: 10.1111/modl.121770026-7902/15/32–48 $1.50/0© 2015 The Modern Language Journal

differences for motion encoding in the twolanguages of Russian–English bilinguals.

MOVING BEYOND TALMY’S TYPOLOGY

Talmy’s (1985, 1991, 2000) typology categorizeslanguages based on mappings between surfacestructures and the following meaning components,which are seen as universal: (a) figure, or themoving object (My Amazon package is arrivingtoday), (b) ground, or the reference-point objectwith respect to which the figure moves, that is, thesource, goal, or location of motion (The mail truckis approaching my building), (c) path, or the coursefollowed by the figure with respect to the ground(The mailman is walking into the building), and (d)motion, which includesmanner and/or cause, thatis, self-propelled vs. caused motion (The mailmanis walking up the stairs with my package) (for othercomponents, see Talmy, 1985, 2000).

Based on the surface encoding (lexicalization) ofthese components, Talmy (1985) divided lan-guages into three groups. The first group, satellite-framed languages (S-languages), includes Finno–Ugric, Germanic, Sino–Tibetan, and Slavic lan-guages that encodemanner ofmotion in themainverb and path in the satellites. Talmy’s (1985,2000) definition of satellites includes prefixes andparticles (e.g., run out) and excludes preposition-al phrases (e.g., run out of the house), butsome other researchers include prepositionsas well (e.g., Regier & Zheng, 2007). Thesecond group, verb-framed languages (V-languages),includes Romance, Semitic, Turkic, and signlanguages that rely on inherently directional(bare) motion verbs to encode path and makethe marking of manner optional (e.g., the Frenchverb descendre ‘descend’) (e.g., Stromqvist &Verhoeven, 2004). To achieve elaboration similarto that in S-languages, speakers of V-languagescombine bare motion verbs with other verbs,adverbs, and additional clauses (e.g., in Frenchdescendre les degres quatre a quatre ‘to descend thestairs two at a time’). The third group involveslanguages, such as Atsugewi and Navajo, thatconflate motion with figure, as in the English torain or to spit. This group, however, is usuallyignored in present-day research that reduces thetypology to the S/V dichotomy.

The type of encoding is linked, in Talmy’s(1985, 2000) view, to information salience insentence processing: The encoding of motioncomponents in the main verb places the meaningin the background (e.g., I drove to New York), whileother types of encoding place the same meaningin the foreground (e.g., I went to New York by car).

Slobin’s (2006) manner salience hypothesis makesan opposite—and a more general—argumentregarding the links betweenmotion lexicalizationand cognitive processing: In this view, S-languages,which encode manner in the main verb and relyon finite, high-frequency verbs, draw attention tomanner, while V-languages, which encode pathin the main verb and manner in nonfinite verbsand low-frequency lexical items, phrases, orclauses, reduce the salience of manner anddraw attention to path. The degree of mannersalience, in this view, is determined by the sizeand diversity of the manner lexicon in the givenlanguage and the frequency of manner encodingin descriptions ofmotion events, with regular andfrequent encoding heightening attention to thiscomponent.

Cross-linguistic research suggests that thelanguage one speaks does make a difference:All children begin at a default starting point,paying equal amount of attention to manner andpath, yet by the age of 3 they begin to displaylanguage-specific patterns of lexicalization ofmotion events (Allen et al., 2007; Maguireet al., 2010). Empirical studies show that childrenand adults speaking S-languages, such as English,German, Icelandic, or Swedish, encode mannermore frequently, use a wider range of mannerverbs, and are more likely to interpret novel verbsas manner verbs than speakers of V-languages,such as Greek, Italian, Japanese, or Spanish(Berman & Slobin, 1994; Brown & Gullberg,2008; Cardini, 2008, 2010; Filipovic, 2011;Maguire et al., 2010; Papafragou & Selimis,2010; Stromqvist & Verhoeven, 2004).

In bilinguals who speak languages that belongto different categories, lexicalization of motionevents in the first language (L1) may influencelexicalization in the second language (L2). Thus,L1 speakers of V-languages (Japanese, Spanish)learning L2 S-languages (Danish, English)were shown to encode manner less frequently inL2 speech than L1 speakers of Danish andEnglish (Brown & Gullberg, 2008; Cadierno,2010; Filipovic, 2011), while L1 speakers of anS-language (English) learning an L2 V-language(Spanish) displayed L1 influence and difficultiesin learning target-like lexicalization of manner inthe L2 (Larranaga et al., 2011). Studies withJapanese–English (Brown&Gullberg, 2008, 2010,2011), Spanish–English (Hohenstein, Eisenberg,& Naigles, 2006), and Turkish–German bilinguals(Daller, Treffers–Daller, & Furman, 2011) alsodocumented bidirectional influence, suggestingthat at higher levels of L2 proficiency, acquisitionof an S-language may increase the salience of

Aneta Pavlenko and Maria Volynsky 33

manner for lexicalization purposes in both the L2and L1.

The findings of these studies are commonlyinterpreted as support for Slobin’s (2006) man-ner salience hypothesis and as evidence thatthe frequency and type of motion encodingaffect the salience of manner for speakingpurposes. Does this mean, however, that speakersof S-languages pay more attention to the mannerof motion than speakers of V-languages who donot encode manner on a regular basis? Thisquestion is addressed in studies that use nonver-bal tasks, such as recognition and similarityjudgments, to determine whether speakers ofS- and V-languages differ in their attention toand memory for manner and path of motion.The results so far have been mixed and inconsis-tent: Some studies found no differences betweenspeakers of S-language English and speakers ofV-languages Italian, Spanish, and Greek (Cardini,2010; Loucks & Pederson, 2011; Papafragou,Massey, & Gleitman, 2002), while others foundthat speakers of S-language English pay moreattention tomanner than speakers of V-languagesJapanese, Spanish, and Greek, but only insome conditions (Filipovic, 2011; Finkbeineret al., 2002; Gennari et al., 2002; Papafragou,Hulbert, & Trueswell, 2008; Papafragou &Selimis, 2010).

The most comprehensive study to date wasconducted by Bohnemeyer, Eisenbeiss, andNarasimhan (2006), who elicited data fromspeakers of 12 V-languages, 4 S-languages, and1 serial-verb language, and found a high degreeof intra-typological variation in participants’reliance on manner and path in similarityjudgments of motion events. As a group, speakersof S-languages did not differ in manner prefer-ence from speakers of V-languages, with theexception of speakers of S-language Polishwho displayed the highest manner bias (85%)among all participants, a finding we will return tolater.

These contradictory results gave rise to twoalternative explanations. Some scholars arguethat cross-linguistic differences in lexicalization ofmotion do not affect nonverbal motion cognition(e.g., Cardini, 2010; Papafragou et al., 2002).Others suggest that language effects in motioncognition are possible; yet the research to datehas been constrained by the theoretical limita-tions of the manner salience hypothesis, method-ological shortcomings of particular tasks, andthe focus on a limited number of languagesand a restricted range of manner and pathcontrasts (e.g., Bohnemeyer et al., 2006; Loucks

& Pederson, 2011). We share these concerns andadd a concern of our own, namely the treatmentof bilinguals as representative speakers of theirL1s in motion lexicon research (for an extendeddiscussion, see Pavlenko, 2014). These concerns,however, are secondary compared to the keyproblem in the research to date, namely thelimitations of Talmy’s (1985, 1991, 2000) typologyas a theoretical framework for research onmotionlanguage and cognition and the lack of a soundalternative framework.

In the past decade, several linguists have raisedconcerns about Talmy’s (1985, 2000) typology:Some identified S-languages with low mannerverb usage and V-languages with low path usage,such as Romansh, while others identified lan-guages, such as Arrernte or Basque, that do noteasily fit within S- and V-categories (e.g., Levinson& Wilkins, 2006; Stromqvist & Verhoeven, 2004).To address such concerns, Slobin (2004b, 2006)proposed a third category—equipollently framedlanguages (E-languages)—where path and man-ner have equal weight. Critics argue, however, thatthe third category does not address the keylimitations of the typology, such as the focus onverbs and satellites at the expense of other formsthat encode motion and the focus on figure,ground, path, and manner at the expense ofcomponents encoded in non-Indo-Europeanlanguages: Tiriyo, for instance, encodes aquaticpostpositions that mark movement into and outof liquid, while Arrernte contains an elaboratecategory of inflections for the encoding ofassociated motion (e.g., ‘do __ act while movingpast’) (Beavers, Levin, & Tham, 2010; Levinson &Wilkins, 2006). The third category also does notaddress concerns raised by significant intra-typological differences within S- and V-languagegroups (Beavers et al., 2010; Bohnemeyer et al.,2006; Hasko & Perelmutter, 2010; Iakovleva,2012).

Most importantly, as noted by both Bohne-meyer et al. (2006) and Loucks and Pederson(2011), a linguistic typology that maps universalmotion components onto linguistic elementsdoes not illuminate cognitive mechanisms thatlink grammatical status with cognitive phenome-na, such as attention and memory biases. In thecase of salience, Talmy (1985, 2000) and Slobin(2006) actually make incompatible assumptionsabout the effects of encoding: The former viewsencoding in the main verbs as backgrounding ofmotion components and the latter as heighteningtheir salience. Slobin’s (2006) manner saliencehypothesis links linguistic forms and languageeffects through frequency but does not specify

34 The Modern Language Journal 99 (2015)

what frequencies are necessary to shape language-specific patterns, what cognitive mechanismsgovern the process, and how we can factor inhigh-frequency nonmanner verbs, such as theEnglish go, come, and get.

In the absence of a much-needed conversationabout the nature of language effects in motioncognition, empirical studies continue to treatEnglish as a prototypical manner language and torely on it as a stand-in, representative of other S-languages (e.g., Cardini, 2010; Filipovic, 2011;Finkbeiner et al., 2002; Gennari et al., 2002;Hohenstein et al., 2006; Larranaga et al., 2011;Loucks & Pederson, 2011; Papafragou et al.,2002, 2008; Papafragou & Selimis, 2010). Thepurpose of the present study is to ask whetherwe are justified in treating English as a canonicalmanner language and, if not, whether we canfind productive ways to move beyond thefrequency of manner encoding and Talmy’stypology in understanding the relationshipbetween language and cognition in the domainof motion.

VERBS OF MOTION IN RUSSIAN ANDENGLISH

To show that English cannot be unproblemati-cally treated as a prototypical manner languageand to put forth an alternative approach to thestudy of motion cognition, we will compareEnglish and Russian on three dimensions ofmotion encoding linked to automaticity ofcognitive processing: (a) obligatoriness of encod-ing of manner, directionality, and aspect, (b)boundedness of motion encoding, and (c)complexity of motion encoding.

The first of these, obligatoriness, involves thedegree to which lexical or grammatical markingof a particular distinction is necessary in thelanguage in question (Lucy, 1992, 1996). Theeffects of obligatoriness have been explored inthe domain of number marking through triadcategorization tasks, where participants have todecide on the similarity between simpleobjects and their shape and material alternates(Gathercole & Min, 1997; Imai & Gentner, 1997;Imai &Mazuka, 2003, 2007; Li, Dunham, & Carey,2009; Lucy, 1992; Sera & Goodrich, 2010;Subrahmanyam & Chen, 2006). These studiesreveal that speakers of the noun class languagesEnglish and Spanish, where plural is markedobligatorily on the majority of nouns, commonlygroup the items on the basis of shape. Thisgrouping is interpreted as a language effectwhereby the obligatory count/mass distinction

draws attention to discreteness of entities. Incontrast, speakers of classifier languages likeChinese, Japanese, Korean, and Yucatec, whereplural is marked optionally on a small number ofnouns, tend to group together entities similar inmaterial, suggesting that the use of mass nounsaccompanied by classifiers increases sensitivity tomaterial or substance. Together, these findingssuggest that obligatoriness of linguistic encodingmay increase perceptual attention to the distinc-tion in question and automaticity of its process-ing, with automaticity referring to fast, efficient,and stabilized patterns of processing (Segalowitz,2010). But what do these findings mean forlinguistic encoding of motion?

Despite the common perception of English as amanner language,manner encoding in English is,in fact, optional. English does not require itsspeakers to encode manner obligatorily—todiscuss motion, English speakers can also appealto high-frequency generics, such as come, go, andget, that cover a wide range of motion events(arriving, departing, entering, exiting, and mov-ing up/down/across) (e.g., Hasko, 2010a; Iakov-leva, 2012). In contrast, Russian does not havesuch wide-coverage generics—its few nonmannermotion verbs, such as the deictic verb pribyt’ (‘toarrive’) or the vertical event verbs podniat’sia(‘ascend’) and spustit’sia (‘to descend’) refer tospecific motion events, making manner encodingobligatory inmost contexts.2 As a consequence, incontexts where English speakers talk about goingor getting somewhere, Russian speakers arerequired to differentiate between motion onfoot (idti/khodit’ ‘to walk’) and motion by meansof transportation (ekhat’/ezdit’ ‘to ride, to drive’).The difference between the two languages,therefore, lies not in the number of manner ornonmanner verbs or the frequency of their usebut in the scope of motion generics (wide inEnglish, narrow in Russian) and, consequently, inthe number and type of motion events that can bedescribed with generics.

The impact of these differences on performancecan be seen in Slobin’s (2004a, 2006) study ofnarratives elicited by Mayer’s (1969) book Frog,where are you?. In descriptions of the owl suddenlyappearing from the hole in the tree, L1 speakers ofDutch, German, and English favored genericdeictic path verbs, such as come or appear, usingmanner verbs, respectively, in 17%, 18%, and 32%of all descriptions. In contrast, L1 Russian speakersused manner verbs, such as vyletela (‘out-flew’),100% of the time. Hasko’s (2010a) study extendedthese findings to other events in Frog, where are you?,showing that L1 English speakers were much more


likely to opt for nonmanner generics (29% of allmotion verbs) than L1 speakers of Russian (11%).

Another obligatory, and this time grammati-cized, distinction encoded in Russian is direction-ality of motion. This category does not appear inTalmy’s (1985) typology and is absent in English,where bare motion verbs refer to motion pro-ceeding in a single direction (e.g., to walk, to run).Russian grammaticizes directionality throughstem variation in a closed class of high-frequencymotion verbs, where each verb pair contains aunidirectional verb that refers to motion proceed-ing in a single direction (e.g., ekhat’ ‘to ride/drivein one direction’) and a multidirectional verb thatrefers to motion proceeding in/from more thanone direction, such as aimless movement, roundtrips, and habitual or repeated motion (e.g., ezdit’‘to ride/drive back and forth, repeatedly’)(Zalizniak & Shmelev, 2000).

The third distinction obligatorily marked inRussianmotion verbs is aspect. In English, aspect isa syntactic category that functions in combinationwith tense; it is not an intrinsic characteristic ofverbs, nor is it marked in the infinitive. In Russian,aspect is a lexico-syntactic category, independentof tense, that characterizes all verb forms,including infinitives, imperatives, and participles.All Russian verbs belong to one of two aspectualcategories: imperfective, which refers to theprocess, state, or habitual action, and perfective,which refers to achievement or accomplishment.Despite some overlapwith English progressive andperfective aspects, these categories do not mirrorthem and do not correspond to tense–aspectcombinations encoded in English (Hasko, 2010a;Pavlenko, 2010; Zalizniak & Shmelev, 2000).

The intersection of aspect and directionalityadds another layer of complexity to the Russianmotion verb system, because base imperfectiveverbs, which form directionality pairs (e.g.,bezhat’/begat’ ‘to be running in a single direc-tion’/‘to run around or back and forth’), can becombined with different prefixes to form a varietyof perfective verbs, such as pobezhat’ (‘to startrunning, to run somewhere’), zabezhat’ (‘to run inand out, to stop by’) or vbezhat’ (‘to run in’). Sinceprefixes also mark other distinctions, the sameprefix can create a perfective and an imperfectiveverb (e.g., sbezhat’ ‘to run down PERFECTIVE’), sbegat’(‘to run somewhere and come back PERFECTIVE’),sbegat’ (‘to be running down IMPERFECTIVE’) (Hasko& Perelmutter, 2010).

These verbs also illustrate the second majordifference between English and Russian, in thecompactness and boundedness of motion encoding.3

Talmy’s (1985) typology treats bounded and

unbounded morphemes equally as satellites, yetfrom a psycholinguistic point of view bounded-ness does make a difference: In the process ofmaking interlingual identifications, L2 learnerssystematically distinguish between bounded andunbounded morphemes (Jarvis & Odlin, 2000;Jarvis & Pavlenko, 2008). This means that the firstimpulse of an L1 English speaker translating randown into L2 Russian would be to match theEnglish morphemes with their Russian counter-parts bezhala (‘ran, was running’) and vniz(‘down’) and not with the prefixed verb sbezhala(‘down-ran’). The prefixed perfective verbs, likesbezhala (‘ran down’), pobezhala (‘started run-ning’), zabezhala (‘ran in and out’), or ubezhala(‘ran away’), may have to be acquired asindividual items. The mapping is further compli-cated by the fact that information encoded inEnglish particles may be encoded in Russian bothin the prefixes and in the prepositions (e.g.,sbezhala vniz ‘down-ran in-down’). The learner’stask is particularly challenging because motionencoding in Russian is simultaneously morecompact (i.e., encoded through bounded mor-phemes) and more distributed than it is inEnglish. In English, the key aspects of motionare encoded in verb stems and particles (as well asprepositions), while in Russian the information isdistributed between the stems, prefixes, suffixes,inflections, and particles (as well as prepositions).

Last but not least, the number of distinctionsobligatorily marked in Russianmotion verbs (e.g.,manner, directionality, aspect, number, gender)also results in much greater complexity, seen ingreater informational load carried by individualRussian motion verbs. This means that in thelexicalization of motion events, L1 Russian speak-ers are required to pay simultaneous attention toseveral types of information that are not obligato-rily encoded in English, most notably manner,directionality, and temporal contours of motionevents (aspect).

The next question to ask is whether thesedifferences in obligatory attention foci havecognitive consequences for the L2 learningprocess. Do they shape different tasks for L1English learners of L2 Russian and L1 Russianlearners of L2 English and different patterns ofcross-linguistic influence?

MOTION LEXICONS OF RUSSIAN–ENGLISHAND ENGLISH–RUSSIAN BILINGUALS

Teachers of Russian as a second or foreignlanguage are fully aware of the challengespresented by the Russian motion lexicon to


speakers of non-Slavic languages. In the English-speaking world, these difficulties gave rise to awhole industry of websites and textbooks dedicat-ed solely to Russian motion verbs (e.g., Alexan-drova & Watt, 2013; Mahota, 1996; Muravyova,1986; Polivin, 2010). Yet, despite this increasedattention, studies show that the L2 Russianmotion lexicon remains a challenge for L1English speakers.

Hasko (2009, 2010b) compared Frog storynarratives elicited from speakers of L1 Russian(n¼ 30) with narratives elicited from AmericanL2 learners of Russian (n¼ 30). The learners hadstudied Russian, on average, for about 5 years andwere enrolled in advanced-level Russian coursesin an intensive summer immersion program,based on rigorous placement tests, including asimulated Oral Proficiency Test. Nevertheless,their narratives revealed a high percentage oferrors in motion encoding, lower lexical diversityin the motion lexicon, and insufficient encodingof manner and directionality, compared to L1Russian speakers.

Among the key errors were two types of L1semantic transfer. The first type stems fromincorrect interlingual identifications betweenEnglish generics, such as go, and manner-specificRussian verbs, such as idti (‘to walk’). Oneparticipant stated, for instance, Sova prishla imnogo pchel prishli (‘The owl walked-in and manybees walked-in PL’). The learner was familiar withthe verb letet’ (‘to fly’) yet in online narration themotion event triggered the generic English come,linked, incorrectly, to the manner-specific Rus-sian idti. This pattern suggests that the L2 learnersstill treat manner of motion as optional and thatRussian verbs do not trigger the same mentalimagery in L1 and L2 Russian speakers.

The second type of transfer stems fromincorrect identifications between inherently uni-directional English motion verbs and multidirec-tional Russian verbs. Thus, another participantstated Liagushka polzala iz banki (‘[The] frog wascrawling around out [of] [the] jar’). Polzala (‘wascrawling around PAST IMPERFECTIVE FEMININE’) is aninappropriate choice for a single finite motionevent, more appropriately described with vypolzlaor vylezla (‘crawled-out’/‘climbed-out PAST PERFEC-

TIVE FEMININE’). This pattern of interlingualidentifications reveals that the learner has notinternalized the distinctions in terms of direc-tionality and aspect (and associated mentalimagery) and focused only on motion compo-nents encoded in English, treating members ofdirectional and aspectual verb pairs as equivalentsof a single English verb.

Gor et al. (2009) compared perception andproduction of Russian motion verbs in Americanlearners of L2 Russian (n¼ 36) with that in L1Russian speakers (n¼ 10) and early bilingualheritage speakers of Russian in the United States(n¼ 24). They found that neither L2 learners norearly bilinguals achieved nativelike mastery of theRussian motion lexicon. In the verb and sentencecompletion tasks, early bilinguals outperformedL2 learners with similar levels of proficiency(likely, as a result of more extensive input) but onthe grammaticality judgment task early bilingualswere more willing to accept incorrect substitu-tions ofmultidirectional verbs with unidirectionalones (likely, as a result of insufficient grammarinstruction).

Similar difficulties have been documented byPolinsky (2008), who found that early bilingualsno longer perceived L1 Russian verbs as perfec-tive or imperfective—instead, following the con-straints of English, they retained only onemember of each aspectual pair, which theytreated as lexical items without specified aspec-tual value. In Pavlenko’s (2010) study, narrativesby four early Russian–English bilinguals dis-played simplification of aspect, directionality,and manner distinctions, interpreted as incom-plete acquisition. Some instances also revealedL2 English influence on L1 Russian: Under theinfluence of the English go, for instance, speakersextended the L1 Russian verb idti (‘to walk’) toreferences to riding/driving, climbing, crawling,and flying, thus patterning with L1 Englishspeakers learning L2 Russian. In contrast, 70Russian–English bilinguals who arrived in theUnited States after the age of 14 did not differfrom L1 Russian speakers in Russia in terms ofaccuracy and lexical diversity of the L1 motionlexicon.

Now, what about L1 Russian learners of L2English? Iakovleva (2012) examined descriptionsof motion in short animated clips in a populationthat mirrored that in Hasko’s (2009, 2010b)study, namely L1 Russian learners of L2 Englishin a classroom context in Russia. The studycompared descriptions of voluntary motionalong three paths (up, down, across) by L1Russian speakers (n¼ 12), L1 English speakers(n¼ 17), and L1 Russian learners of L2 English(n¼ 12) at intermediate and advanced levels ofproficiency, with ages of L2 acquisition between5 and 11 years. Comparative corpus analysisdemonstrated that L1 English speakers reliedon the canonical S-language pattern (i.e., encod-ing manner in verbs and path in satellites) inall three tasks, while L1 Russian speakers


displayed three different patterns: S-framing formotion across, conflation of manner and pathfor motion up, and V-framing, that is, relianceon generics, for downward motion. Thesefindings confirm that the typological profile ofRussian is highly complex and variable andthat it does not easily fit in the traditionalS-language category. L1 Russian learners of L2English were mostly target-like, especially indescriptions of up- and across-events, suggestingthat the lesser complexity of the English motionsystem facilitates acquisition and constrainsnegative transfer. At the same time, they stilldisplayed subtle traces of L1 influence, in termsof types of information encoded. For instance,one participant produced a sentence A girl iscrossing the lake on skates or she is crossing the lake byskating. The participant’s attempt to specify themanner of motion in both clauses reveals thattheir performance is still shaped by the auto-matic habit to encode information aboutmanner.

Together, these findings suggest that differ-ences in obligatoriness, boundedness, andcomplexity of motion components shape differ-ent tasks for speakers of Russian and Englishlearning each other’s languages. L1 Englishspeakers learning L2 Russian have to learn to paymore attention to manner, aspect, and direc-tionality in order to make their encodingautomatic. They also have to restructure linguis-tic patterns of motion encoding, learning toencode multiple new types of information in amore distributed way, through prefixes, verbstems, suffixes, inflections, and prepositions. Incontrast, L1 Russian speakers learning L2English begin by relying on their automaticattention biases and learn to pay less attention tomanner and directionality and to encode pathinformation through particles and prepositions.What we do not know yet is how the two motionlexicons interact in bilingual speakers, becausethe studies to date have focused on a singlelanguage of bilingual speakers, either L1 (e.g.,Pavlenko, 2010) or L2 (e.g., Hasko, 2009, 2010b;Iakovleva, 2012). In what follows, we expand thisline of research and examine lexicalization ofmotion in both languages of Russian–Englishbilinguals.

RESEARCH DESIGN

Research Questions

Based on the findings to date, we havearticulated four research questions for our study:4

Participants

Ninety-nine participants took part in the study:(1) L1 speakers of Russian (n¼ 31) were under-graduates at Tomsk State University, Russia;Russian was their native and dominant language,yet they also had some (instructed) knowledge ofEnglish and a few reported some knowledge ofother languages (Buriat, French, German, Japa-nese, Kazakh), (2) L1 speakers of English (n¼ 38)were undergraduates at Temple University,Philadelphia; English was their native and domi-nant language and all reported low levels ofcompetence in foreign languages (French,German, Italian, Japanese, Latin, Spanish), (3)Russian–English bilinguals (REB)wereundergrad-uate (n¼ 24) and graduate students (n¼ 6) atBryn Mawr College, Chestnut Hill College, andTempleUniversity, Philadelphia; somehad limitedcompetence in languages other than Englishand Russian. Clearly, the majority of our L1speakers are not monolingual under a strictdefinition and our bilinguals could be consideredmultilingual, yet we do not see the L2 or L3 as aconfounding variable, because of low levels offoreign language competence and because theseL2s and L3s were mainly V-languages that do notrequire obligatory encoding of either manner ordirectionality and thus could not enhance sensitiv-ity to these distinctions. All groups included maleand female participants but wemadeno attempt tocontrol for or to investigate the effects of gender,because previous studies on lexicalization ofmotion, as already discussed, have not identifiedany gender effects in this area of the lexicon.

As seen in Table 1, bilingual participants weresubdivided into three groups based on the AoAand LoR in the L2 context: (1) 10 early bilinguals,

RQ1. What are the similarities and differencesbetween L1 speakers of Russian and Englishin lexicalization of motion in the context ofelicited narratives?

RQ2. What are the similarities and differencesbetween the motion lexicon of L1 Englishspeakers and the L2 English motion lexiconof Russian–English bilinguals?

RQ3. What are the similarities and differencesbetween the motion lexicon of L1 Russianspeakers and the L1 Russian motion lexiconof Russian–English bilinguals?

RQ4. What, if any, role is played by the combinedeffects of the Age of Arrival (AoA) and theLength of Residence (LoR) in the L2context in the acquisition of L2 Englishmotion lexicon and the maintenance of theL1 Russian motion lexicon by Russian–English bilinguals?


(2) 10 childhood bilinguals, and (3) 10 latebilinguals. While combining the two variablesdoes not allow us to attribute the effectsunambiguously to one or the other, this seemedlike the only reasonable approach given the factthat participants who arrived youngest had alsobeen in theUnited States longest.One of the earlybilinguals was born in the United States, and therest of the early and childhood bilinguals arrivedin the United States as members of Russian-speaking immigrant families from Russia (8),Ukraine (5), Moldova (2), Belarus (1), Georgia(1), Latvia (1), or Uzbekistan (1). They usedRussian at home with family members and, in afew cases, with friends and colleagues at work. Allattended secondary school in the United Statesand used English for interactional and education-al purposes. The key differences between thesetwo subgroups were in the AoA and the LoR(Table 1) and in the self-reported levels ofproficiency in the two languages (Table 2).

Late bilinguals differed from the other twogroups in their learning trajectory, interactionalcontexts, and self-reported proficiencies. All tencame from Russian-speaking families in Belarus(4), Kazakhstan (2), Russia (2), or Ukraine (2),four arrived in the United States as immigrants,

four as students, and two as green-card holders.Most had been exposed to English prior to theirarrival through secondary school and college-level instruction; consequently, their actual age ofL2 acquisition is earlier than the AoA adoptedhere, yet we decided to keep the definition ofthe AoA consistent for all three groups because wehad no means to control for the quality and typeof English instruction prior to their arrival inthe United States. All reported using Russian totalk to family members and Russian-speakingfriends and for leisure activities (books, TV,movies, internet). They also regularly used Englishfor everyday interaction, education, and work.

Methods and Procedures

Narratives were elicited from all participantsindividually with the use of the book Frog, where areyou? (Mayer, 1969), selected because it facilitatedcomparisons with other Russian–English Frogstory corpora (e.g., Hasko, 2009, 2010a, 2010b).The participants were asked to describe, pictureby picture, what they saw happening in the book.Narratives of bilingual participants were elicitedtwice, once in each language, with elicitationconducted in separate sessions, with a minimum

TABLE 1Study Participants

Participant GroupAge (Years) atTime of Testing

Age (Years) ofArrival in L2Context (AoA)

Length (Years) ofResidence in L2Context (LoR) Proficiency

L1 Russian Speakers (n¼ 31) 20.7 (19–23)L1 English Speakers (n¼ 38) 19.6 (18–22)Russian–English Bilinguals (n¼ 30) 22.0 (18–32)Early Russian–English Bilinguals 21.3 (18–32) 3.3 (0–6) 17.9 (14–28) Russian: 4.4

(n¼ 10) English: 7.0Childhood Russian–English 19.6 (18–26) 9.2 (7–13) 10.5 (8–14) Russian: 5.2

Bilinguals (n¼ 10) English: 6.8Late Russian–English Bilinguals 25.1 (19–31) 19.5 (14–29) 5.5 (2–12) Russian: 6.7

(n¼ 10) English: 5.6

Note. Proficiency based on self-ratings using a scale from 0 to 7, with 0meaning ‘not at all’ and 7 ‘native or native-like’ (ranges in parentheses).

TABLE 2Self-Reported Levels of Russian and English Proficiency

Participants

Speaking Listening Reading Writing

Russian English Russian English Russian English Russian English

Early Bilinguals 5.0 7.0 6.3 7.0 3.3 7.0 2.9 7.0Childhood Bilinguals 5.8 6.8 6.5 6.9 4.6 6.9 4.0 6.5Late Bilinguals 6.9 5.3 7.0 5.8 6.5 5.9 6.4 5.3

Note. Ratings based on a 0 to 7 Likert scale with 0 meaning ‘not at all’ and 7 ‘native or native-like’ (ranges inparentheses).


interval of 2 weeks and a maximum of 2 years. Ineach group, five randomly selected participantsfirst performed the task in Russian and the otherfive in English. The task was performed in thecontext of a larger study where participants alsoperformed other tasks, such as naming andtypicality judgments of drinking containers.Sociodemographic data were collected via awritten language background questionnaire anda tape-recorded semi-structured language learn-ing history interview. The interviews were con-ducted in the language of subsequent narrativeelicitation, in order to ensure the activation of thelanguage in question.

Data Analysis

All narratives were transcribed in the languageof the telling by one of the two authors and cross-checked by the other. We then created corporacontaining voluntary motion verbs that referred tothe figure changing position or location. Consis-tent with this definition, we included Englishphrasal verbs (e.g., sit down, stand up) and Russianperfective verbs (e.g., sest’ ‘to sit down,’ vstat’ ‘tostand up’) and excluded verbs referring to causedmotion, English action verbs, and Russian imper-fective verbs (e.g., to sit/sidet’, to stand/stoiat’)because they involve no change in position. Onlyverbs on which both researchers agreed wereincluded in the analysis.

Next, all of the verbs in the respective corporawere divided into types, counted as tokens, andcategorized in terms of manner, path, and, inRussian, directionality. In Russian, perfective andimperfective verbs and verbs with different affixeswere counted as separate types, even if they hadthe same verb root, because they referred todifferent types of motion, so that pribezhat’ (‘toarrive by running’), vybezhat’ (‘to run out [with apossibility of return]’) and ubezhat’ (‘to run away’)would be counted as different verbs. The sameapproach was adopted in English where phrasalverbs with the same root and different particleswere counted as different verbs (e.g., get in, getout). Once again, all categorization was based onconsensus between the two researchers.

Quantitative analyses, based on type and tokenfrequencies, compared four corpora—L1Russian,L1 English, Bilingual L1 Russian, and Bilingual L2English—on lexical diversity, type of path encod-ing, and the frequency of motion verb use andmanner encoding. Throughout, we used anindependent samples t-test to compare L1 Russianand L1 English corpora; one-way ANOVAs forthree-way comparisons among L1 Russian, L1

English, and Bilingual L1 Russian (or L2 English),and a paired samples t-test for REB (Russian–English Bilingual) corpora. To examine differ-ences between the three REB groups (early,childhood, and late bilinguals), we used Krus-kall–Wallis tests with a Bonferroni correction.

To examine lexical diversity, we have adoptedDugast’s Uber formula, which is better suited forrelatively small samples with narratives of differ-ent length (Jarvis, 2002):

Log, here, stands for logarithm, which is aquantity representing the power to which a fixednumber (the base) must be raised to produce agiven number. In this formula, the squaredlogarithm of all the verbs produced by a partici-pant is divided by the difference of the logarithmof all the words and logarithm of the word typesproduced by a participant. Logarithms are usedfor maximum-likelihood estimation to find anonlinear relationship between variables. Tocarry out the analysis, we calculated the totalnumber of motion verbs (tokens) used by eachparticipant. Then, we calculated the number oftypes of motion verbs used by each participant.After tabulating the total number of tokens andtypes of motion verbs in each narrative, the totalnumber of tokens and types was calculated foreach group of the participants and then the Uberindex was calculated for each participant and theformula was applied to determine whether therewere statistically significant differences in lexicaldiversity among the four groups.

Qualitative analyses involved two steps. First, weidentified lexical errors and deviations fromstandard Russian and English usage in the REBcorpora: Instances counted as errors or deviationsonly if the two researchers were in agreement andif no L1 speaker made the same lexical choice inthe same context. Next, we identified instances ofcross-linguistic influence, that is, lexicalizationchoices that differed from those made by L1speakers of the language in question andpatterned with those made by the speakers ofthe other language.

RESULTS

Lexical Diversity of the Motion Lexicons andFrequency of Motion Verb Use

In the analysis of motion verb use, independentsamples t-tests showed that L1 Russian speakers

(log tokens)2Uber index¼U¼ ––––––––––––––––––––––––––––––(log tokens – log types)


used significantly more motion verbs (M¼ 23.2)than L1 English speakers (M¼ 16.4) (t[67]¼ 4.2,p<.000, r2¼ .20), as well as more motion verbtypes (136 vs. 87), displaying significantly higherlexical diversity of the motion lexicon (M¼ 15.8)than L1 English speakers (M¼ 11.7) (t[66]¼ 2.4,p<.05). These findings, summarized in Table 3,are consistent those obtained byHasko (2010a) ina comparative analysis of L1 Russian and L1English Frog story corpora.

One-way ANOVAs indicated significant differ-ences in the number of motion verbs among L1Russian, L1 English, and REB L1 Russian corpora(F[2,96]¼ 8.1, p<.05, h2¼ .14) and among L1Russian, L1 English, and REB L2 English corpora(F[2,96]¼ 9.3, p<.05, h2¼ .16). Tukey’s HSD posthoc analyses revealed that bilinguals performedsimilarly to L1 English speakers and differed fromL1 Russian speakers both in L1 Russian and L2English (p< 0.5). A paired t-test found nosignificant differences in the number of motionverbs used by bilinguals in L1 Russian (M¼ 18.2,SD¼ 8.0) and L2 English (M¼ 16.8, SD¼ 7.9).

A similar pattern was identified in the analysis oflexical diversity (see Figure 1). One-way ANOVAsrevealed statistically significant differences incomparisons of L1 Russian, L1 English, and REBL1 Russian corpora (F¼ 7.5, p¼ .001) and of L1Russian, L1 English, and REB L2 English corpora(F¼ 11.0, p¼ .000), with the post hoc analysisshowing that L1 Russian speakers had a signifi-cantly higher Uber index of lexical diversity(U¼ 11.3) than L1 English speakers (U¼ 9.1)and bilinguals in L1 Russian (U¼ 9.9) and L2English (U¼ 8.2). No differences were foundbetween bilinguals’ L1 and L2 corpora and the

L1 English corpus, nor were there any significantdifferences in lexical diversity between bilinguals’L1 and L2 corpora (t[29]¼ .62, p> .05).

These findings are interesting because theyshow that, in the context of significant differencesbetween L1 Russian and L1 English speakers,bilinguals display a pattern of convergence intheir two languages that is proceeding in thedirection of L2 English, as seen in the decrease inlexical diversity and the types of motion verbs inthe L1 Russian lexicon.

Manner Encoding

In the analysis of manner encoding, indepen-dent samples t-tests revealed that L1 Russianspeakers used a significantly higher percentage ofmanner verbs (96.5%) than L1 English speakers

FIGURE 1Lexical Diversity of Motion Vocabulary in the FourCorpora

TABLE 3Verbs of Motion in Russian and English Narratives

Corpus Sizeand Mean

Narrative Length

Verbs ofMotion

(Proportion)

Verbs ofMotion(Types)

MannerEncodingin the

Motion Corpus

Path Encodingin the MotionCorpus (PathSegments)

L1 Russian (n¼ 31) 15,239 719 (4.7%) 136 694 (96.5%)Mean¼ 491.6 Mean¼ 23.2 Mean¼ 22.4 Mean¼ 29.6SD¼ 173.9 SD¼ 7.1 SD¼ 7.0 SD¼ 10.0

L1 English (n¼ 38) 19,430 624 (3.2%) 87 407 (65.2%)Mean¼ 511.3 Mean¼ 16.4 Mean¼ 10.7 Mean¼ 12.9SD¼ 241.0 SD¼ 6.3 SD¼ 4.2 SD¼ 6.0

REB L1 Russian (n¼ 30) 11,264 545 (4.8%) 96 537 (98.5%)Mean¼ 375.4 Mean¼ 18.2 Mean¼ 17.9 Mean¼ 24.3SD¼ 187.3 SD¼ 8.0 SD¼ 7.9 SD¼ 11.0

REB L2 English (n¼ 30) 15,896 505 (3.2%) 66 315 (62.4%)Mean¼ 529.9 Mean¼ 16.8 Mean¼ 10.5 Mean¼ 14.3SD¼ 218.6 SD¼ 7.9 SD¼ 5.4 SD¼ 7.0


(65.2%) (t[67]¼ 12.3, p<.001) and significantlymore manner verbs per narrative (M¼ 22.4) thanL1 English speakers (M¼ 10.7) (t[67]¼ –11.70,p<.05, r2¼ .52). L1 English narratives containedseveral high-frequency nonmanner verbs, suchas come, go, get, and escape, while L1 Russiannarratives contained only one verb that could beinterpreted as a nonmanner motion verb, osta-novit’sia (‘to stop’). L1 Russian-speaking narra-tors also made more distinctions in terms ofmanner, using 129 types of manner verbs ascompared to 62 types in the L1 English corpus.An independent samples t-test showed that L1Russian narratives displayed significantly higherdiversity of manner verbs (M¼ 15.9, SD¼ 4.5)than L1 English narratives (M¼ 7.7, SD¼ 3,0)(t[67]¼ -9.0, p<.05, r2¼ .54). These findings—consistent with the findings ofHasko (2010a) andIakovleva (2012)—suggest that L1 Russian speak-ers encode manner more frequently and makemore fine-grained distinctions in this domainthan L1 speakers of English.

One-way ANOVAs indicated a significant dif-ference in the number ofmanner verbs among L1English, L1 Russian, and REB L2 English corpora(F[2,96]¼ 47.8, p<.05, h2¼ .49) and among L1English, L1 Russian, and REB L1 Russian corpora(F[2,96]¼ 29.6, p<.05, h2¼ .38). A post hocanalysis showed that bilinguals speaking L1Russian and L2 English used significantly fewermanner verbs than L1 Russian speakers; nodifferences were found between L1 Englishspeakers and bilinguals speaking L2 English.The obligatoriness of manner encoding in L1Russian was maintained in the bilingual corpus(98.5%) but it was not transferred to L2 English,where the frequency of manner encodingapproximated that of L1 English speakers(62.4% vs 65.2%). A paired samples t-test revealeda significant difference in the number ofmanner verbs used in the L1 Russian (M¼ 17.9,SD¼ 7.9) and the L2 English (M¼ 10.5, SD¼ 5.4)(t[29]¼ 5.8, p¼ .000) of bilingual speakers.In terms of the types of manner verbs, one-way ANOVAs indicated differences among L1English, L1 Russian, and REB L2 English corpora(F[2,96]¼ 63.3, p<.05, h2¼ .56) and amongL1 English, L1 Russian, and REB L1 Russiancorpora (F[2,96]¼ 30.9, p<.05, h2¼ .39). Posthoc Tukey’s HSD tests revealed significantdifferences between all corpora except for L1English andL2English. The L1Russian lexicon ofbilingual speakers displayed a significantly higherdiversity of manner verbs (M¼ 11.4, SD¼ 5.4;93 types) than L2 English (M¼ 6.6, SD¼ 3.1;40 types) (t[29]¼ 4.6, p¼ .000).

Qualitative analyses of manner encodingrevealed no influence of L1 Russian on L2English. L2 English influence on L1 Russian wasfound in semantic extensions and attrition ofcategory boundaries and obligatory distinctions.Most prominently, under the influence of thegeneric English go, five of the ten early bilingualsextendedmanner verbs idti/khodit’ (‘to walk’) andtheir derivatives to climbing, flying, jumping, andcrawling as in the following example:

EXAMPLE 1

Together, these findings show that the L1Russian of bilingual speakers displays lower fre-quency and reduced lexical diversity of themannerlexicon, situated in between diversity levels dis-played by L1 speakers of Russian and English. Theobligatory encoding of manner in bilinguals’ L1Russian remains automatic but some high-frequen-cy Russian manner verbs are incorrectly assignedthe meanings of English generics. In L2 English,bilinguals display lower sensitivity to manner,following the L1 English pattern.

Path and Directionality Encoding

An analysis of path segments (path verbs,satellites, adverbs, prepositions) demonstratedthat L1 English speakers tended to encode paththrough one-segment clauses (87%), while L1Russian speakers employed more two-segmentpath clauses (55%), which is consistent with thehigher complexity and distributed nature ofmotion encoding in Russian. The independentsamples t-test showed that L1 Russian speakersused a significantly higher number of pathsegments per narrative (M¼ 29.6) than L1English speakers (M¼ 12.9) (t[67]¼� 8.5,p<.05, r2¼ .52). Replicating the findings ofHasko’s (2010a) study, these differences suggestthat L1 Russian speakers attend to both mannerand path of motion to a greater degree than L1English speakers.

Bilinguals displayed language-specific patternsin both languages, favoring one-segment clausesin L2 English (86.7%) and two-segment pathclauses (most commonly a prefixþpreposition)in L1 Russian (56.4%). Examples of such two-segmented path clauses follow:

маленькое животное выходит из дыркуmalen’koe zhivotnoe vykhodit iz dyrku(a) small animal out-walks from (the) hole‘A small animal is walking out of the hole.’(an appropriate lexical choice here is vylezaet [‘out-climbs’/‘crawls out’])


EXAMPLE 2

One-way ANOVAs revealed significant differ-ences in the number of path segments for bothcomparisons, L1 English, L1 Russian, and REB L2English corpora (F[2,96]¼ 45.9, p<.05, h2¼ .48)and L1 English, L1 Russian, and REB L1 Russiancorpora (F[2,96]¼ 30.78, p<.05, h2¼ .39). Therewere no differences between L1 Russian speakersand REB L1 Russian corpora nor between L1English and REB L2 English. A paired t-test founda significant difference in path segments used inL1 Russian (M¼ 24.3) and L2 English (M¼ 14.3)(t[29]¼ 5.47, p¼ .000). These findings suggestthat bilinguals maintain language-specific pat-terns of path encoding.

Since directionality is encoded in Russian butnot in English, a comparison of usage ofunidirectional and multidirectional verbs wascarried out between L1 Russian and REB L1Russian corpora. A comparison of the usage ofunidirectional verbs by L1 Russian speakers(M¼ 16.6) and bilinguals in L1 Russian(M¼ 13.7) revealed no differences (t[59]¼ 1.5,p¼ .92), nor did the comparison of usage ofmultidirectional verbs in L1 Russian (M¼ 6.7)and REB L1 Russian (M¼ 4.5) (t[59]¼ 1.8,p¼ .60). Bilinguals also maintained the prefer-ence for unidirectional perfective verbs (59.9% ofthe L1 Russian corpus, 66.5% of the REB corpus).These quantitative patterns, however, do not tellus much about the accuracy of individual lexicalchoices. Qualitative analysis found that, to markthe beginning of an action, some bilingualsopted, incorrectly, for verbal constructions withmultidirectional imperfective verbs:

EXAMPLE 3

Together, these findings suggest that in L2English bilinguals approximate L1 English pat-terns without displaying any L1Russian influence,while in L1 Russian they maintain language-specific lexicalization of path and obligatoryencoding of directionality, with some loss oflexical precision and/or L2 English influence.

AoA/LoR Effects on Bilinguals’ Motion Lexicons

The previous analyses treat Russian–Englishbilinguals as a group, but we also wanted toexamine potential differences among the threebilingual groups. Contrary to our expectations, aseries of Kruskall–Wallis tests revealed that therewere no significant differences among the threegroups on any aspects of motion encoding in L1Russian, even though the Uber index of lexicaldiversity was highest in late bilinguals. In L2English, there was a statistically significantdifference in lexical diversity (x2¼ 10.6,p¼ 0.005) between early and childhood bilin-guals (Z¼� 2.9, p¼ 0.004), and early and latebilinguals (x2¼� 2.7, p¼ 0.007). Early bilin-guals also used significantly more types ofEnglish manner verbs than late bilinguals(x2¼ 7.6, p¼ 0.006), a result that can beexplained by differences in English proficiency,which, in turn, stem from the earlier AoA andthe longer LoR in the United States. Qualitativeanalyses of deviations from standard usagerevealed only 18 errors in the L2 English corpus,all of them produced by four late bilinguals. Itappears that early, childhood, and the majorityof late bilinguals have fully acquired Englishpatterns of motion lexicalization. As a conse-quence, the discussion that follows will focus onmotion encoding in the L1 Russian of bilingualspeakers.

As seen in Table 4, the L1 Russian corpuscontained 81 errors, produced by 26 participants(10 early bilinguals, 9 childhood bilinguals, 5 latebilinguals). These errors were divided into threecategories, based on the aspects of motion eventsthey involved: (a) aspect and directionality (39%),(b) manner (30%), and (c) path (31%).

The first category involved spatiotemporalcontours encoded through aspect and direction-ality, with particular difficulties displayed in theuse of prefixed perfective verbs to mark thebeginning of action. We have already illustratedthis pattern in Example 3, where the speakerused an auxiliary verb and a multidirectionalimperfective verb instead of a perfective verb.Example 4 illustrates a different approach to thesame lexicalization problem, namely a verbal

(a) он [мальчик] залез на камень on [mal’chik] zalez na kamen’he [a boy] up-climbed on (the) stone‘He climbed on the stone.’

(b) мальчик у бегает от филинаmal’chik ubegaet ot filina(the) boy (is) away-running from (the)

(eagle-)owl‘A boy is running away from the owl.’

они будут летать за собакойoni budut letat’ za sobakoithey will (be)

flying (around)after (the) dog, Dat

‘They will be flying after the dog.’(an appropriate lexical choice here is poletiat [‘will fly,in one direction’])


construction with the verb nachat’ (‘to start, tobegin’) and a multidirectional imperfective verbused, incorrectly, tomark the beginning of action:

EXAMPLE 4

The second category of errors involved the lossof lexical precision in the marking of manner. Insome contexts, participants extended the Russianverbs idti/khodit’ (‘to walk’/‘to be walking’) andtheir derivatives to other types of motion underthe influence of the generic English go, as seenin Example (1). In other contexts, bilingualsextended the meanings of other Russian mannerverbs to contexts to which they do not apply, asseen in Example 5:

EXAMPLE 5

The third category of errors involved incorrectmarking of path through prefixes and preposi-tions, as seen in Example 6:

EXAMPLE 6

Both this and the previous examples show thatwhile overall the L1 Russian motion lexiconremains stable, some participants are beginningto lose certain subtle distinctions, such as thedistinction between upast’ (‘to fall down’) andvypast’ (‘to fall out of an enclosed entity’). 53% ofall the errors occur in early bilinguals but, giventhe fact that they differed from other groups inboth AoA and LoR, we cannot attribute thefindings to one or the other variable, only to theircombined effect. In the absence of longitudinaldata, it is also impossible to determine withcertainty whether these errors reflect incompleteacquisition of the L1 Russian motion lexicon, L2English influence on the previously acquired L1Russian, attrition of the previously acquired L1Russian distinctions, or a combination of all three.

DISCUSSION AND CONCLUSIONS

The findings of the present study show that L1Russian speakers segment motion events in amore fine-grained way and pay greater attentionto both path and manner of motion than L1English speakers. More specifically, in Frog storynarratives, L1 Russian speakers marked mannerin a near-obligatory fashion, used significantlymore types and tokens of motion verbs, andencoded more complex temporal contours (as-pect) and locative trajectories (path and direc-tionality) than speakers of L1 English. Thesedifferences, consistent with the findings of Hasko(2010a) and Iakovleva (2012), suggest that thereliance on English as a language of conveniencein studies of motion language and cognition mayhave skewed the findings to date because Englishdoes not require obligatory manner encoding.

Following Lucy (1992, 1996), we find thedifference between obligatory and optionalencoding of manner important for research onlanguage and cognition and L2 acquisition. Ifthe encoding is optional, its frequency, be it inthe lexicon or in speech, is a poor predictor ofattention biases, not least because it may varyacross contexts and tasks. Obligatory or near-

TABLE 4Errors in the L1 Russian Motion Lexicon of Russian–English Bilinguals

Types of ErrorsEarly REB(# of cases)

Childhood REB(# of cases)

Late REB(# of cases) Total

Aspect and Directionality 19 9 4 32 (39%)Manner 14 9 1 24 (30%)Path 10 8 7 25 (31%)Total 43 (53%) 26 (32%) 12 (15%) 81

олень начал скакатьolen’ nachal skakat’(the) deer started hopping/galloping‘The deer started to hop/gallop.’(an appropriate lexical choice here is poskakal [‘setoff hopping’/‘galloping’])

мальчик запрыгнул на деревоmal’chik zaprygnul na derevo(the) boy jumped on (the) tree‘The boy jumped on the tree.’(an appropriate lexical choice here is zalez[‘climbed up’])

собака упала из домаsobaka upala iz doma(the) dog down-fell from on (the) tree‘The dog fell down from the house.’(an appropriate lexical choice here is vypala [‘out-fell’])


obligatory marking is a more stable predictor andwe hypothesize that it is the near-obligatorymarking of manner in Polish that accountsfor the differences in attention to mannerbetween speakers of Polish and speakers of otherS-languages revealed by Bohnemeyer et al.(2006). This means that, as far as manner isconcerned, studies of language effects in motioncognition may be more successful if they com-pared memory and attention for manner inspeakers of languages with optional mannerencoding, like French or Spanish, and speakersof languages with obligatory or near-obligatorymanner encoding, like Polish or Russian.

Yet the reliance on the S/V dichotomy is onlyone of the limitations in the study of languageeffects in motion cognition and L2 acquisition.The second and more serious limitation, in ourview, is the focus onuniversalmotion componentsthat are so salient in visual perception that neitherattention normemory are likely to be significantlymediated by language. A more promising line ofinquiry in the study of motion cognition and L2acquisition involves language-specific motioncomponents, such as directionality in Russianor associated motion in Arrernte. Undoubtedly,these components can be expressed in otherlanguages through a variety of means; yet, asargued by Lucy (1992, 1996), what matters mostin the study of language effects on cognition isnot what can be expressed but what must beexpressed. Thus, future studies could productivelyexamine whether L1 Russian speakers perceiveunidirectional and multidirectional motionevents as categorically distinct, remember locativetrajectories better than speakers of languages thatdo not encode directionality, or display bettermemory for motion events in general, as a resultof the greater complexity in the semantic domainof motion and automaticity of its obligatorydistinctions.

While we do not know yet whether obligatori-ness and complexity of encoding influencenonverbal cognition, we do know that they affectL2 learning outcomes on two levels: cognitive(meaning components) and linguistic (surfaceencoding). American learners of L2 Russianfaced with the requirement to automaticallymark manner, aspect, and directionality ofmotion display both linguistic difficulties (inmapping unbounded particles, in a distributedmanner, onto bounded and unbounded mor-phemes) and cognitive difficulties (in increasingattention to distinctions that have to be markedautomatically and linking motion verbs withthe appropriate mental imagery [Hasko, 2009,

2010b]). Erroneous interlingual identificationsof Russian manner verbs, such as idti (‘to walk’),with English generics, such as to go, revealdisregard for the manner of motion and thusL1 transfer that is conceptual, rather thanlinguistic, in nature.

A different situation obtains with L1 Russianlearners of L2 English. While our analysesrevealed differences in lexical diversity of the L2lexicon among the three bilingual subgroups,they did not identify any constraints placed by L1Russian on acquisition of the L2 English motionlexicon. Russian–English bilinguals patternedwith L1 English speakers in the frequency ofmotion verb use andmanner encoding and in thepreference for one-segment clauses in encodingof path. The lack of L1 Russian influence on L2English motion lexicon is not surprising becausethe relationship between the two languages offersno opportunities for negative L1 transfer: Englishencodes fewer motion components than Russian,does not mark them obligatorily, and hasunbounded morphemes that learners can identi-fy with Russian prepositions. As a consequence,our bilingual participants have learned to parsemotion events in a more holistic manner and todecrease attention to manner, appealing togeneric verbs.

In L1 Russian, all three groups of Russian–English bilinguals maintained the obligatoryfocus on manner and directionality and thepreference for two-segment path encoding andunidirectional perfective verbs. Replicating theresults of Pavlenko’s (2010) study conducted withdifferent elicitation stimuli, these findings raisean intriguing possibility that automaticity ofprocessing of obligatory distinctions makesthem less susceptible to L2 influence and attritioneffects in late bilinguals. At the same time, earlybilinguals used fewer motion verbs, displayedlower lexical diversity than L1 Russian speakers,and made a variety of errors.

Now, what, if any, relevance do these findingshave for the foreign language classroom? Whilestudies of bilingualism in naturalistic settings canonly indirectly inform us about foreign languageteaching and learning, the findings of our studywould not surprise any language instructor—teachers of Russian know perfectly well thatRussian motion verbs are difficult to learn andteachers of English know just as well that theyneed to push their students beyond the genericcome and go to the riches of the English phrasalverb system. The contribution of our study, as wesee it, is in highlighting the links among motionencoding, attention, and cognition and in


emphasizing the fact that the difficulties studentsexperience are not simply linguistic but alsoconceptual and attentional. Russian requires L1English speakers to pay simultaneous attention toa variety of aspects of motion and to make newobligatory distinctions, be it more generally, interms of aspect and directionality, or morespecifically, in terms of individual motion events,such as falling down (e.g., upast’ ‘fall down’ vsvypast’ ‘fall down out of an entity’). In turn,English may not require L1 Russian speakers toacquire new distinctions but it requires them todecrease attention to some of the distinctionsencoded in Russian, to restructure some ofconceptual groupings formed through L1 Rus-sian, and to learn to encode locative trajectoriesthrough unbounded satellites.

We argue that, to help learners of both L2Russian and L2 English, instruction cannot belimited to decontextualized form-focused exer-cises, such as descriptions or fill-in-the-blanks.Rather, the focus of instruction needs to shiftto conceptual distinctions—the informationload carried by individual forms and mappingsbetween these forms and their dynamic multi-modal representations—with clips and videosreplacing static textbook pictures (for an extend-ed version of this argument and sample exercises,see Pavlenko &Driagina, 2009). In the L2 Russianclassroom, for instance, attention exercises couldrequire students to watch short video clips, toattend to the key aspects of motion events thatneed to be encoded, and to reflect on theseaspects. Comprehension exercises might requirestudents to explain why characters in short storiesor video clips opted for particular lexical choicesand to decode information contained in singlemotion verbs, such as prikhodil (‘came’), whichcaptures a motion event, where a single malecame for a visit on foot and already left.Production exercises need to place the forms inmeaningful discursive and narrative contextsrelevant to and engaging for the learners, ratherthan in endless series of actions performed bycartoon characters, common for Russian motionverb materials. Some tasks may ask learners todescribe a particular video clip and to comparetheir own lexical choices with those of L1 Russianspeakers (e.g., Pavlenko & Driagina, 2009), whileothers could focus on miscommunication thatstems from selection of inappropriate prefixes orparticles and the lack of lexical precision. The keyto teaching motion vocabulary, as we see it, is inplacing it in the meaningful and dynamic contextthat facilitates cognitive restructuring, develop-ment of appropriate mental imagery, and inte-

gration of linguistic forms and multi-modalconceptual representations.

ACKNOWLEDGMENTS

We thank the editors and the two anonymousreviewers for their insightful and helpful commentson the original version of this manuscript. All remainingerrors and inconsistencies are our own.

NOTES

1 For alternative approaches, see Bohnemeyer andPederson (2011), Bylund (2011), and Gullberg (2011).

2 Throughout, our transliteration of Russian wordsfollows the Library of Congress conventions.

3 Our use of the term ‘boundedness’ is limited tomorphological boundedness and does not extend toevent boundedness, that is spatial or temporal aspects ofmotion events.

4 The study is based on the dissertation by the secondauthor, which was part of a larger study carried out bythe first author, in collaboration with the second authorand Drs. Victoria Hasko, Barbara Malt, and NinaVyatkina.

REFERENCES

Alexandrova, N., & Watt, A. (2013). Russian step by stepverbs of motion. Create Space Independent Publish-ing Platform.

Allen, S., Ozyurek, A., Kita, S., Brown, A., Furman, R.,Ishizuka, T., & Fujii, M. (2007). Language-specificand universal influences in children’s syntacticpackaging of manner and path: A comparison ofEnglish, Japanese, andTurkish.Cognition, 102, 16–48.

Beavers, J., Levin, B., & Tham, S. (2010). The typology ofmotion expressions revisited. Journal of Linguistics,46, 331–377.

Berman, R., & Slobin, D. (1994). Relating eventsin narrative: A cross-linguistic developmental study.Hillsdale, NJ: Lawrence Erlbaum.

Bohnemeyer, J., Eisenbeiss, S., & Narasimhan, B.(2006). Ways to go: Methodological consider-ations in Whorfian studies on motion events. EssexResearch Reports in Linguistics, 50, 1–20.

Bohnemeyer, J., & Pederson, E. (Eds.). (2011). Eventrepresentation in language and cognition. Cambridge:Cambridge University Press.

Brown, A., & Gullberg, M. (2008). Bidirectional cross-linguistic influence in L1–L2 encoding of mannerin speech and gesture: A study of Japanesespeakers of English. Studies in Second LanguageAcquisition, 30, 225–251.


Brown, A., & Gullberg, M. (2010). Changes in encodingof path of motion in a first language duringacquisition of a second language. Cognitive Lin-guistics, 21, 263–286.

Brown, A., & Gullberg, M. (2011). Bidirectional cross-linguistic influence in event conceptualization?Expressions of path among Japanese learners ofEnglish. Bilingualism: Language and Cognition, 14,79–94.

Bylund, E. (2011). Language-specific patterns in eventconceptualization: Insights from bilingualism. InA. Pavlenko (Ed.), Thinking and speaking in twolanguages (pp. 108–142). Bristol, UK: MultilingualMatters.

Cadierno, T. (2010). Motion in Danish as a secondlanguage: Does the learner’s L1 make a differ-ence? In Z. Han & T. Cadierno (Eds.), Linguisticrelativity in SLA: Thinking for speaking (pp. 1–33).Bristol, UK: Multilingual Matters.

Cardini, F. (2008). Manner of motion saliency: Aninquiry into Italian. Cognitive Linguistics, 19, 533–570.

Cardini, F. (2010). Evidence against Whorfian effects inmotion conceptualization. Journal of Pragmatics,42, 1442–1459.

Daller, M., Treffers–Daller, J., & Furman, R. (2011).Transfer of conceptualization patterns in bilin-guals: The construal of motion events in Turkishand German. Bilingualism: Language and Cognition,14, 95–119.

Filipovic, L. (2011). Speaking and remembering in oneor two languages: Bilingual vs. monolinguallexicalization and memory for motion events.International Journal of Bilingualism, 15, 466–485.

Finkbeiner, M., Nicol, J., Greth, D., & Nakamura, K.(2002). The role of language in memory foractions. Journal of Psycholinguistic Research, 31, 447–457.

Gathercole V., & Min, H. (1997). Word meaning biasesor language-specific effects: Evidence from En-glish, Spanish, and Korean. First Language, 17, 31–56.

Gennari, S., Sloman, S., Malt, B., & Fitch, T. (2002).Motion events in language and cognition. Cogni-tion, 83, 49–79.

Gor, K., Cook, S., Malyushenkova, V., & Vdovina, T.(2009). Verbs of motion in highly proficientlearners and heritage speakers of Russian. Slavicand Eastern European Journal, 53, 386–408.

Gullberg, M. (2011). Thinking, speaking, and gesturingabout motion in more than one language. In A.Pavlenko (Ed.), Thinking and speaking in twolanguages (pp. 143–169). Bristol, UK: MultilingualMatters.

Hasko, V. (2009). The locus of difficulties in theacquisition of Russian verbs of motion by highlyproficient learners. Slavic and Eastern EuropeanJournal, 53, 360–385.

Hasko, V. (2010a). Semantic composition of motionverbs in Russian and English: The case of intra-typological variability. In V. Hasko & R. Perelmut-

ter (Eds.), New approaches to Slavic verbs of motion(pp. 197–224). Philadelphia/Amsterdam: JohnBenjamins.

Hasko, V. (2010b). The role of thinking for speaking inadult L2 speech: The case of (non) unidirection-ality encoding by American learners of Russian. InZ. Han & T. Cadierno (Eds.), Linguistic relativity inSLA: Thinking for speaking (pp. 34–58). Bristol, UK:Multilingual Matters.

Hasko, V., & Perelmutter, R. (Eds.). (2010). Newapproaches to Slavic verbs of motion. Philadelphia/Amsterdam: John Benjamins.

Hohenstein, J., Eisenberg, A., & Naigles, L. (2006). Is hefloating across or crossing afloat? Cross-influenceof L1 and L2 in Spanish–English bilingual adults.Bilingualism: Language and Cognition, 9, 249–261.

Iakovleva, T. (2012). Typological constraints in foreignlanguage acquisition: The expression of voluntarymotion by upper intermediate and advancedRussian learners of English. Language, Interaction,and Acquisition, 3, 231–260.

Imai, M., & Gentner, D. (1997). A cross-linguistic studyof early word meaning: Universal ontology andlinguistic influence. Cognition, 62, 169–200.

Imai, M., & Mazuka, R. (2003). Reevaluating linguisticrelativity: Language-specific categories and therole of universal ontological knowledge in theconstrual of individuation. In D. Gentner & S.Goldin–Meadow (Eds.), Language in mind: Advan-ces in the study of language and thought (pp. 429–464). Cambridge, MA: The MIT Press.

Imai, M., & Mazuka, R. (2007). Language-relativeconstrual of individuation constrained by univer-sal ontology: Revisiting language universals andlinguistic relativity. Cognitive Science, 31, 385–413.

Jarvis, S. (2002). Short texts, best-fitting curves and newmeasures of lexical diversity. Language Testing,19(1), 57–84.

Jarvis, S., &Odlin, T. (2000).Morphological type, spatialreference, and language transfer. Studies in SecondLanguage Acquisition, 22, 535–556.

Jarvis, S., & Pavlenko, A. (2008). Cross-linguistic influencein language and cognition. New York/London:Routledge.

Larranaga, P., Treffers–Daller, J., Tidball, F., & Ortega,M. (2011). L1 transfer in the acquisition ofmanner and path in Spanish by native speakersof English. International Journal of Bilingualism, 16,117–138.

Levinson, S., & Wilkins, D. (Eds.). (2006). Grammars ofspace: Explorations in cognitive diversity. Cambridge:Cambridge University Press.

Li, P., Dunham, Y., &Carey, S. (2009).Of substance: Thenature of language effects on entity construal.Cognitive Psychology, 58, 487–524.

Loucks, J., & Pederson, E. (2011). Linguistic and non-linguistic categorization of complex motionevents. In J. Bohnemeyer & E. Pederson (Eds.),Event representation in language and cognition(pp. 108–133). Cambridge: Cambridge UniversityPress.


Lucy, J. (1992).Grammatical categories and cognition: A casestudy of the linguistic relativity hypothesis. Cambridge:Cambridge University Press.

Lucy, J. (1996). The scope of linguistic relativity: Ananalysis and review of empirical research. In J.Gumperz& S. Levinson (Eds.),Rethinking linguisticrelativity (pp. 37–69). Cambridge: CambridgeUniversity Press.

Maguire, M., Hirsh–Pasek, K., Michnick Golinkoff, R.,Imai, M., Haryu, E., Vanegas, S., Okada, H.,Pulverman, R., & Sanchez–Davis, B. (2010). Adevelopmental shift from similar to language-specific strategies in verb acquisition: A compari-son of English, Spanish, and Japanese. Cognition,114, 299–319.

Mahota, W. (1996). Russian motion verbs for intermediatestudents. New Haven, CT: Yale University Press.

Mayer, M. (1969). Frog, where are you? New York: Dial.Muravyova, L. (1986). Verbs of motion in Russian (5th ed.).

Moscow: Russky Yazyk Publishers.Papafragou, A., Hulbert, J., & Trueswell, J. (2008). Does

language guide event perception? Evidence fromeye movements. Cognition, 108, 155–184.

Papafragou,A.,Massey, C.,&Gleitman,L. (2002). Shake,rattle, ’n roll: The representation of motion inlanguage and cognition. Cognition, 84, 289–219.

Papafragou, A., & Selimis, S. (2010). Event categoriza-tion and language: A cross-linguistic study ofmotion. Language and Cognitive Processes, 25, 224–260.

Pavlenko, A. (2010). Verbs of motion in L1 Russian ofRussian–English bilinguals. Bilingualism: Languageand Cognition, 13, 49–62.

Pavlenko, A. (2014). The bilingual mind and what it tells usabout language and thought. Cambridge: CambridgeUniversity Press.

Pavlenko, A., &Driagina, V. (2009).Advancing in Russianthrough narration. University Park, PA: CALPERPublications.

Polinsky, M. (2008). Heritage language narratives. In D.Brinton, O. Kagan, & S. Bauckus (Eds.), Heritagelanguage education: A new field emerging (pp. 149–164). New York: Routledge.

Polivin, A. (2010). Russian verbs of motion: The workbook.Create Space Independent Publishing Platform.

Regier, T., & Zheng, M. (2007). Attention to endpoints:A cross-linguistic constraint on spatial meaning.Cognitive Science, 31, 705–719.

Segalowitz, N. (2010). Cognitive bases of second languagefluency. New York/London: Routledge.

Sera, M., & Goodrich, W. (2010). Who thinks that apiece of furniture refers to a broken couch?

Count–mass constructions and individuation inEnglish and Spanish. Cognitive Linguistics, 21, 419–442.

Slobin, D. (1996a). From “thought and language” to“thinking for speaking.” In J. Gumperz & S.Levinson (Eds.), Rethinking linguistic relativity(pp. 70–114). Cambridge: Cambridge UniversityPress.

Slobin, D. (1996b). Two ways to travel: Verbs of motionin English and Spanish. In M. Shibatani & S.Thompson (Eds.), Grammatical constructions: Theirform and meaning (pp. 195–219). Oxford, UK:Clarendon Press.

Slobin, D. (2004a). How people move: Discourse effectsof linguistic typology. In C. Moder & A. Marti-novic–Zic (Eds.), Discourse across languages andcultures (pp. 195–210). Philadelphia/Amsterdam:John Benjamins.

Slobin, D. (2004b). The many ways to search for a frog:Linguistic typology and the expression of motionevents. In S. Stromqvist & L. Verhoeven (Eds.),Relating events in narrative: Typological and contextualperspectives (pp. 219–257). Mahwah, NJ: LawrenceErlbaum.

Slobin, D. (2006). What makes manner of motionsalient? Explorations in linguistic typology, dis-course, and cognition. In M. Hickmann & S.Robert (Eds.), Space in languages: Linguistic systemsand cognitive categories (pp. 59–81). Philadelphia/Amsterdam: John Benjamins.

Stromqvist, S., & Verhoeven, L. (Eds.). (2004). Relatingevents in narrative: Typological and contextual perspec-tives. Mahwah, NJ: Lawrence Erlbaum.

Subrahmanyam, K., & Chen, H. (2006). A cross-linguistic study of children’s noun learning: Thecase of object and substance words. First Language,26, 141–160.

Talmy, L. (1985). Lexicalization patterns: Semanticstructure in lexical forms. In T. Shopen (Ed.),Language typology and syntactic description. Vol. 3Grammatical categories and the lexicon (pp. 57–149).Cambridge: Cambridge University Press.

Talmy, L. (1991). Path to realization: A typology of eventconflation. In Proceedings of the Seventeenth AnnualMeeting of the Berkeley Linguistics Society (pp. 480–519). Berkeley, CA: Berkeley Linguistics Society.

Talmy, L. (2000). Toward a cognitive semantics. Cam-bridge, MA: The MIT Press.

Zalizniak, A., & Shmelev, A. [Зализняк, А., & Шмелев, А.].(2000). Введение в русскуюаспектологию [Intro-duction to the study of Russian aspect]. Moscow:Iazyki russkoi kul’tury.


Motion Encoding in Russian and English: Moving Beyond · PDF fileKeywords: Russian; motion...

Documents

Transcript of Motion Encoding in Russian and English: Moving Beyond · PDF fileKeywords: Russian; motion...