Minimal Word Research Articles

The uptake of spectral and temporal cues in vowel perception is rapidly influenced by context

Speech perception is dependent on auditory information within phonemes such as spectral or temporal cues. The perception of those cues, however, is affected by auditory information in surrounding context (e.g., a fast context sentence can make a target vowel sound subjectively longer). In a two-by-two design the current experiments investigated when these different factors influence vowel perception. Dutch listeners categorized minimal word pairs such as /tɑk/–/taːk/ (“branch”–“task”) embedded in a context sentence. Critically, the Dutch /ɑ/–/aː/ contrast is cued by spectral and temporal information. We varied the second formant (F2) frequencies and durations of the target vowels. Independently, we also varied the F2 and duration of all segments in the context sentence. The timecourse of cue uptake on the targets was measured in a printed-word eye-tracking paradigm. Results show that the uptake of spectral cues slightly precedes the uptake of temporal cues. Furthermore, acoustic manipulations of the context sentences influenced the uptake of cues in the target vowel immediately. That is, listeners did not need additional time to integrate spectral or temporal cues of a target sound with auditory information in the context. These findings argue for an early locus of contextual influences in speech perception.

Multicriteria-Based Computer-Aided Pronunciation Quality Evaluation of Sentences

The problem of the sentence-based pronunciation evaluation task is defined in the context of subjective criteria. Three subjective criteria (that is, the minimum subjective word score, the mean subjective word score, and first impression) are proposed and modeled with the combination of word-based assessment. Then, the subjective criteria are approximated with objective sentence pronunciation scores obtained with the combination of word-based metrics. No a priori studies of common mistakes are required, and class-based language models are used to incorporate incorrect and correct pronunciations. Incorrect pronunciations are automatically incorporated by making use of a competitive lexicon and the phonetic rules of students' mother and target languages. This procedure is applicable to any second language learning context, and subjective-objective sentence score correlations greater than or equal to 0.5 can be achieved when the proposed sentence-based pronunciation criteria are approximated with combinations of word-based scores. Finally, the subjective-objective sentence score correlations reported here are very comparable with those published elsewhere resulting from methods that require a priori studies of pronunciation errors.

Listeners retune phoneme categories across languages.

Native listeners adapt to noncanonically produced speech by retuning phoneme boundaries by means of lexical knowledge. We asked whether a second language lexicon can also guide category retuning and whether perceptual learning transfers from a second language (L2) to the native language (L1). During a Dutch lexical-decision task, German and Dutch listeners were exposed to unusual pronunciation variants in which word-final /f/ or /s/ was replaced by an ambiguous sound. At test, listeners categorized Dutch minimal word pairs ending in sounds along an /f/-/s/ continuum. Dutch L1 and German L2 listeners showed boundary shifts of a similar magnitude. Moreover, following exposure to Dutch-accented English, Dutch listeners also showed comparable effects of category retuning when they heard the same speaker speak her native language (Dutch) during the test. The former result suggests that lexical representations in a second language are specific enough to support lexically guided retuning, and the latter implies that production patterns in a second language are deemed a stable speaker characteristic likely to transfer to the native language; thus retuning of phoneme categories applies across languages.

ALGORITHMIC COMBINATORICS ON PARTIAL WORDS

Algorithmic combinatorics on partial words, or sequences of symbols over a finite alphabet that may have some do-not-know symbols or holes, has been developing in the past few years. Applications can be found, for instance, in molecular biology for the sequencing and analysis of DNA, in bio-inspired computing where partial words have been considered for identifying good encodings for DNA computations, and in data compression. In this paper, we focus on two areas of algorithmic combinatorics on partial words, namely, pattern avoidance and subword complexity. We discuss recent contributions as well as a number of open problems. In relation to pattern avoidance, we classify all binary patterns with respect to partial word avoidability, we classify all unary patterns with respect to hole sparsity, and we discuss avoiding abelian powers in partial words. In relation to subword complexity, we generate and count minimal Sturmian partial words, we construct de Bruijn partial words, and we construct partial words with subword complexities not achievable by full words (those without holes).

Introducing Communicative Event As A Tool To Communicate Via the Medium Of Language: The Case of Job Advertisement

This study is going to investigate the language style of classified advertisements of job vacancies in order to understand how advertisers manipulate language in convincing the job-seekers to apply for the advertised jobs. Furthermore, the study tries to examine the ways that advertisements’ language may be concisely used to convey the meaning in the fewest words and avoid blurring the job seekers. To gain a clear picture of the form and content of job advertisements the views of (Lunde, 1974) and (Fairclough, 1989) are exploited to clarify how job advertisements may be perceived as a class of communicative events in order to communicate via the medium of language with potential job seekers and persuade them towards applying for the advertised jobs. The data for this qualitative study which came from the local English newspaper (Star) in the month of August were analyzed through content analysis. Being grounded in Lund (1974), who believes that an advertisement should: attract attention, arouse interests, stimulate desires, create conviction and get action, the sample advertisements were examined and the results revealed that all are in accordance with Lund’s AIDEA formula. An analysis of the surface structure of the samples did not reveal how the basic structure of English sentences was adapted. The spaces allocated to these types of advertisements were so limited. Thus, the advertisers manipulate language in a way that they can convey their purposes with the use of minimum words. This may only be achieved by the use of the key words to convey the message and nominalizations.

Codes from incidence matrices of graphs

We examine the p-ary codes, for any prime p, from the row span over $${\mathbb {F}_p}$$ of |V| × |E| incidence matrices of connected graphs Γ = (V, E), showing that certain properties of the codes can be directly derived from the parameters and properties of the graphs. Using the edge-connectivity of Γ (defined as the minimum number of edges whose removal renders Γ disconnected) we show that, subject to various conditions, the codes from such matrices for a wide range of classes of connected graphs have the property of having dimension |V| or |V| − 1, minimum weight the minimum degree δ(Γ), and the minimum words the scalar multiples of the rows of the incidence matrix of this weight. We also show that, in the k-regular case, there is a gap in the weight enumerator between k and 2k − 2 of the binary code, and also for the p-ary code, for any prime p, if Γ is bipartite. We examine also the implications for the binary codes from adjacency matrices of line graphs. Finally we show that the codes of many of these classes of graphs can be used for permutation decoding for full error correction with any information set.

Filler Syllables in the Acquisition of Hebrew: A Prosodic Account

In this paper we review the phenomenon of filler syllables in the early speech of RM, a monolingual Hebrew-acquiring girl. Our findings support former claims, in particular those made in Demuth (2001a), which tie the emergence of filler syllables with the process of prosodic structure acquisition. We present longitudinal corpus-based data that are analyzed against pre-defined stages in RM’s acquisition of prosodic units to corroborate our predictions. We show that the distributional patterns of filler syllables can be attributed to prosodic requirements such as minimal word and foot type preferences during early stages of acquisition, and the emergence of concatenative morpho-syntax at later stages, where prosodic units above the phonological word begin to appear. In our view, a proper analysis of filler syllables is possible with “pure” phonological devices that rely on the prosodic hierarchy (Selkirk 1978, 1986, Nespor and Vogel 1986), since this hierarchy already encapsulates an underlying interaction with morpho-syntactic domains.Our criteria for defining filler syllables do not presuppose that fillers are limited to any specific position, yet we take measures to ensure that actual fillers are not overlooked while general epenthesis processes are not confused with augmentation processes (only the latter are considered as fillers). We do so by detection of vowel insertions that stem from vocalization of approximant coda targets, cluster simplification processes and coda resyllabification. We believe that our strategy in considering subsets of fillers enables to verify attested trends and consider their complexities in full.

Speaking of Word Count…

About a year ago, we informed the potential authors of future manuscripts submitted to sections of the Journal of Food Science that we would be limiting word count on manuscripts. In this action, we were trying to decrease the number of printed pages particularly for JFS while enhancing the readability of the papers. In a survey of printed papers at that time, Amanda Ferguson discovered that most manuscripts to JFS were a little over 5,000 in word count, and manuscripts in Comprehensive Reviews in Food Science and Food Safety were all over 10,000 words. With those data as the guideline, we set a JFS research manuscript word count limit (excluding tables and figures) at 5,000 words and minimum word count for CRFSFS at 10,000 words. Starting November 2011, authors were asked to voluntarily submit word count on manuscripts and in January 2011, words count was required. Needless to say, we had some flexibility in applying the word count rule. On the other hand, I received not one letter to the editor decrying the use of word count or complaining that it was too limiting. Nor did I receive any oral argument while I attended the IFT AMFE in June. I take this inaction as a sign of not only acceptance but of support for word count limitation. With that in mind, I asked Amanda to again do a survey of word count on our accepted manuscripts. The data for a sampling of papers published in 2011 are in and they tell the following story. Of the 141 papers in the JFS research sections, only 18 were over the 5,000 word limit, and only 3 were more than 1000 words over. The mean of 4500 word count for the research sections of JFS compares to the October 2010 reported mean of 5140 and range of 2350-8560. Only 1 review paper in JFS was over the 10,000 word limit. The effect of the decrease in word count was to reduce the page count per paper from 7.3 to 7.0. In other words, the change achieved the result we were hoping for and there were not untoward consequences. Now, for the kicker, should we be looking at a 4,500 word count limit for research papers submitted to JFS for 2012? Send your thoughts to me at dlund@cals.wisc.edu.

Investigation of improved approaches to bayes risk decoding

Bayes risk (BR) decoding methods have been widely investigated in the speech recognition area due to its flexibility and complexity compared with the maximum a posteriori (MAP) method regarding to minimum word error (MWE) optimization. This paper investigates two improved approaches to the BR decoding, aiming at minimizing word error. The novelty of the proposed methods is shown in the explicit optimization of the objective function, the value of which is calculated by an improved forward algorithm on the lattice. However, the result of the first method is obtained by an expectation maximization (EM) like iteration, while the result of the second one is achieved by traversing the confusion network (CN), both of which lead to an optimized objective function value with distinct approaches. Experimental results indicate that the proposed methods result in an error reduction for lattice rescoring, compared with the traditional CN method for lattice rescoring.

Listeners retune phoneme boundaries across languages

Listeners can flexibly retune category boundaries of their native language to adapt to non-canonically produced phonemes. This only occurs, however, if the pronunciation peculiarities can be attributed to stable and not transient speaker-specific characteristics. Listening to someone speaking a second language, listeners could attribute non-canonical pronunciations either to the speaker or to the fact that she is modifying her categories in the second language. We investigated whether, following exposure to Dutch-accented English, Dutch listeners show effects of category retuning during test where they hear the same speaker speaking her native language, Dutch. Exposure was a lexical-decision task where either word-final [f] or [s] was replaced by an ambiguous sound. At test listeners categorized minimal word pairs ending in sounds along an [f]-[s] continuum. Following exposure to English words, Dutch listeners showed boundary shifts of a similar magnitude as following exposure to the same phoneme variants...

Codes and designs from triangular graphs and their line graphs

Abstract For any prime p, we consider p-ary linear codes obtained from the span over $\mathbb{F}_p $ p of rows of incidence matrices of triangular graphs, differences of the rows and adjacency matrices of line graphs of triangular graphs. We determine parameters of the codes, minimum words and automorphism groups. We also show that the codes can be used for full permutation decoding.

Codes from the incidence matrices of graphs on 3-sets

We examine the p-ary linear codes from incidence matrices of the three uniform subset graphs with vertex set the set of subsets of size 3 of a set of size n, with adjacency defined by two vertices as 3-sets being adjacent if they have zero, one or two elements in common, respectively. All the main parameters of the codes and the nature of the minimum words are obtained, and it is shown that the codes can be used for full error-correction by permutation decoding. We examine also the binary codes of the line graphs of these graphs.

Codes from the incidence matrices and line graphs of Hamming graphs $H^k(n,2)$ for $k \geq 2$

We examine the $p$-ary codes, for any prime $p$, that can be obtained from incidence matrices and line graphs of the Hamming graphs, $H^k(n,m)$, for $k \geq 2$. For $m=2$, we obtain the main parameters of the codes from the incidence matrices, including the minimum weight and the nature of the minimum words. We show that all the codes can be used for full permutation decoding.

Aniterative approach to Bayes risk decoding and system combination

We describe a novel approach to Bayes risk (BR) decoding for speech recognition, in which we attempt to find the hypothesis that minimizes an estimate of the BR with regard to the minimum word error (MWE) metric. To achieve this, we propose improved forward and backward algorithms on the lattices and the whole procedure is optimized recursively. The remarkable characteristics of the proposed approach are that the optimization procedure is expectation-maximization (EM) like and the formation of the updated result is similar to that obtained with the confusion network (CN) decoding method. Experimental results indicated that the proposed method leads to an error reduction for both lattice rescoring and lattice-based system combinations, compared with CN decoding, confusion network combination (CNC), and ROVER methods.

Topic Mining based on Word Posterior Probability in Spoken Document

For speech recognition system, there are three kinds of result representations as one-best, N -best and Lattice. Since lattice has multi-path which can reduce the effect of recognition error rate, it is widely applied nowadays. In fact, there are amount of redundancies in lattice, which leads to the increasing of complexity of latter algorithm based on it. Additionally, for the decoding algorithm, it is acted as maximum a posterior probability (MAP) which can only guarantee the posterior probability of the whole sentence is of maximum. For MAP does not mean the highest syllable recognition rate, here, confusion network is introduced in topic mining system. In the clustering during confusion network, the minimum word error rule is adopted, which is proper to topic mining system since the least meaningful unit is word in Chinese and word information is most important in topic mining. In this paper, a simplified confusion network generation algorithm is proposed to handle some problems caused by insertion error during recognition. Then based on the confusion network, a word list extraction approach is proposed, in which, the dictionary is adopted to judge whether the consecutive arc in confusion sets is a word. At this stage, the error word information produced by error recognition rate can be corrected to some extent. After the competition part in word list extraction on confusion network, a final word list with posterior probability can be obtained. Furthermore, this kind of posterior probability can be combined in topic mining system. SVD and NMF are adopted here to decompose the term-document matrix on the word list of confusion network. From the experiments, it can be drawn that the proposed approach based on confusion network can achieve better performance than that of one-best and N -best. Additionally, the modified weight which combined posterior probability into term-document matrix can further improve the system performance.

Growing words in the free group on two generators

This paper is concerned with minimal-length representatives of equivalence classes of words in $F_2$ under Aut $F_2$. We give a simple inequality characterizing words of minimal length in their equivalence class. We consider an operation that “grows” words from other words, increasing the length, and we study root words—minimal words that cannot be grown from other minimal words. Root words are “as minimal as possible” in the sense that their characterization is the boundary case of the minimality inequality. The property of being a root word is respected by equivalence classes, and the length of each root word is divisible by 4.

E-043 Novel use of the Penumbra Aspiration System in three cases of severe dural venous sinus thrombosis

IntroductionDural sinus venous thrombosis is a rare but potentially devastating disease. Current therapies include intravenous heparin, direct mechanical thrombectomy and local chemical thrombolysis. Despite aggressive management of this disease with...

Depth of reading vocabulary in hearing and hearing-impaired children

The main point of our study was to examine the vocabulary knowledge of pupils in grades 3–6, and in particular the relative reading vocabulary disadvantage of hearing-impaired pupils. The achievements of 394 pupils with normal hearing and 106 pupils with a hearing impairment were examined on two vocabulary assessment tasks: a lexical decision task and a use decision task. The target words in both tasks represent the vocabulary children should have at the end of primary school. The results showed that most hearing pupils reached this norm, whereas most hearing-impaired pupils did not. In addition, results showed that hearing-impaired pupils not only knew fewer words, but that they also knew them less well. This lack of deeper knowledge remained even when matching hearing and hearing-impaired children on minimal word knowledge. Additionally, comparison of the two tasks demonstrated the efficacy of the lexical decision task as a measure of lexical semantic knowledge.

Prosodic fusion and minimality in Kabardian

AbstractThe Northwest Caucasian language Kabardian displays a typologically unusual process of word formation, whereby two lexical roots fuse to form a single prosodic word whose phonological behaviour is parallel to prosodic words containing a single root. It is shown that this process of fusion, which is subject to a number of phonological and morphosyntactic restrictions, reflects a typologically unusual response to a cross-linguistically common minimal word requirement banning monomoraic prosodic words. Rather than employing segmental lengthening or insertion to ensure that minimality is satisfied, Kabardian chooses to violate the one-to-one mapping between grammatical and prosodic words. A further complication is the scalar nature of minimality in Kabardian: while the impossibility of fusion in certain prosodic and morphosyntactic contexts allows monomoraic prosodic words to surface, a more stringent minimality restriction ensures that all prosodic words have at least one mora.

Nonlinear Index Coding Outperforming the Linear Optimum

The following source coding problem was introduced by Birk and Kol: a sender holds a word x isin {0, 1}n, and wishes to broadcast a codeword to n receivers, Rn,..., Rn. The receiver Ri is interested in xi, and has prior side information comprising some subset of the n bits. This corresponds to a directed graph G on n vertices, where i j is an edge Ri Ri knows the bit xj. An index code for G is an encoding scheme which enables each Ri to always reconstruct xi, given his side information. The minimal word length of an index code was studied by Bar-Yossef, Birk, Jayram, and Kol (FOCS'06). They introduced a graph parameter, minrk2(G), which completely characterizes the length of an optimal linear index code for G. They showed that in various cases linear codes attain the optimal word length, and conjectured that linear index coding is in fact always optimal. In this work, we disprove the main conjecture of Bar-Yossef, Birk, Jayram, and Kol in the following strong sense: for any epsiv > 0 and sufficiently large n, there is an n-vertex graph G so that every linear index code for G requires codewords of length at least nepsiv and yet a nonlinear index code for G has a word length of ne. This is achieved by an explicit construction, which extends Alon's variant of the celebrated Ramsey construction of Frankl and Wilson. In addition, we study optimal index codes in various, less restricted, natural models, and prove several related properties of the graph parameter minrk(G).